Math mode - v.2.

20
Herbert Voß∗

October 9, 2006

Abstract
It is often said that TEX was designed for mathematical or technical
purposes. This may be true when we remember the reasons why
Donald Knuth created TEX. But nowadays there are many examples in
which TEX is used for publications with no mathematical or technical
background content. However, writing publications with such material
is one of the important advantages of TEX. Because it seems impossible
to know all existing macros and options of (LA)TEX and the several
additional packages, especially of AMSmath. This is the reason why
I have attempted to gather all the relevant facts in this paper. An
advanced german version of this paper is available as a book [25].

Please report typos or any other comments to this documentation to

[email protected].
This file can be redistributed and/or modified under the terms of
the LaTeX Project Public License Distributed from CTAN archives in
directory CTAN://macros/latex/base/lppl.txt.

∗
Thanks for the feedback to: Hendri Adriaens; Alexander Boronka; Walter Brown;
Christian Faulhammer; José Luis Gómez Dans; Zongbao Fang; Azzam Hassam; Henning
Heinze; Martin Hensel; Morten Høgholm; M. Kalidoss; Dan Lasley; Angus Leeming; Tim
Love; Dan Luecking; Hendrik Maryns; Heinz Mezera; David Neuway; Joachim Punter; Carl
Riehm; Will Robertson; Christoph Rumsmüller; José Carlos Santos; Arnaud Schmittbuhl;
Rainer Schöpf; Jens Schwaiger; Uwe Siart; Martin Sievers; Heiko Stamer; Uwe Stöhr;
Carsten Thiel; Juan Luis Varona; David Weenink; Philipp Wook; Michael Zedler; Zou
Yuan-Chuan; and last but not least a special thanks to Monika Hattenbach for her excellent
job of proofreading.

1
CONTENTS CONTENTS

Contents
Page
I Standard LATEX math mode 10

1 Introduction 10

2 The Inlinemode 10
2.1 Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Fraction command . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Math in Chapter/Section Titles . . . . . . . . . . . . . . . . . 11
2.4 Equation numbering . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 Framed math . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.6 Linebreak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.7 Whitespace . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.8 AMSmath for the inline mode . . . . . . . . . . . . . . . . . . 13

3 Displaymath mode 13
3.1 equation environment . . . . . . . . . . . . . . . . . . . . . . 13
3.2 eqnarray environment . . . . . . . . . . . . . . . . . . . . . . 14
3.2.1 Short commands . . . . . . . . . . . . . . . . . . . . . 15
3.3 Equation numbering . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.1 Changing the style . . . . . . . . . . . . . . . . . . . . 16
3.3.2 Resetting a counter style . . . . . . . . . . . . . . . . . 16
3.3.3 Equation numbers on the left side . . . . . . . . . . . 17
3.3.4 Changing the equation number style . . . . . . . . . . 17
3.3.5 More than one equation counter . . . . . . . . . . . . 17
3.4 Labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.5 Frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4 array environment 20
4.1 Cases structure . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2 arraycolsep . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

5 Matrix 22

6 Super/Subscript and limits 24

6.1 Multiple limits . . . . . . . . . . . . . . . . . . . . . . . . . . 24
6.2 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

7 Roots 25

Mathmode.tex 2
CONTENTS CONTENTS

8 Brackets, braces . . . 26
8.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
8.1.1 Braces over several lines . . . . . . . . . . . . . . . . . 28
8.1.2 Middle bar . . . . . . . . . . . . . . . . . . . . . . . . 29
8.2 New delimiters . . . . . . . . . . . . . . . . . . . . . . . . . . 29
8.3 Problems with parentheses . . . . . . . . . . . . . . . . . . . . 30

9 Text in math mode 30

10 Font commands 31
10.1 Old-style font commands . . . . . . . . . . . . . . . . . . . . . 31
10.2 New-style font commands . . . . . . . . . . . . . . . . . . . . 31

11 Space 31
11.1 Math typesetting . . . . . . . . . . . . . . . . . . . . . . . . . 31
11.2 Additional horizontal spacing . . . . . . . . . . . . . . . . . . 33
11.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
11.4 Dot versus comma . . . . . . . . . . . . . . . . . . . . . . . . 34
11.5 Vertical whitespace . . . . . . . . . . . . . . . . . . . . . . . . 35
11.5.1 Before/after math expressions . . . . . . . . . . . . . . 35
11.5.2 Inside math expressions . . . . . . . . . . . . . . . . . 36

12 Styles 38

13 Dots 39

14 Accents 39
14.1 Over- and underbrackets . . . . . . . . . . . . . . . . . . . . . 40
14.1.1 Use of \underbracket{...} . . . . . . . . . . . . . . . 41
14.1.2 Overbracket . . . . . . . . . . . . . . . . . . . . . . . . 41
14.2 Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

15 Exponents and indices 42

16 Operators 43

17 Greek letters 44

18 Pagebreaks 45

19 \stackrel 45

20 \choose 46

21 Color in math expressions 46

Mathmode.tex 3
CONTENTS CONTENTS

22 Boldmath 47
22.1 Bold math titles and items . . . . . . . . . . . . . . . . . . . . 48

23 Multiplying numbers 48

24 Other macros 49

II AMSmath package 50

25 align environments 51
25.1 The default align environment . . . . . . . . . . . . . . . . . 52
25.2 alignat environment . . . . . . . . . . . . . . . . . . . . . . . 53
25.3 flalign environment . . . . . . . . . . . . . . . . . . . . . . . 54
25.4 xalignat environment . . . . . . . . . . . . . . . . . . . . . . 55
25.5 xxalignat environment . . . . . . . . . . . . . . . . . . . . . 56
25.6 aligned environment . . . . . . . . . . . . . . . . . . . . . . . 56
25.7 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

26 Other environments 57
26.1 gather environment . . . . . . . . . . . . . . . . . . . . . . . 57
26.2 gathered environment . . . . . . . . . . . . . . . . . . . . . . 57
26.3 multline environment . . . . . . . . . . . . . . . . . . . . . . 59
26.4 split environment . . . . . . . . . . . . . . . . . . . . . . . . 61
26.5 Specials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
26.6 cases environment . . . . . . . . . . . . . . . . . . . . . . . . 63
26.7 Matrix environments . . . . . . . . . . . . . . . . . . . . . . . 64

27 Vertical whitespace 65

28 Dots 65

29 fraction commands 66
29.1 Standard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
29.2 Binoms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

30 Roots 68
30.1 Roots with \smash command . . . . . . . . . . . . . . . . . . 68

31 Accents 68

32 \mod command 69

33 Equation numbering 69
33.1 Subequations . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

Mathmode.tex 4
CONTENTS CONTENTS

34 Labels and tags 71

35 Limits 71
35.1 Multiple limits . . . . . . . . . . . . . . . . . . . . . . . . . . 72
35.2 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
35.3 \sideset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

36 Operator names 74

37 Text in math mode 75

37.1 \text command . . . . . . . . . . . . . . . . . . . . . . . . . 75
37.2 \intertext command . . . . . . . . . . . . . . . . . . . . . . 76

38 Extensible arrows 77

39 Frames 78

40 Greek letters 79

41 Miscellaneous commands 79

42 Problems with amsmath 79

III TEX and math 82

43 Length registers 82
43.1 \abovedisplayshortskip . . . . . . . . . . . . . . . . . . . . 82
43.2 \abovedisplayskip . . . . . . . . . . . . . . . . . . . . . . . 82
43.3 \belowdisplayshortskip . . . . . . . . . . . . . . . . . . . . 82
43.4 \belowdisplayskip . . . . . . . . . . . . . . . . . . . . . . . 82
43.5 \delimiterfactor . . . . . . . . . . . . . . . . . . . . . . . . 82
43.6 \delimitershortfall . . . . . . . . . . . . . . . . . . . . . . 83
43.7 \displayindent . . . . . . . . . . . . . . . . . . . . . . . . . 83
43.8 \displaywidth . . . . . . . . . . . . . . . . . . . . . . . . . . 84
43.9 \mathsurround . . . . . . . . . . . . . . . . . . . . . . . . . . 84
43.10\medmuskip . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
43.11\mkern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
43.12\mskip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
43.13\muskip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
43.14\muskipdef . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
43.15\nonscript . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
43.16\nulldelimiterspace . . . . . . . . . . . . . . . . . . . . . . 85
43.17\predisplaysize . . . . . . . . . . . . . . . . . . . . . . . . . 85
43.18\scriptspace . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Mathmode.tex 5
CONTENTS CONTENTS

43.19\thickmuskip . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
43.20\thinmuskip . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
43.21\medmuskip . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

44 Math font macros 86

44.1 \delcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
44.2 \delimiter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
44.3 \displaystyle . . . . . . . . . . . . . . . . . . . . . . . . . . 86
44.4 \fam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
44.5 \mathaccent . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
44.6 \mathbin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
44.7 \mathchar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
44.8 \mathchardef . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
44.9 \mathchoice . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
44.10\mathclose . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
44.11\mathcode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
44.12\mathop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
44.13\mathopen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
44.14\mathord . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
44.15\mathpunct . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
44.16\mathrel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
44.17\scriptfont . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
44.18\scriptscriptfont . . . . . . . . . . . . . . . . . . . . . . . 89
44.19\scriptscriptstyle . . . . . . . . . . . . . . . . . . . . . . . 90
44.20\scriptstyle . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
44.21\skew . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
44.22\skewchar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
44.23\textfont . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
44.24\textstyle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

45 Math macros 90
45.1 \above . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
45.2 \abovewithdelims . . . . . . . . . . . . . . . . . . . . . . . . 91
45.3 \atop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
45.4 \atopwithdelims . . . . . . . . . . . . . . . . . . . . . . . . . 91
45.5 \displaylimits . . . . . . . . . . . . . . . . . . . . . . . . . 91
45.6 \eqno . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
45.7 \everydisplay . . . . . . . . . . . . . . . . . . . . . . . . . . 92
45.8 \everymath . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
45.9 \left . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
45.10\leqno . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
45.11\limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
45.12\mathinner . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
45.13\nolimits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

Mathmode.tex 6
CONTENTS CONTENTS

45.14\over . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
45.15\overline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
45.16\overwithdelims . . . . . . . . . . . . . . . . . . . . . . . . . 93
45.17\radical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
45.18\right . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
45.19\underline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
45.20\vcenter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

46 Math penalties 94
46.1 \binoppenalty . . . . . . . . . . . . . . . . . . . . . . . . . . 94
46.2 \displaywidowpenalty . . . . . . . . . . . . . . . . . . . . . 94
46.3 \postdisplaypenalty . . . . . . . . . . . . . . . . . . . . . . 95
46.4 \predisplaypenalty . . . . . . . . . . . . . . . . . . . . . . . 95
46.5 \relpenalty . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

IV Other packages 96

47 List of available math packages 96

47.1 accents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
47.2 amscd – commutative diagrams . . . . . . . . . . . . . . . . . 97
47.3 amsopn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
47.4 bigdel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
47.5 bm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
47.6 braket . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
47.7 cancel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
47.8 cool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
47.9 delarray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
47.10empheq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
47.11esint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
47.12eucal and euscript . . . . . . . . . . . . . . . . . . . . . . . 105
47.13exscale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
47.14mathtools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
47.15nicefrac . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
47.16relsize . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
47.17xypic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

V Math fonts 109

48 Computer modern 109

49 Latin modern 109

50 Palatino 110

Mathmode.tex 7
CONTENTS CONTENTS

51 Palatino – microimp 110

52 cmbright 111

53 minion 111

VI Special symbols 112

54 Integral symbols 112

55 Harpoons 113

56 Bijective mapping arrow 114

57 Stacked equal sign 114

58 Other symbols 115

VII Examples 116

59 Tuning math typesetting 116

60 Matrix 117
60.1 Identity matrix . . . . . . . . . . . . . . . . . . . . . . . . . . 117
60.2 System of linear equations . . . . . . . . . . . . . . . . . . . . 117
60.3 Matrix with comments on top . . . . . . . . . . . . . . . . . . 118

61 Cases structure 118

61.1 Cases with numbered lines . . . . . . . . . . . . . . . . . . . . 118

62 Arrays 119
62.1 Quadratic equation . . . . . . . . . . . . . . . . . . . . . . . . 120
62.2 Vectors and matrices . . . . . . . . . . . . . . . . . . . . . . . 121
62.3 Cases with (eqn)array environment . . . . . . . . . . . . . . . 121
62.4 Arrays inside arrays . . . . . . . . . . . . . . . . . . . . . . . 122
62.5 Colored cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
62.6 Boxed rows and columns . . . . . . . . . . . . . . . . . . . . . 124

63 Over- and underbraces 125

63.1 Braces and roots . . . . . . . . . . . . . . . . . . . . . . . . . 125
63.2 Overlapping braces . . . . . . . . . . . . . . . . . . . . . . . . 125
63.3 Vertical alignment . . . . . . . . . . . . . . . . . . . . . . . . 126
63.4 Alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Mathmode.tex 8
CONTENTS CONTENTS

64 Integrals 128

65 Vertical alignment 128

65.1 Example 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
65.2 Example 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

66 Node connections 132

67 Special Placement 133

67.1 Formulas side by side . . . . . . . . . . . . . . . . . . . . . . . 133
67.2 Itemize environment . . . . . . . . . . . . . . . . . . . . . . . 135

VIII Lists, bibliography and index 137

List of Figures 138

List of Tables 139

Bibliography 140

Index 142

Mathmode.tex 9
 2 THE INLINEMODE

Part I
Standard LATEX math mode
1 Introduction
The following sections describe all the math commands which are available
without any additional package. Most of them also work with special packages
and some of them are redefined. At first some important facts for typesetting
math expressions.

2 The Inlinemode
As the name says this are always ´ b math expressions which are in a standard
textline, like this one: f (x) = a sinx x dx. There are no limitations for the
height of the math expressions, so that the layout may  be very lousy
 if you
a b c
insert a big matrix in an inline mode like this: A =  d e f . In this
g h i
 
a b c
case it is better to use the \smallmatrix environment A = d e f from the
g h i
AMSmath package (see section 26.7 on page 64) or the displaymath mode
(section 3 on page 13).
This inline mode is possible with three different commands:

1 \(\ sum _{ i =1}^{ n } i =\ frac {1}{2} n \ cdot ( n +1) \)

Pn 1
i=1 i = 2 n · (n + 1)
\\[10 pt ]
2 $\ sum _{ i =1}^{ n } i =\ frac {1}{2} n \ cdot ( n +1)
Pn 1 $\\[10 pt ]
i=1 i = 2 n · (n + 1) 3 \ begin { math }
Pn 1 4 \ sum _{ i =1}^{ n } i =\ frac {1}{2} n \ cdot ( n +1)
i=1 i = 2 n · (n + 1) 5 \ end { math }

1. \( ... \) , the problem is that \( is not a robust macro (see sec- \(...\)
tion 2.3 on the following page).

2. $ ... $ $...$

3. \begin{math} ... \end{math}, also not robust \begin{math}

...
In general $...$ is the best choice, but this does not work in environments \end{math}
like verbatim or alltt. In this case \(...\) works.

Mathmode.tex 10
2.1 Limits 2 THE INLINEMODE

2.1 Limits
In the inline mode the limits are by default only in super or subscript mode
and 1
´ ∞ 1the fractions are always in the scriptstyle font size. For example:
1 x2 dx = 1, which is not too big for the textline. You can change this \limits
with the command \limits, which must follow a math operator2 like an \int
integral (\int), a sum (\sum), a product (\prod) or a limes (\lim). But this \lim
´∞ 1 \prod
x2
dx = 1 does not look very nice in a text line when it appears between \sum
1
two lines, especially when there are multiline limits.3

2.2 Fraction command

For inlined formulas the fractions are by default in the scriptstyle (see tabular 8
a
on page 38), which is good for typesetting y = b+1 , because the linespacing
is nearly the same, but not optimal, when the formula shows some important \frac
facts. There are two solutions to get a better reading:
1. choose the display mode instead of the inline mode, which is the better
one;
a
2. set the fontstyle to \displaystyle, which makes the fraction y =
b+1
more readable but the linespacing increases which is always a bad
solution and should only be used when the first solution makes no
sense.4

a a $ y =\ frac { a }{ b +1}={\ displaystyle \ frac { a }{ b +1}}$

y= b+1 = 1
b+1

2.3 Math in \part, \chapter, \section, ... titles like f (x) =

Q n 1


i=1 i − 2i

All commands which appear in positions like contents, index, header, ... must
be robust5 which is the case for $...$ but not for \(...\). If you do not
have any contents, index, a.s.o. you can write the mathstuff in \chapter,
\section, a.s.o without any restriction. Otherwise use \protect\( and
\protect\) or the $...$ version.
The whole math expression appears in the default font shape and not in
bold like the other text. Section 22.1 on page 48 describes how the math
expressions can be printed also in bold. \texorpdfstring
1
See section 12 on page 38.
2
To define a new operator see page 74
3
For more information about limits see section 6.1 on page 24 or section 35 on page 71.
4
For an abbreviation see section 29 on page 66, there is a special \dfrac macro.
5
robust means that the macro is not expanded before it is moved into for example the
tableofcontents file (*.toc). No robustness is often a problem, when a macro is part of
another macro.

Mathmode.tex 11
2.4 Equation numbering 2 THE INLINEMODE

There are problems with the hyperref package when there is no text
part in a title. It is possible with the command \texorpdfstring to tell
hyperref to use different commands, one for the title and another one for
the bookmarks:

\texorpdfstring{<TeX part>}{<hyperref part>}

1 \ texorpdfstring {$\ int f ( x ) \ ,\ mathrm { d } x $}{ Integral function }

2.4 Equation numbering

It is obvious that the numbering of inline mathstuff makes no sense!

2.5 Framed math

With the \fbox macro everything of inline math can be framed, like the
following one:

Qn 1 \ fbox {$ f ( x ) =\ prod _{ i =1}^{ n }\ left (i -\ frac {1}{2 i

1


f (x) = i=1 i− 2i }\ right ) $}

Parameters are the width of \fboxsep and \fboxrule, the predefined

values from the file latex.ltx are:

1 \ fboxsep = 3 pt
2 \ fboxrule = .4 pt
Qn 1


The same is possible with the \colorbox f (x) = i=1 i− 2i from the
color package.

1 \ colorbox { yellow }{$ f ( x ) =\ prod _{ i =1}^{ n }\ left (i -\ frac {1}{2 i }\ right ) $}

2.6 Linebreak
LATEX can break an inline formula only when a relation symbol (=, <, >, . . .)
or a binary operation symbol (+, −, . . .) exists and at least one of these
symbols appears at the outer level of a formula. Thus $a+b+c$ can be broken
across lines, but ${a+b+c}$ not.

• The default: f (x) = an xn + an−1 xn−1 + an−2 xn−2 + . . . + ai xi +

a2 x2 + a1 x1 + a0

• The same inside a group {...}: f (x) = an xn + an−1 xn−1 + an−2 xn−2 + . . . + ai xi + a2 x2 + a1 x1 +

• Without any symbol: f (x) = an (an−1 (an−2 (. . .) . . .) . . .)

Mathmode.tex 12
 2.7 Whitespace 3 DISPLAYMATH MODE

If it is not possible to have any mathsymbol, then split the inline formula
in two or more pieces ($...$ $...$). If you do not want a linebreak for the
whole document, you can set in the preamble:

\relpenalty=9999
\binoppenalty=9999

which is the extreme case of grudgingly allowing breaks in extreme cases.

2.7 Whitespace
LATEX defines the length \mathsurround with the default value of 0pt. This
length is added before and after an inlined math expression (see table 1).

´∞ 1
1 foo \ fbox {$ f ( x ) =\ int _1^{\ infty }\ frac {1}{ x
foo f (x) = 1 x2
dx = 1 bar ^2}\ ,\ mathrm { d } x =1 $} bar

foo \ rule {20 pt }{\ ht \ strutbox }\ fbox {$ f ( x )

´∞
1
1 =\ int _1^{\ infty }\ frac {1}{ x ^2}\ ,\ mathrm { d
foo f (x) = 1 x2
dx = 1 bar
} x =1 $}\ rule {20 pt }{\ ht \ strutbox } bar

\ setlength {\ mathsurround }{20 pt }

´∞
1
1 foo \ fbox {$ f ( x ) =\ int _1^{\ infty }\ frac {1}{ x
foo f (x) = 1 x2
dx = 1 bar 2
^2}\ ,\ mathrm { d } x =1 $} bar

Table 1: Meaning of \mathsurround

2.8 AMSmath for the inline mode

None of the AMSmath-functions are available in inline mode.

3 Displaymath mode
This means, that every formula gets its own paragraph (line). There are some
differences in the layout to the one from the title of 2.3.

3.1 equation environment

For example:

1 \ begin { equation }
n   2 f ( x ) =\ prod _{ i =1}^{ n }\ left (i -\ frac
Y 1 {1}{2 i }\ right )
f (x) = i− (1)
2i 3 \ end { equation }
i=1

Mathmode.tex 13
3.2 eqnarray environment 3 DISPLAYMATH MODE

The delimiters \begin{equation} ... \end{equation} are the only

difference to the inline version. There are some equivalent commands for the
display-math mode: \begin{displaymath}
...
1. \begin{displaymath}. . . \end{displaymath}, same as \[ . . . \] \end{displaymath}

2. \[...\]. (see above) the short form of a displayed formula, no number \[...\]
n  
Y 1
f (x) = i−
2i
i=1

displayed, no number. Same as 1.

3. \begin{equation}...\end{equation} \begin{equation}
...
n  
Y 1 \end{equation}
f (x) = i− (2)
2i
i=1

displayed, a sequential equation number, which may be reset when

starting a new chapter or section.

(a) There is only one equation number for the whole environment. \nonumber
(b) There exists no star-version of the equation environment because
\[. . . \] is the equivalent. With the tag \nonumber it is possible
to suppress the equation number:
1 \ begin { equation }
2 f ( x ) = [...] \ nonumber
f (x) = [...] 3 \ end { equation }

3.2 eqnarray environment

This is by default an array with three columns and as many rows as you like.
It is nearly the same as an array with a rcl column definition. \begin{eqnarray}
It is not possible to change the internal behaviour of the eqnarray ...
environment without rewriting the environment. It is always an implicit array \end{eqnarray}
with three columns and the horizontal alignment right-center-left (rcl)
and small symbol sizes for the middle column. All this can not be changed
by the user without rewriting the whole environment in latex.ltx.

1 \ begin { eqnarray *}
2 \ mathrm { left } & \ mathrm { middle } & \
left middle right mathrm { right }\\
√ 3 \ frac {1}{\ sqrt { n }}= & \ frac {\ sqrt { n
1 n n
√ = n =
√ }}{ n }= & \ frac { n }{ n \ sqrt { n }}
n n n 4 \ end { eqnarray *}

Mathmode.tex 14
3.2 eqnarray environment 3 DISPLAYMATH MODE

The eqnarray environment should not be used as an array. As seen in

the above example the typesetting is wrong for the middle column. The
numbering of eqnarray environments is always for every row, means, that
four lines get four different equation numbers (for the labels see section 3.4
on page 18):

1 \ begin { eqnarray }
2 y & = & d \ label { eq :2}\\
y = d (3) 3 y & = & cx + d \\
y = cx + d (4) 4 y & = & bx ^{2}+ cx + d \\
5 y & = & ax ^{3}+ bx ^{2}+ cx +
y = bx2 + cx + d (5) d \ label { eq :5}
y = ax3 + bx2 + cx + d (6) 6 \ end { eqnarray }

Toggling numbering off/on for all rows is possible with the starred version
of eqnarray.
1 \ begin { eqnarray *}
2 y & = & d \ label { eq :3}\\
y = d 3 y & = & cx + d \\
y = cx + d 4 y & = & bx ^{2}+ cx + d \\
5 y & = & ax ^{3}+ bx ^{2}+ cx + d \
y = bx2 + cx + d label { eq :4}
y = ax3 + bx2 + cx + d 6 \ end { eqnarray *}

Toggling off/on for single rows is possible with the above mentioned
\nonumber tag at the end of a row (before the newline command). For
example:
1 \ begin { eqnarray }
2 y & = & d \ nonumber \\
y = d 3 y & = & cx + d \ nonumber \\
y = cx + d 4 y & = & bx ^{2}+ cx + d \
nonumber \\
y = bx2 + cx + d 5 y & = & ax ^{3}+ bx ^{2}+ cx + d
3 2 \ end { eqnarray }
y = ax + bx + cx + d (7) 6

3.2.1 Short commands

It is possible to define short commands for the eqnarray environment
1 \ makeatletter
2 \ newcommand {\ be }{ %
3 \ begingroup
4 % \ setlength {\ arraycolsep }{2 pt }
5 \ eqnarray %
6 \ @ifstar {\ nonumber }{} %
7 }
8 \ newcommand {\ ee }{\ endeqnarray \ endgroup }
9 \ makeatother

Mathmode.tex 15
3.3 Equation numbering 3 DISPLAYMATH MODE

Now you can write the whole equation as

ˆ 1 \be
sin x 2 f(x) &=& \int\frac{\sin x}{x}\,\mathrm{d}x
f (x) = dx (8)
x 3 \ee

or, if you do not want to have a numbered equation as

ˆ 1 \be*
sin x 2 f(x) &=& \int\frac{\sin x}{x}\,\mathrm{d}x
f (x) = dx
x 3 \ee

3.3 Equation numbering

For all equations which can have one or more equation numbers (for every \nonumber
line/row) the numbering for the whole equation can be disabled with switching
from the unstarred to the star version. This is still for the whole formula and
doesn’t work for single rows. In this case use the \nonumber tag.
• This doc is written with the article-class, which counts the equations
continuously over all parts/sections. You can change this behaviour in
different ways (see the following subsections).
• In standard LATEX it is a problem with too long equations and the
equation number, which may be printed with the equation one upon
the other. In this case use the AMSmath package, where the number is
set above or below of a too long equation (see equation 28 on page 28).
• For counting subequations see section 33.1 on page 70.

3.3.1 Changing the style

\theequation
With the beginning of Section 25.2 on page 53 the counting changes from
“44” into the new style “II-51”. The command sequence is
1 \ renewcommand \ theequation {\ thepart -\ arabic { equation }}

See section 33 on page 69 for the AMSmath command.

3.3.2 Resetting a counter style

Removing a given reset is possible with the remreset.6 Write into the
preamble \@removefromreset
1 \ makeatletter
2 \ @removefromreset { equation }{ section }
3 \ makeatother

or anywhere in the text.

Now the equation counter is no longer reset when a new section starts.
You can see this after section 26.4 on page 61.
6
CTAN://macros/latex/contrib/supported/carlisle/remreset.sty

Mathmode.tex 16
3.3 Equation numbering 3 DISPLAYMATH MODE

3.3.3 Equation numbers on the left side

Choose package leqno7 or have a look at your document class, if such an
option exists.

3.3.4 Changing the equation number style

The number style can be changed with a redefinition of

\def\@eqnnum{{\normalfont \normalcolor (\theequation)}}

For example: if you want the numbers not in parentheses write

1 \ makeatletter
2 \ def \ @eqnnum {{\ normalfont \ normalcolor \ theequation }}
3 \ makeatother

For AMSmath there is another macro, see section 33 on page 69.

3.3.5 More than one equation counter

You can have more than the default equation counter. With the following
code you can easily toggle between roman and arabic equation counting.
1 % code by Heiko Oberdiek
2 \ makeatletter
3 % Roman counter
4 \ newcounter { roem }
5 \ renewcommand {\ theroem }{\ roman { roem }}
6
7 % save the original counter
8 \ newcommand {\ c@org@eq }{}
9 \ let \ c@org@eq \ c@equation
10 \ newcommand {\ org@theeq }{}
11 \ let \ org@theeq \ theequation
12
13 % \ setroem sets roman counting
14 \ newcommand {\ setroem }{
15 \ let \ c@equation \ c@roem
16 \ let \ theequation \ theroem }
17
18 % \ setarab the arabic counting
19 \ newcommand {\ setarab }{
20 \ let \ c@equation \ c@org@eq
21 \ let \ theequation \ org@theeq }
22 \ makeatother

The following examples show how it works:

7
CTAN://macros/latex/unpacked/leqno.sty

Mathmode.tex 17
3.4 Labels 3 DISPLAYMATH MODE

1 \ begin { align }
2 f ( x ) &= \ int \ sin x \ ,\ mathrm { d } x \ label { eq :
arab 1}\\
ˆ 3 g ( x ) &= \ int \ frac {1}{ x }\ ,\ mathrm { d } x
f (x) = sin x dx (9) 4 \ end { align }
ˆ 5 %
1 6 \ setroem
g(x) = dx (10)
x 7 %
8 \ begin { align }
9 F ( x ) &= -\ cos x \\
F (x) = − cos x (i) 10 G ( x ) &=\ ln x \ label { eq : rom 1}
11 \ end { align }
G(x) = ln x (ii) 12 %
13 \ setarab
14 %
f 0 (x) = sin x (11) 15 \ begin { align }
1 16 f ^{\ prime } ( x ) &= \ sin x \\
g 0 (x) = (12) 17 g ^{\ prime } ( x ) &= \ frac {1}{ x }\ label { eq :
x arab 2}
18 \ end { align }

There can be references to these equations in the usual way, like eq.9, 12
and for the roman one eq.ii.

3.4 Labels
Every numbered equation can have a label to which a reference is possible.

• There is one restriction for the label names, they cannot include one of
LATEX’s command characters.8
• The label names are replaced by the equation number.
\tag
If you do not want a reference to the equation number but to a self defined
name then use the AMSmath command \tag..., which is described in
section 34 on page 71.

3.5 Frames
Similiar to the inline mode, displayed equations can also be framed with
the \fbox command, like equation 13. The only difference is the fact, that
the equation must be packed into a parbox or minipage. It is nearly the
same for a colored box, where the \fbox{...} has to be replaced with
\colorbox{yellow}{...}. The package color.sty must be loaded and
–important – the calc package to get a correct boxwidth.
ˆ ∞
1
f (x) = 2
dx = 1 (13)
1 x

8
$_ˆ\&%{}

Mathmode.tex 18
3.5 Frames 3 DISPLAYMATH MODE

1 \ noindent \ fbox {\ parbox {\ linewidth -2\ fboxsep -2\ fboxrule }{ %

2 \ begin { equation }\ label { eq : frame 0}
3 f ( x ) =\ int _1^{\ infty }\ dfrac {1}{ x ^2}\ ,\ mathrm { d } x =1
4 \ end { equation } %
5 }}

If the equation number should not be part of the frame, then it is a

bit complicated. There is one tricky solution, which puts an unnumbered
equation just beside an empty numbered equation. The \hfill is only useful
for placing the equation number right aligned, which is not the default. The
following four equations 14-17 are the same, only the second one written
with the \myMathBox macro which has the border and background color as
optional arguments with the defaults white for background and black for
the frame. If there is only one optional argument, then it is still the one for
the frame color (15).
1 \ makeatletter
2 \ def \ myMathBox {\ @ifnextchar [{\ my@MBoxi }{\ my@MBoxi [ black ]}}
3 \ def \ my@MBoxi [#1]{\ @ifnextchar [{\ my@MBoxii [#1]}{\ my@MBoxii
[#1][ white ]}}
4 \ def \ my@MBoxii [#1][#2]#3#4{ %
5 \ par \ noindent %
6 \ fcolorbox {#1}{#2}{ %
7 \ parbox {\ linewidth -\ labelwidth -2\ fboxrule -2\ fboxsep }{#3}
%
8 }%
9 \ parbox {\ labelwidth }{ %
10 \ begin { eqnarray }\ label {#4}\ end { eqnarray } %
11 }%
12 \ par %
13 }
14 \ makeatother

f (x) = x2 + x (14)

f (x) = x2 + x (15)

f (x) = x2 + x (16)

f (x) = x2 + x (17)

1 \ begin { equation }\ label { eq : frame 2}

2 f ( x ) = x ^2 + x
3 \ end { equation }
4 \ myMathBox [ red ]{\[ f ( x ) = x ^2 + x \]}{ eq : frame 3}

Mathmode.tex 19
 4 ARRAY ENVIRONMENT

5 \ myMathBox [ red ][ yellow ]{\[ f ( x ) = x ^2 + x \]}{ eq : frame 4}

6 \ myMathBox {\[ f ( x ) = x ^2 + x \]}{ eq : frame 5}

If you are using the AMSmath package, then try the solutions from
section 39 on page 78.

4 array environment
\begin{array}
This is simply the same as the eqnarray environment only with the possibility ...
of variable rows and columns and the fact, that the whole formula has \end{array}
only one equation number and that the array environment can only be
part of another math environment, like the equation environment or the
displaymath environment. With @{} before the first and after the last colomn
the additional space \arraycolsep is not used, which maybe important when
using left aligned equations.

a) y = c (constant) 
b) y = cx + d (linear)

Polynomes (18)
c) y = bx2 + cx + d (square)
d) y = ax3 + bx2 + cx + d (cubic)


1 \ begin { equation }
2 \ left . %
3 \ begin { array }{ @ {} r@ {\ quad } ccrr@ {}}
4 \ textrm { a }) & y & = & c & ( constant ) \\
5 \ textrm { b }) & y & = & cx + d & ( linear ) \\
6 \ textrm { c }) & y & = & bx ^{2}+ cx + d & ( square ) \\
7 \ textrm { d }) & y & = & ax ^{3}+ bx ^{2}+ cx + d & ( cubic )
8 \ end { array } %
9 \ right \} \ textrm { Polynomes }
10 \ end { equation }

The horizontal alignment of the columns is the same as the one from the
tabular environment.
For arrays with delimiters see section 47.9 on page 103.

4.1 Cases structure

If you do not want to use the AMSmath package then write your own cases
structure with the array environment:

Mathmode.tex 20
 4.2 arraycolsep 4 ARRAY ENVIRONMENT

1 \begin{equation}
2 x=\left\{ \begin{array}{cl}
3 0 & \textrm{if A=...}\\
4 1 & \textrm{if B=...}\\
5 x & \textrm{this runs with as much text as you like, but without an
raggeright text.}\end{array}\right.
6 \end{equation}


 0 if A=...
x= 1 if B=...
x this runs with as much text as you like, but without an raggeright text.

(19)
It is obvious, that we need a \parbox if the text is longer than the possible
linewidth.
1 \begin{equation}
2 x = \left\{%
3 \begin{array}{l>{\raggedright}p{.5\textwidth}}%
4 0 & if A=...\tabularnewline
5 1 & if B=...\tabularnewline
6 x & \parbox{0.5\columnwidth}{this runs with as much text as you like
, %
7 because an automatic linebreak is given with %
8 a raggedright text. Without this %
9 \raggedright command, you’ll get a formatted %
10 text like the following one ... but with a parbox ... it works}
11 \end{array}%
12 \right. %
13 \end{equation}



 0 if A=...
1 if B=...







 this runs with as much text as you
like, because an automatic linebreak is

x= (20)
given with a raggedright text.
x


Without this command, you’ll get a







 formatted text like the following one

... but with a parbox ... it works

4.2 arraycolsep
\arraycolsep
All the foregoing math environments use the array to typeset the math
expression. The predefined separation between two columns is the length
\arraycolsep|, which is set by nearly all document classes to 5pt, which
seems to be too big. The following equation is typeset with the default value
and the second one with \arraycolsep=1.4pt
ˆ
sin x
f (x) = dx
x

Mathmode.tex 21
 5 MATRIX

ˆ
sin x
f (x) = dx
x
If this modification should be valid for all arrays/equations, then write
it into the preamble, otherwise put it into a group or define your own
environment as done in section 3.2.1 on page 15.
1 \ bgroup
2 \ arraycolsep =1.4 pt
3 \ begin { eqnarray }
4 f ( x ) & = & \ int \ frac {\ sin x }{ x }\ ,\ mathrm { d } x
5 \ end { eqnarray }
6 \ egroup

1 \ makeatletter
2 \ newcommand {\ be }{ %
3 \ begingroup
4 \ setlength {\ arraycolsep }{1.4 pt }
5 [ ... ]

5 Matrix
\begin{matrix}
TEX knows two macros and LATEX one more for typesetting a matrix: ...
\end{matrix}
1 $\begin{matrix} \bordermatrix
A B C 2 A & B & C \\
d e f 3 d & e & f \\
4 1 & 2 & 3 \\
1 2 3 5 \end{matrix}$

1 $\bordermatrix{%
0 1 2 2 & 0 & 1 & 2 \cr
0 & A & B & C \cr
  3
0 A B C
4 1 & d & e & f \cr
1 d e f 5 2 & 1 & 2 & 3 \cr
2 1 2 3 6 }$

The first two macros are listed here for some historical reason, because
the array environment or especially the AMSmath package offers the same
or better macros/environments. Nevertheless it is possible to redefine the
\bordermatrix macro to get other parentheses and a star version which takes
the left top part as matrix:
1 $\bordermatrix{%
1 2 2 & 1 & 2 \cr
1 & x1 & x2 \cr
  3
1 x1 x2
4 2 & x3 & x4 \cr
2  x3 x4  5 3 & x5 & x6
3 x5 x6 6 }$

Mathmode.tex 22
 5 MATRIX

1 $\bordermatrix[{[]}]{%
1 2 2 & 1 & 2 \cr
1 & x1 & x2 \cr
  3
1 x1 x2
4 2 & x3 & x4 \cr
2  x3 x4  5 3 & x5 & x6
3 x5 x6 6 }$

1 $\bordermatrix[\{\}]{%
1 2 2 & 1 & 2 \cr
  3 1 & x1 & x2 \cr
1  x1 x2 
4 2 & x3 & x4 \cr
2 x3 x4 5 3 & x5 & x6
3 x5 x6
 
6 }$

1 $\bordermatrix*{%
  2 x1 & x2 & 1 \cr
x1 x2 1 3 x3 & x4 & 2 \cr
 x3 x4  2 4 x5 & x6 & 3 \cr
5 1 & 2
x5 x6 3 6 }$
1 2
1 $\bordermatrix*[{[]}]{%
  2 x1 & x2 & 1 \cr
x1 x2 1 3 x3 & x4 & 2 \cr
 x3 x4  2 4 x5 & x6 & 3 \cr
5 1 & 2
x5 x6 3 6 }$
1 2
1 $\bordermatrix*[\{\}]{%
  2 x1 & x2 & 1 \cr
 x1 x2 1 3 x3 & x4 & 2 \cr
x3 x4 2 4 x5 & x6 & 3 \cr
  5 1 & 2
x5 x6 3 6 }$
1 2
There is now an optional argument for the parenthesis with () as the
default one. To get such a behaviour, write into the preamble:
1 \ makeatletter
2 \ newif \ if@borderstar
3 \ def \ bordermatrix {\ @ifnextchar *{ %
4 \ @borderstartrue \ @bordermatrix@i }{\ @borderstarfalse \ @bordermatrix@i
*} %
5 }
6 \ def \ @bordermatrix@i *{\ @ifnextchar [{\ @bordermatrix@ii }{\
@bordermatrix@ii [() ]}}
7 \ def \ @bordermatrix@ii [#1]#2{ %
8 \ begingroup
9 \ m@th \ @tempdima 8.75\ p@ \ setbox \ z@ \ vbox { %
10 \ def \ cr {\ crcr \ noalign {\ kern 2\ p@ \ global \ let \ cr \ endline }} %
11 \ ialign {$##$\ hfil \ kern 2\ p@ \ kern \ @tempdima & \ thinspace %
12 \ hfil $##$\ hfil && \ quad \ hfil $##$\ hfil \ crcr \ omit \ strut %
13 \ hfil \ crcr \ noalign {\ kern -\ baselineskip }#2\ crcr \ omit %

Mathmode.tex 23
 6 SUPER/SUBSCRIPT AND LIMITS

14 \ strut \ cr }} %
15 \ setbox \ tw@ \ vbox {\ unvcopy \ z@ \ global \ setbox \ @ne \ lastbox } %
16 \ setbox \ tw@ \ hbox {\ unhbox \ @ne \ unskip \ global \ setbox \ @ne \ lastbox } %
17 \ setbox \ tw@ \ hbox { %
18 $\ kern \ wd \ @ne \ kern -\ @tempdima \ left \ @firstoftwo #1 %
19 \ if@borderstar \ kern 2 pt \ else \ kern -\ wd \ @ne \ fi %
20 \ global \ setbox \ @ne \ vbox {\ box \ @ne \ if@borderstar \ else \ kern 2\ p@ \ fi } %
21 \ vcenter {\ if@borderstar \ else \ kern -\ ht \ @ne \ fi %
22 \ unvbox \ z@ \ kern -\ if@borderstar 2\ fi \ baselineskip } %
23 \ if@borderstar \ kern -2\ @tempdima \ kern 2\ p@ \ else \ ,\ fi \ right \
@secondoftwo #1 $ %
24 }\ null \;\ vbox {\ kern \ ht \ @ne \ box \ tw@ } %
25 \ endgroup
26 }
27 \ makeatother

The matrix environment macro cannot be used together with the AMSmath
package, it redefines this environment (see section 26.7 on page 64).

6 Super/Subscript and limits

Writing amin and amax gives the same depth for the subscript, but writing
them in upright mode with \mbox gives a different depth: amin and amax .
The problem is the different height, which can be modified in several ways

• $a_{\mbox{\vphantom{i}max}}: amin and amax ;

• $a_{\mathrm{max}}: amin and amax ;

• $a_{\max}: amin and amax . Both are predefined operators (see sec-
tion 16 on page 43).

6.1 Multiple limits

\atop
For general information about limits read section 2.1 on page 11. With
the TEX command \atop command multiple limits for a \sum or \prod are
possible. The syntax is:

above 1 \[ {above \atop below} \]

below
which is nearly the same as a fraction without a rule. This can be enhanced
to a\atop b\atop c and so on. For equation 21 do the following steps:
1 \begin{equation}\label{eq:atop}
X 2 \sum_{{1\le j\le p\atop {%
aij bjk cki (21)
3 {1\le j\le q\atop 1\le k\le r}}}%
1≤j≤p
1≤j≤q 4 }a_{ij}b_{jk}c_{ki}
1≤k≤r 5 \end{equation}

Mathmode.tex 24
6.2 Problems 7 ROOTS

\shortstack
which is not the best solution because the space between the lines is too
big. The AMSmath package provides several commands for limits (section 35
on page 71) and the \underset and \overset commands (see section 41 on
page 79).

6.2 Problems
X
aij bjk cki (22)
1≤j≤p
1≤j≤q
1≤k≤r

The equation 22 shows that the horizontal alignment is not optimal,

because the math expression on the right follows at the end of the limits
which are a unit together with the sum symbol. There is an elegant solution
with AMSmath, described in subsection 35.2 on page 72. If you do not want
to use AMSmath, then use \makebox. But there is a problem when the
general fontsize is increased, \makebox knows nothing about the actual math
font size. Equation 23a shows the effect and equation 23b the view without
the boxes.
X X
aij bjk cki (23a) aij bjk cki (23b)
1≤j≤p 1≤j≤p
1≤j≤q 1≤j≤q
1≤k≤r 1≤k≤r

1 \ begin { equation }
2 \ sum _{\ makebox [0 pt ]{$ %
3 {{\ scriptscript style 1\ le j \ le p \ atop { %
4 {1\ le j \ le q \ atop 1\ le k \ le r }}}} %
5 $}} a _{ ij } b _{ jk } c _{ ki }
6 \ end { equation }

7 Roots
The square root \sqrt is the default for LATEX and the n-th root can be
inserted with the optional parameter \sqrt[n]{...}. . \sqrt
√
\sqrt{x} x
√
3
\sqrt[3]{x} x

There is a different typesetting in roots. Equation 24 on the following

page has different heights for the roots, whereas equation 25 on the next
page has the same one. This is possible with the \vphantom command, which \vphantom
reserves the vertical space (without a horizontal one) of the parameter height.

Mathmode.tex 25
 8 BRACKETS, BRACES . . .

1 \begin{equation}
2 \sqrt{a}\,%
√ √ q 3 \sqrt{T}\,%
a T 2αkB1 T i (24) 4 \sqrt{2\alpha k_{B_1}T^i}\label{eq:root1}
5 \end{equation}

1 \begin{equation}\label{eq:root2}
q q q 2 \sqrt{a\vphantom{k_{B_1}T^i}}\,%
3 \sqrt{T\vphantom{k_{B_1}T^i}}\,%
a T 2αkB1 T i (25) 4 \sqrt{2\alpha k_{B_1}T^i}
5 \end{equation}

The typesetting looks much better, especially when the formula has
different roots in a row, like equation 24. Using AMSmath with the \smash
command9 gives some more possibilities for the typesetting of roots (see
section 30 on page 68).

8 Brackets, braces and parentheses

This is one of the major problems inside the math mode, because there is often
a need for different brackets, braces and parentheses in different size. At first
we had to admit, that there is a difference between the characters “()[]/\
{} | k bc de hi ↑⇑ ↓⇓ lm” and their use as an argument of the \left and \leftX
\right command, where LATEX stretches the size in a way that everything \rightX
between the pair of left and right parentheses is smaller than the parentheses
themselves. In some cases10 it may be useful to choose a fixed height, which
is possible with the \big-series. Instead of writing \leftX or \rightX one of
the following commands can be chosen:
\bigX
default ()[]/\{}|k bc de hi ↑⇑ ↓⇓lm \BigX

    
 x~ w x~
\bigX  w y y \biggX
 hi ./no  jk lm DE x~ w x~ \BiggX
\BigX
 w w w
 w y y
      x~ w x~
 w w w
\biggX 

w w w
w y y
! "# ,-()  $% &' *+ x~ wx~
 w ww
\BiggX
 w ww
 w ww
 w yy
Only a few commands can be written in a short form like \big(. The “X”
has to be replaced with one of the following characters or commands from
table 3 on page 28, which shows the parentheses character, its code for the
use with one of the “big” commands and an example with the code for that. \biglX
For all commands there exists a left/right version \bigl, \bigr, \Bigl \bigrX
and so on, which only makes sense when writing things like:
9
The \smash command exists also in LATEX but without an optional argument, which
makes the use for roots possible.
10
See section 8.1.1 on page 28 for example.

Mathmode.tex 26
 8 BRACKETS, BRACES . . .

  1 \begin{align}
a
× × (26) 2 \biggl)\times \frac{a}{b} \times\biggr(
b 3 \end{align}
  4 \begin{align}
a 5 \bigg)\times \frac{a}{b} \times\bigg(
× × (27) 6 \end{align}
b

LATEX takes the \biggl) as a mathopen symbol, which has by default

another horizontal spacing.
In addition to the above commands there exist some more: \bigm, \Bigm,
\biggm and \Biggm, which work as the standard ones (without the addtional
“m”) but add some more horizontal space between the delimiter and the \bigmX
formula before and after (see table 2). \bigmX

  
1  3 1 $\bigg(\dfrac{1}{3}\bigg|\dfrac
{3}{4}\bigg)$
34
  
1  3 1 $\bigg(\dfrac{1}{3}\biggm|\dfrac
{3}{4}\bigg)$
34

Table 2: Difference between the default \bigg and the \biggm command

Char Code Example Code

 2

() () 3 a2 + bc 3\Big( aˆ2+bˆ{cˆ2}\Big)
h 2
i
[] [] 3 a2 + bc 3\Big[ aˆ2+bˆ{cˆ2}\Big]
. 2
/
/\ /\backslash 3 a2 + bc 3\Big/
aˆ2+bˆ{cˆ2}\Big\backslash
n 2
o
{} \{\} 3 a2 + bc 3\Big\{ aˆ2+bˆ{cˆ2}\Big\}

 2 c 2
|k | \Vert 3 a + b 3\Big|aˆ2+bˆ{cˆ2}\Big\Vert
j 2
k
bc \lfloor 3 a2 + bc 3\Big\lfloor aˆ2+bˆ{cˆ2}
\rfloor \Big\rfloor
l 2
m
de \lceil\rceil 2
3 a +b c 3\Big\lceil aˆ2+bˆ{cˆ2}
\Big\rceil
D 2
E
hi \langle\rangle3 a2 + bc 3\Big\langle
aˆ2+bˆ{cˆ2}\Big\rangle
x ~
2w
↑⇑ \uparrow 3a2 + bc w 3\Big\uparrow

\Uparrow aˆ2+bˆ{cˆ2}\Big\Uparrow

Mathmode.tex 27
8.1 Examples 8 BRACKETS, BRACES . . .

Char Code Example Code

 w
 2 c 2w
↓⇓ \downarrow 3ya + b  3\Big\downarrow aˆ2+bˆ{cˆ2}
\Downarrow \Big\Downarrow
x ~
2w
lm \updownarrow 3ya2 + bc  3\Big\updownarrow aˆ2+bˆ{cˆ2}

\Updownarrow \Big\Updownarrow

Table 3: Use of the different parentheses for the “big” com-

mands

8.1 Examples
8.1.1 Braces over several lines
The following equation in the single line mode looks like

 
1 X
∆(fij f ij ) = 2  χij (σi − σj )2 + f ij ∇j ∇i (∆f ) + ∇k fij ∇k f ij + f ij f k [2∇i Rjk − ∇k Rij ]
2
i<j
(28)
and is too long for the text width and the equation number has to be placed
under the equation.11 With the array environment the formula can be split
in two smaller pieces:

X
1 ij  χij (σi − σj )2 + f ij ∇j ∇i (∆f )+
2 ∆(fij f ) = 2 (29)
i<j
+∇k fij ∇k f ij + f ij f k [2∇i Rjk − ∇k Rij ]

It is obvious that there is a problem with the right closing parentheses.

Because of the two pairs “\left( ... \right.” and “\left. ... \right)”
they have a different size because every pair does it in its own way. Using
the Bigg command changes this into a better typesetting:

X
1 ij
2 ∆(fij f ) =2 χij (σi − σj )2 + f ij ∇j ∇i (∆f )+
i<j ! (30)
+∇k fij ∇k f ij + f ij f k [2∇ i Rjk − ∇k Rij ]

11
In standard LATEX the equation and the number are printed one over the other for
too long formulas. Only AMSmath puts it one line over (left numbers) or under (right
numbers) the formula.

Mathmode.tex 28
8.2 New delimiters 8 BRACKETS, BRACES . . .

1 {\ arraycolsep =2 pt
2 \ begin { equation }
3 \ begin { array }{ rcl }
4 \ frac {1}{2}\ Delta ( f _{ ij } f ^{ ij }) & = & 2\ Bigg ({\
displaystyle
5 \ sum _{ i < j }}\ chi _{ ij }(\ sigma _{ i } -\ sigma _{ j }) ^{2}+ f ^{ ij } %
6 \ nabla _{ j }\ nabla _{ i }(\ Delta f ) +\\
7 & & +\ nabla _{ k } f _{ ij }\ nabla ^{ k } f ^{ ij }+ f ^{ ij } f ^{ k }[2
8 \ nabla _{ i } R _{ jk } -\ nabla _{ k } R _{ ij }]\ Bigg )
9 \ end { array }
10 \ end { equation }
11 }

Section 26.5 on page 63 shows another solution for getting the right size
for parentheses when breaking the equation in smaller pieces.

" ∞  
µ X Re n
B(r, φ, λ) = Jn Pn (sφ)
r r
n=2
n 
!#
Re n
X 
+ (Cnm cos mλ + Snm sin mλ)Pnm (sφ)
r
m=1

1 \ begin { align *}
2 B (r ,\ phi ,\ lambda ) = & \ ,\ dfrac {\ mu }{ r }
3 \ Bigg [\ sum _{ n =2}^{\ infty } \ Bigg ( \ left ( \ dfrac { R _ e }{ r } \
right ) ^ n J _ nP _ n ( s \ phi ) \\
4 & +\ sum _{ m =1}^ n \ left ( \ dfrac { R _ e }{ r } \ right ) ^ n
5 ( C _{ nm }\ cos m \ lambda + S _{ nm }\ sin m \ lambda ) P _{ nm }( s \ phi ) \
Bigg ) \ Bigg ]
6 \ end { align *}

8.1.2 Middle bar

See section 47.6 on page 99 for examples and the use of package braket.

8.2 New delimiters

The default delimiters are defined in the file fontmath.ltx which is stored in
general in [TEXMF]/tex/latex/base/fontmath.ltx. If we need for example
a thicker vertical symbol than the existing \vert symbol we can define in
the preamble:
1 \ D e c l ar e M a t h D el i m i t e r {\ Norm }
2 {\ mathord }{ largesymbols }{"3 E }{ largesymbols }{"3 E }

The character number 3E16 (decimal 62) from the cmex10 font is the
small thick vertical rule. Now the new delimiter \Norm can be used in the
usual way:

Mathmode.tex 29
8.3 Problems with parentheses 9 TEXT IN MATH MODE

∗BLA∗

  1 $\left\Norm *BLA* \right\Norm$

∗BLA∗

  2

 


 
 ∗BLU B∗ 
 3 $\left\Norm \dfrac{*BLA*}{*BLUB*} \right\Norm$

8.3 Problems with parentheses

\delimitershortfall
It is obvious that the following equation has not the right size of the paren- \delimiterfactor
thesis in the second integral, the inner one should be a bit smaller than the
outer one.
\[
ˆ ˆ
1
β 2 \int_\gamma F’(z) dz =\int_\alpha^\beta
F 0 (z)dz = F 0 (γ(t))·γ 0 (t)dt 3 F’\left(\gamma (t)\right)\cdot\gamma ’(t)dt
γ α 4 \]

The problem is that TEX controlls the height of the parenthesis with
\delimitershortfall and \delimiterfactor, with the default values
\delimitershortfall=5pt
\delimiterfactor=901
\delimiterfactor/1000 is the relative size of the parenthesis for a given
formula environment. They could be of \delimitershortfall too short.
These values are valid at the end of the formula, the best way is to set them
straight before the math environment or globally for all in the preamble.
{\delimitershortfall=-1pt
ˆ ˆ β
1
2 \[
F 0 (z)dz = F0 γ(t) ·γ 0 (t)dt


3 \int_\gamma F’(z) dz =\int_\alpha^\beta
γ α 4 F’\left(\gamma (t)\right)\cdot\gamma ’(t)dt
5 \]}

9 Text in math mode

Standard text in math mode should be written in upright shape and not in the
italic one. This shape is reserved for the variable names: I am text inside math.
(see also ( 7 on page 33). There are different ways to write text inside math. \textstyle
\mbox
• \mathrm. It is like math mode (no spaces), but in upright mode \mathrm
• \textrm. Upright mode with printed spaces (real textmode)
• \mbox. The font size is still the one from \textstyle (see section 12
on page 38), so that you have to place additional commands when you
use \mbox in a super- or subscript for limits.
Inserting long text is possible with a \parbox, which can be aligned as
usual to the top, bottom or center, e.g.,
a + b + c + d + ef = g+h+i+j+k this is a very long (31)
description of a for-
mula

Mathmode.tex 30
 11 SPACE

1 \ begin { eqnarray }
2 a + b + c + d + ef & = & g + h + i + j + k %
3 \ qquad \ textrm {\ parbox [ t ]{.25\ linewidth }{ %
4 this is a very long description of a formula } %
5 }
6 \ end { eqnarray }

Additional commands for text inside math are provided by AMSmath

(see section 37 on page 75).

10 Font commands
10.1 Old-style font commands
Should never be used, but are still present and supported by LATEX. The
default syntax for the old commands is
1 {\ XX test }

Table 4 shows what has to be replaced for the XX. The major difference to
the new style is that these \XX are toggling the actual math mode into the
“XX” one, whereas the new commands start a group which switches back to
the mode before at its end.
\bf test \cal T EST \it test \rm test \tt test

Table 4: Old font style commands

10.2 New-style font commands

\mathrm
The default syntax is \mathfrak
1 \ mathXX { test }
\mathcal
\mathsf
Table 5 shows what has to be replaced for the XX. See section 47.12 on \mathbb
page 105 for additional packages. \mathtt
\mathit
\mathbf
11 Space
11.1 Math typesetting
\thinmuskip
LATEX defines the three math lengths12 with the following values13 : \medmuskip
1 \ thinmuskip =3 mu \thickmuskip
2 \ medmuskip =4 mu plus 2 mu minus 4 mu
3 \ thickmuskip =5 mu plus 5 mu

12
For more information see: http://www.tug.org/utilities/plain/cseq.html
13
see fontmath.ltx

Mathmode.tex 31
11.1 Math typesetting 11 SPACE

Command Test
default ABCDEF GHIJKLM N OP QRST U V W XY Z
abcdef ghijklmnopqrstuvwxyz
\mathfrak ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
\mathcala ABCDEFGHIJ KLMN OPQRST UVWX YZ
\mathsf ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
\mathbba ABCDEFGHIJKLMNOPQRSTUVWXYZ
\mathtt ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
\mathit ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
\mathrm ABCDEFGHIJKLMNOPQRSTUVWXYZ
abcdefghijklmnopqrstuvwxyz
\mathbf ABCDEFGHIJKLMNOPQRSTUVWXYZ

\mathdsb


 !#"$%'&)(+*-,
abcdefghijklmnopqrstuvwxyz

a Not available for lower letters. For mathcal exists a non free font for lower letters
(http://www.yandy.com)
b Needs package dsfont.sty

Table 5: Fonts in math mode

where mu is the abbreviation for math unit.

1
1mu = em
18

default f (x) = x2 + 3x0 · sin x

\thinmuskip=0mu f (x) = x2 + 3x0 · sinx
\medmuskip=0mu f (x) = x2 +3x0 ·sin x
\thickmuskip=0mu f (x)=x2 + 3x0 · sin x
all set to zero f (x)=x2 +3x0 ·sinx

Table 6: The meaning of the math spaces

These lengths can have all glue and are used for the horizontal spacing
in math expressions where TEX puts spaces between symbols and operators.
The meaning of these different horizontal skips is shown in table 6. For a
better typesetting LATEX inserts different spaces between the symbols.

\thinmuskip space between ordinary and operator atoms

\medmuskip space between ordinary and binary atoms in display and text
styles

Mathmode.tex 32
11.2 Additional horizontal spacing 11 SPACE

\thickmuskip space between ordinary and relation atoms in display and

text styles

11.2 Additional horizontal spacing

\thinspace
\medspace
Positive Space Negative Space \thickspace
$ab$ a b \negthinspace
\negmedspace
$a b$ a b \negthickspace
$a\ b$ a b
$a\mbox{\textvisiblespace}b$ a ␣b
$a\,b$ ($a\thinspace b$) a b $a\! b$ a b
$a\: b$ ($a\medspace b$) a b $a\negmedspace b$ ab
$a\; b$ ($a\thickspace b$ a b $a\negthickspace b$ ab
$a\quad b$ a b
$a\qquad b$ a b
$a\hspace{0.5cm}b$ a b $a\hspace{-0.5cm}b$ ba
$a\kern0.5cm b$ a b $a\kern-0.5cm b$ ba
$a\hphantom{xx}b$ a b
$axxb$ a xx b

Table 7: Spaces in math mode

LaTeX defines the following short commands:

\def\>{\mskip\medmuskip}
\def\;{\mskip\thickmuskip}
\def\!{\mskip-\thinmuskip}

In math mode there is often a need for additional tiny spaces between
di di
variables, e.g., L written with a tiny space between L and looks nicer:
dt dt
di
L . Table 7 shows a list of all commands for horizontal space which can be
dt
used in math mode. The “space” is seen “between” the boxed a and b. For
all examples a is \boxed{a} and b is \boxed{b}. The short forms for some
spaces may cause problems with other packages. In this case use the long \hspace
form of the commands. \hphantom
\kern
11.3 Problems
Using \hphantom in mathmode depends to on object. \hphantom reserves only
the space of the exact width without any additional space. In the following

Mathmode.tex 33
11.4 Dot versus comma 11 SPACE

example the second line is wrong: & \hphantom{\rightarrow} b\\. It does

not reserve any additional space.
1 \begin{align*}
a→b 2 a & \rightarrow b\\
b 3 & \hphantom{\rightarrow} b\\
4 & \mkern\thickmuskip\hphantom{\rightarrow}\mkern\thickmuskip b\\
b 5 & \mathrel{\hphantom{\rightarrow}} b
b 6 \end{align*}

This only works when the math symbol is a mathrel one, otherwise you
have to change the horizontal space to \medmuskip or \thinmuskip or to
use an empty group after the \hphantom command. For more informations
about the math objects look into fontmath.ltx or amssymb or use the \show
macro, which prints out the type of the mathsymbol, e.g., \show\rightarrow
with the output:
1 > \ rightarrow =\ mathchar "3221.
2 l .20 \ show \ rightarrow

The first digit represents the type:

0: ordinary
1: large operator
2: binary operation
3: relation
4: opening
5: closing
6: punctuation
7: variable family
Grouping a math symbol can change the behaviour in horizontal spacing.
Compare 50 × 1012 and 50×1012 , the first one is typeset with $50\times10^{12}$
and the second one with $50{\times}10^{12}$. Another possibilty is to use
the numprint package.14

11.4 Dot versus comma

\mathpunct
In difference to a decimal point and a comma as a marker of thousands a lot \mathord
of countries prefer it vice versa. To get the same behaviour the meaning of
dot and comma has to be changed:

1, 234, 567.89 default (32)

1.234.567, 89 vice versa, wrong spacing (33)
1. 234. 567,89 correct spacing (34)

14
CTAN://macros/latex/contrib/numprint/

Mathmode.tex 34
11.5 Vertical whitespace 11 SPACE

1 1 ,234 ,567.89 & \ textrm { default }\\

2 1.234.567 ,89 & \ textrm { vice versa , wrong spacing }\\
3 1\ mathpunct {.}234\ mathpunct {.}567{ ,}89 & \ textrm { correct
spacing }

The original definitions from fontmath.ltx15 are

\DeclareMathSymbol{,}{\mathpunct}{letters}{"3B}
\DeclareMathSymbol{.}{\mathord}{letters}{"3A}

\mathord and \mathpunct can be changed for a documentwide other be-

haviour. In the above equation 34 the comma is only set in a pair of braces
{,}, which is the same as writing \mathord{,} because LATEX handles every-
thing inside of parenthises as a formula, which gets the same spacing.
It is also possible to use the package icomma16 for a documentwide correct
spacing.

11.5 Vertical whitespace

11.5.1 Before/after math expressions
There are four predefined lengths, which control the vertical whitespace of
displayed formulas:

\abovedisplayskip=12pt plus 3pt minus 9pt

\abovedisplayshortskip=0pt plus 3pt
\belowdisplayskip=12pt plus 3pt minus 9pt
\belowdisplayshortskip=7pt plus 3pt minus 4pt

The short skips are used if the formula starts behind the end of the foregoing
last line. Only for demonstration the shortskips are set to 0pt in the following
examples and the normal skips to 20pt without any glue:
The line ends before. ˆ
sin x
f (x) = dx (35)
x
The line doesn’t end before the formula.
ˆ
sin x
f (x) = dx (36)
x

And the next line starts as usual with some text ...

15
Located in texmf/tex/latex/base/
16
CTAN:// macros/latex/contrib/was/

Mathmode.tex 35
11.5 Vertical whitespace 11 SPACE

1 \ a b o v e d i s p l a y s h o r t s k i p =0 pt
2 \ b e l o w d i s p l a y s h o r t s k i p =0 pt
3 \ abovedisplayskip =20 pt
4 \ belowdisplayskip =20 pt
5 \ noindent The line ends before .
6 \ begin { equation }
7 f ( x ) = \ int \ frac {\ sin x }{ x }\ ,\ mathrm { d } x
8 \ end { equation }
9 \ noindent The line doesn ’ t end before the formula .
10 \ begin { equation }
11 f ( x ) = \ int \ frac {\ sin x }{ x }\ ,\ mathrm { d } x
12 \ end { equation }
13 \ noindent And the next line starts as usual with some text
...
fleqn class
When using the fleqn classoption for left aligned equations the math option
environments equation and \[. . . \] are typeset as a list. This is the rea-
son why the vertical space is defined by the length registers for a list, es-
pecially \topsep, instead of \abovedisplayskip and \belowdisplayskip.
This doesn’t effect the eqnarray environment.

11.5.2 Inside math expressions

\\[<length>] This works inside the math mode in the same way as in the
text mode.
\jot
\jot The vertical space between the lines for all math expressions which
allow multiple lines can be changed with the length \jot, which is predefined
as

\newdimen\jot \jot=3pt

The following three formulas show this for the default value, \setlength\jot{0pt}
and \setlength\jot{10pt}.

y = d
y = d y = d
1 1 1
y = c +d y = c +d y = c +d
x x x
1 1 1
y = b 2 + cx + d y = b 2 + cx + d y = b + cx + d
x x x2

Defining a new environment with a parameter makes things easier, because

changes to the length are locally.
1 \ newenvironment { mathspace }[1]{ %
2 \ setlength {\ jot }{#1} %
3 \ ignorespaces %

Mathmode.tex 36
 12 STYLES

4 }{ %
5 \ i g n o re s p a c e s af t e r e n d %
6 }

\arraystretch
\arraystretch The vertical space between the lines for all math expressions
which contain an array environment can be changed with the command
\arraystretch, which is predefined as

\renewcommand\arraystretch{1}

Renewing this definition is global to all following math expressions, so it

should be used in the same way as \jot.

\vskip Another spacing for single lines is possible with the \vskip macro:
1 \[
2 \begin{pmatrix}
3 0 & 1 & 1 & 0 & 0 & 1 \\
0 1 1 0 0 1
 
4 1 & 0 & 0 & 1 & 1 & 0 \\
1 0 0 1 1 0 5 \noalign{\vskip2pt}
 
 1  6 0 & 1 & 1 & 0 & \dfrac{1}{\sqrt{2}} & 1\\
0 1 1 0 √ 1

 2 
 7 \noalign{\vskip2pt}
1 0 1 0 1 0 8 1 & 0 & 1 & 0 & 1 & 0 \\
0 1 0 1 0 1 9 0 & 1 & 0 & 1 & 0 & 1 \\
10 \end{pmatrix}
11 \]

Package setspace To have all formulas with another vertical spacing, one
can choose the package setspace and redefining some of the math macros,
e.g.,
1 \ newcommand *\ Array [2][1]{\ setstretch {#1}\ array {#2}}
2 \ let \ endArray \ endarray

1 \[
2 \begin{Array}[2]{cc}
a= b 3 a =&b\\
4 a =&b\\
a= b 5 a =&b
6 \end{Array}
7 \]
a= b 8
9 text $\begin{Array}{cc}
10 a =&b\\
a= b
11 a =&b\\
text a = b text 12 a =&b
a= b 13 \end{Array}$ text

Mathmode.tex 37
 12 STYLES

Mode Inline Displayed

´ ˆ
default f (t) = T 1
dt T 1
2π sin ωt f (t) = dt
2π sin ωt
ˆ ˆ
T 1 T 1
\displaystyle f (t) = dt f (t) = dt
2π sin ωt 2π sin ωt
T
´ 1
\scriptstyle f (t) = 2π sin ωt dt T
´ 1
f (t)= 2π sin ω
dt
t

\scriptscriptstyle T ´ 1 dt
T ´
f (t)= 2π
sin ω
t f (t)= 2π 1 dt
sin ω
t
´
\textstyle f (t) = T 1
dt T
´ 1
2π sin ωt f (t) = dt
2π sin ωt

Table 8: Math styles

12 Styles
This depends on the environment in which they are used. An inline formula
has a default math fontsize called \textstyle, which is smaller than the \textstyle
one for a display formula (see section 3), which is called \displaystyle. \displaystyle
Beside this predefinition there are two other special fontstyles for math, \scriptstyle
\scriptstyle and \scriptscriptstyle. They are called “style” in difference \scripscriptstyle
to “size”, because they have a dynamic character, their real fontsize belongs to
the environment in which they are used. A fraction for example is by default
in scriptstyle when it is in an inline formula like this ab , which can be changed
a
to . This may be in some cases useful but it looks in general ugly because
b
the line spacing is too big. These four styles are predefined and together in a
logical relationship. It is no problem to use the other styles like large, \Large,
. . . outside the math environment. For example a fraction written with
\Huge:
a (\Huge$\frac{a}{b}$). This may cause some problems when
b
you want to write a displayed formula in another fontsize, because it also
affects the interline spacing of the preceding part of the paragraph. If you
end the paragraph, you get problems with spacing and page breaking above
the equations. So it is better to declare the font size and then restore the
baselines: ˆ 2 1
dx = 0.5 (37)
1 x2

1 \ makeatletter
2 \ newenvironment { smallequation }[1]{ %
3 \ skip@ =\ baselineskip
4 #1 %
5 \ baselineskip =\ skip@

Mathmode.tex 38
 14 ACCENTS

6 \ equation
7 }{\ endequation \ i g n o re spacesafterend }
8 \ makeatother
9
10 \ begin { smallequation }{\ tiny }
11 \ int _1^2\ ,\ frac {1}{ x ^2}\ ,\ mathrm { d } x =0.5
12 \ end { smallequation }

If you use this the other way round for huge fontsizes, don’t forget to load
package exscale (see section 47.13 on page 105). Also see this section for
diffent symbol sizes.

13 Dots
\cdots
In addition to the above decorations there are some more different dots which \dots
are single commands and not by default over/under a letter. It is not easy \dotsb
to see the differences between some of them. Dots from lower left to upper \dotsc
. \dotsi
right are possible with \reflectbox{$\ddots$} . . \dotsm
\dotso
..
\cdots · · · \ddots . \dotsb · · · \dotsc . . . \dotsi · · · \ldots
. \vdots
\dotsm ··· \dotso ... \ldots ... \vdots ..

Table 9: Dots in math mode

14 Accents
The letter “a” is only for demonstration. The table 10 shows all in standard
LATEX available accents and also the ones placed under a character. With
package amssymb it is easy to define new accents. For more information see
section 31 on page 68 or other possibilities at section 47.1 on page 96.

\acute á \bar ā \breve ă

\bar ā \breve ă
...
\check ǎ \dddot a \ddot ä
\dot ȧ \grave à \hat â
\mathring å \overbrace
z}|{
a \overleftarrow ←
−
a
\overleftrightarrow ←
→
a \overline a \overrightarrow →
−
a
\tilde ã \underbar a \underbrace a
|{z}
\underleftarrow a
←
− \underleftrightarrow ←
→a \underline a
\underrightarrow a
→
− \vec ~a \widehat a
b
\widetilde a
e

Table 10: Accents in math mode

Mathmode.tex 39
 14.1 Over- and underbrackets 14 ACCENTS

The letters i and j can be substituted with the macros \imath and \jmath
when an accents is placed over these letters and the dot should disappear:
...
~ı  ($\vec{\imath}\ \dddot{\jmath}$).
Accents can be used in different ways, e.g., strike a single character with a
horizontal line like $\mathaccent‘-A$: A- or $\mathaccent\mathcode‘-A$:
− In section 47.7 on page 100 is a better solution for more than one character.
A.

14.1 Over- and underbrackets

There are no \underbracket and \overbracket commands in the list of
accents. They can be defined in the preamble with the following code.
1 \ makeatletter
2 \ def \ underbracket { %
3 \ @ifnextchar [{\ @underbracket }{\ @underbracket [\ @bracketheight ]} %
4 }
5 \ def \ @underbracket [#1]{ %
6 \ @ifnextchar [{\ @under@bracket [#1]}{\ @under@bracket [#1][0.4 em ]} %
7 }
8 \ def \ @under@bracket [#1][#2]#3{ % \ message { Underbracket : #1 ,#2 ,#3}
9 \ mathop {\ vtop {\ m@th \ ialign {##\ crcr $\ hfil \ displaystyle {#3}\ hfil $
%
10 \ crcr \ noalign {\ kern 3\ p@ \ nointerlineskip }\ upbracketfill {#1}{#2}
11 \ crcr \ noalign {\ kern 3\ p@ }}}}\ limits }
12 \ def \ upbracketfill #1#2{$\ m@th \ setbox \ z@ \ hbox {$\ braceld $}
13 \ edef \ @bracketheight {\ the \ ht \ z@ }\ bracketend {#1}{#2}
14 \ leaders \ vrule \ @height #1 \ @depth \ z@ \ hfill
15 \ leaders \ vrule \ @height #1 \ @depth \ z@ \ hfill \
bracketend {#1}{#2}$}
16 \ def \ bracketend #1#2{\ vrule height #2 width #1\ relax }
17 \ makeatother

1. \underbrace{...} is an often used command:

x2 + 2x + 1 = f (x) (38)
| {z }
(x + 1)2

2. Sometimes an underbracket is needed, which can be used in more ways

than \underbrace{...}. An example for \underbracket{...}:

Hate Science 1 → 2 → 3 → 4 → 5 → 6 → 7 → 8 → 9 → 10 Love Science

low medium high

Mathmode.tex 40
14.2 Vectors 14 ACCENTS

14.1.1 Use of \underbracket{...}

The \underbracket{...} command has two optional parameters:

• the line thickness in any valid latex unit, e.g., 1pt

• the height of the edge brackets, e.g., 1em

using without any parameters gives the same values for thickness and height
as predefined for the \underbrace command.

1. $\underbracket{foo~bar}$ f oo bar

2. $\underbracket[2pt]{foo~bar}$ f oo bar

3. $\underbracket[2pt][1em] {foo~bar}$ f oo bar

14.1.2 Overbracket
In addition to the underbracket an overbracket is also useful, which can be
used in more ways than \overbrace{...}. For example:

Hate Science 1 → 2 → 3 → 4 → 5 → 6 → 7 → 8 → 9 → 10 Love Science

low medium high

The \overbracket{...} command has two optional parameters:

• the line thickness in any valid latex unit, e.g., 1pt

• the height of the edge brackets, e.g., 1em

using without any parameters gives the same values for thickness and height
as predefined for the \overbrace command.

1. $\overbracket {foo\ bar}$ f oo bar

2. $\overbracket[2pt] {foo\ bar}$ f oo bar

3. $\overbracket[2pt] [1em] {foo\ bar}$ f oo bar

Mathmode.tex 41
14.2 Vectors 15 EXPONENTS AND INDICES

\vv{...} \overrightarrow{...}
#»
a →
−
a
#» −→
abc abc
#»
ı →
−
ı
#» →
−
Ax Ax

Table 11: Vectors with package esvec (in the right column the default one
from LATEX)

14.2 Vectors
Especially for vectors there is the package esvect17 package, which looks
better than the \overrightarrow, e.g.,
Look into the documentation for more details about the package esvect.

15 Exponents and indices

The two active characters _ and ^ can only be used in math mode. The
following character will be printed as an index ($y=a_1x+a_0$: y = a1 x+a0 )
or as an exponent ($x^2+y^2=r^2$: x2 + y 2 = r2 ). For more than the next
character put it inside of {}, like $a_{i-1}+a_{i+1}<a_i$: ai−1 + ai+1 < ai .
Especially for multiple exponents there are several possibilities. For
example:

4
3 4 3
((x2 )3 )4 = ((x2 ) ) = x2 (39)

1 (( x ^2) ^3) ^4 =
2 {({( x ^2) }^3) }^4 =
3 {\ left ({\ left ( x ^2\ right ) }^3\ right ) }^4

For variables with both exponent and indice index the order is not
important, $a_1^2$ is exactly the same than $a^2_1$: a21 = a21 . By default
all exponents and indices are set as italic characters. It is possible to change
this behaviour to get upright characters. The following example shows this
for the indices.
1 $A_{abc_{xyz}123def}^{abc123def}aa$
2
3 \makeatletter
Aabc123def
abcxyz 123def aa 4 \catcode‘\_\active
5 \def_#1{\sb{\operator@font#1}}
Aabc123def
abcxyz 123def aa 6 \makeatother
7
8 $A_{abc_{xyz}123def}^{abc123def}aa$

17
CTAN://macros/latex/contrib/esvect/

Mathmode.tex 42
 16 OPERATORS

16 Operators
They are written in upright font shape and are placed with some additional
space before and after for a better typesetting. With the AMSmath package it
is possible to define one’s own operators (see section 36 on page 74). Table 12
and 13 show a list of the predefined ones for standard LATEX.
` W V
\coprod U \bigvee T \bigwedge S
\biguplus ´ \bigcap ´ \bigcup Q
\intop P \int N \prod L
\sum J \bigotimes ¸ \bigoplus ¸
\bigodot F \ointop \oint
\bigsqcup \smallint ∫

Table 12: The predefined operators of fontmath.ltx

The difference between \intop and \int is that the first one has by default
over/under limits and the second subscript/superscript limits. Both can be
changed with the \limits or \nolimits command. The same behaviour
happens to the \ointop and \oint Symbols.

\log log \lg lg \ln ln

\lim lim \limsup lim sup \liminf lim inf
\sin sin \arcsin arcsin \sinh sinh
\cos cos \arccos arccos \cosh cosh
\tan tan \arctan arctan \tanh tanh
\cot cot \coth coth \sec sec
\csc csc \max max \min min
\sup sup \inf inf \arg arg
\ker ker \dim dim \hom hom
\det det \exp exp \Pr Pr
\gcd gcd \deg deg \bmod mod
\pmod{a} (mod a)

Table 13: The predefined operators of latex.ltx

For more predefined operator names see table 20 on page 98. It is easy to
define a new operator with
1 \ makeatletter
2 \ newcommand \ foo {\ mathop {\ operator@font foo }\ nolimits }
3 \ makeatother

Now you can use \foo in the usual way:

1 \[ \foo_1^2 = x^2 \]
foo21 = x2

Mathmode.tex 43
 17 GREEK LETTERS

In this example \foo is defined with \nolimits, means that limits are placed
in superscript/subscript mode and not over under. This is still possible with
\limits in the definition or the equation:
2
\[ \foo\limits_1^2 = x^2 \]
foo = x2
1

AMSmath has an own macro for a definition, have a look at section 36

on page 74.

17 Greek letters
The AMSmath package simulates a bold font for the greek letters, it writes
a greek character twice with a small kerning. The \mathbf{<character>}
doesn’t work with lower greek character. See section 40 on page 79 for the
\pmb macro, which makes it possible to print bold lower greek letters. Not
all upper case letters have own macro names. If there is no difference to
the roman font, then the default letter is used, e.g., A for the upper case
of α. Table 14 shows only those upper case letters which have own macro
names. Some of the lower case letters have an additional var option for an
alternative.
lower default upper default \mathbf \mathit
\alpha α
\beta β
\gamma γ \Gamma Γ Γ Γ
\delta δ \Delta ∆ ∆ ∆
\epsilon 
\varepsilon ε
\zeta ζ
\eta η
\theta θ \Theta Θ Θ Θ
\vartheta ϑ
\iota ι
\kappa κ
\lamda λ \Lambda Λ Λ Λ
\mu µ
\nu ν
\xi ξ \Xi Ξ Ξ Ξ
\pi π \Pi Π Π Π
\varpi $
\rho ρ
\varrho %
\sigma σ \Sigma Σ Σ Σ
\varsigma ς

Mathmode.tex 44
 19 \STACKREL

lower default upper default \mathbf \mathit

\tau τ
\upsilon υ \Upsilon Υ Υ Υ
\phi φ \Phi Φ Φ Φ
\varphi ϕ
\chi χ
\psi ψ \Psi Ψ Ψ Ψ
\omega ω \Omega Ω Ω Ω
Table 14: The greek letters

Bold greek letters are possible with the package bm (see section 47.5 on
page 99) and if they should also be upright with the package upgreek:
$\bm{\upalpha}, \bm{\upbeta} ... $ α, β...
A useful definition maybe:
1 \ usepackage { upgreek }
2 \ makeatletter
3 \ newcommand {\ bfgreek }[1]{\ bm {\ @nameuse { up #1}}}
4 \ makeatother

Then $\bfgreek{mu}$ will allow you to type µ to obtain an upright

boldface µ.

18 Pagebreaks
\allowdisplaybreaks
By default a displayed formula cannot have a pagebreak. This makes some
sense, but sometimes it gives a better typesetting when a pagebreak is
possible.

\allowdisplaybreaks

\allowdisplaybreaks enables TEX to insert pagebreaks into displayed

formulas whenever a newline command appears. With the command \displaybreak
it is also possible to insert a pagebreak at any place.

19 \stackrel
\stackrel puts a character on top of another one which may be important if
∧
a used symbol is not predefined. For example “ =” (\stackrel{\wedge}{=}).
The syntax is \stackrel
1 \ stackrel { top }{ base }

Such symbols may be often needed so that a macro definition in the

preamble makes some sense:

Mathmode.tex 45
 21 COLOR IN MATH EXPRESSIONS

1 \ newcommand {\ eqdef }{ %
2 \ ensuremath { %
3 \ stackrel {\ mathrm { def }}{=} %
4 }%
5 }

With the \ensuremath command we can use the new \eqdef command in
text and in math mode, LATEX switches automatically in math mode, which
saves some keystrokes like the following command, which is written without
def
the delimiters ($...$) for the math mode = , only \eqdef with a space at
the end. In math mode together with another material it may look like
def
~x = (x1 , . . . , xn ) and as command sequence
1 $\ vec { x }\ eqdef \ left ( x _{1} ,\ ldots , x _{ n }\ right ) $

The fontsize of the top is one size smaller than the one from the base, but
it is no problem to get both the same size, just increase the top or decrease
the base.

20 \choose
\choose is like \atop with delimiters or like \frac without the fraction line
and also with delimiters. It is often used for binoms and has the following
syntax: \choose
1 { above \ choose below }

The two braces are not really important but it is safe to use them.
     
m+1 m m
= + (40)
n n k−1
1 {{ m +1 \ choose n }}={{ m \ choose n }}+{{ m \ choose k -1}}\ label { eq :
choose }

See section 29.2 on page 67 for the AMSmath equivalents and enhance-
ments.

21 Color in math expressions

There is no difference in using colored text and colored math expressions.
With

\usepackage{color}

Mathmode.tex 46
 22 BOLDMATH

in the preamble the macro \textcolor{<color>}{<text or math>} exists.

ˆ∞
1
f (x) = dx = 1 (41)
x2
1
\textcolor
1 \ begin { equation }
2 \ textcolor { blue }{ f ( x ) } = \ int \ limits _1^{\ infty }\ textcolor { red
}{\ frac {1}{ x ^2}}\ ,\ mathrm { d } x =1
3 \ end { equation }

If all math expressions should be printed in the same color, then it is

better to use the everydisplay macro (section 24 on page 49).

22 Boldmath
\mathversion
Writing a whole formula in bold is possible with the command sequence \boldmath
\boldmath ... \unboldmath, which itself must be written in textmode (out- \unboldmath
side the formula) or with the command {\mathversion{bold} ... }.
X X
aij bjk cki aij bjk cki
1≤j≤p 1≤j≤p
1≤j≤q 1≤j≤q
1≤k≤r 1≤k≤r

1 \ boldmath
2 \[
3 \ sum _{ %
4 \ makebox [0 pt ]{$ %
5 {{\ scriptscriptstyle 1\ le j \ le p \ atop { %
6 {1\ le j \ le q \ atop 1\ le k \ le r }}}} %
7 $} %
8 } a _{ ij } b _{ jk } c _{ ki }
9 \]
10 \ unboldmath

The \mathversion macro defines a math style which is valid for all
following math expressions. If you want to have all math in bold then use
this macro instead of \boldmath. But it is no problem to put \mathversion
inside a group to hold the changes locally.

y(x) = ax3 + bx2 + cx + d (42)

1 {\ mathversion { bold } %
2 \ begin { equation }
3 y ( x ) = ax ^3+ bx ^2+ cx + d
4 \ end { equation }}

Mathmode.tex 47
22.1 Bold math titles and items 23 MULTIPLYING NUMBERS

Single characters inside a formula can be written in bold with \mathbf, but
only in upright mode, which is in general not useful as shown in equation 43.
It is better to use package bm (see section 47.5 on page 99).
X
aij bjk cki (43)
1≤j≤p
1≤j≤q
1≤k≤r

22.1 Bold math expressions as part of titles and items

By default the titles in sections, subsections, a.s.o. are printed in bold. Same
for the description environment. The problem is that a math expression in
one of these environments is printed in default font shape, like the following
example for a section and description environment:

22 Function f (x) = x2
This is y = f (x) Only a demonstration.

And z = f (x, y) Another demonstration.

With a redefinition of the \section and \item macros it is possible to
get everything in bold font.

22 Function f (x) = x2
This is y = f (x) Only a demonstration.

And z = f (x, y) Another demonstration.

1 \ let \ itemOld \ item

2 \ makeatletter
3 \ renewcommand \ item [1][]{ %
4 \ def \ @tempa {#1}
5 \ ifx \ @tempa \ @empty \ itemOld \ else \ boldmath \ itemOld [#1]\
unboldmath \ fi %
6 }
7 \ makeatother
8 \ let \ sectionOld \ section
9 \ renewcommand \ section [2][\ empty ]{ %
10 \ boldmath \ sectionOld [#1]{#2}\ unboldmath %
11 }

23 Multiplying numbers
When the dot is used as the decimal marker as in the United States, the
preferred sign for the multiplication of numbers or values of quantities is a
cross (\times × ), not a half-high and centered dot (\cdot · ).

Mathmode.tex 48
 24 OTHER MACROS

When the comma is used as the decimal marker as in Europe, the preferred
sign for the multiplication of numbers is the half-high dot. The multiplication
of quantity symbols (or numbers in parentheses or values of quantities in
parentheses) may be indicated in one of the following ways: ab, a · b, a × b.
For more information see “Nist Guide to SI Units -More on Printing and
Using Symbols and Numbers in Scientific and Technical Documents” 18 or the
German DIN 1304, Teil 1.

24 Other macros
\everymath
There are some other macros which are not mentioned in the foregoing text. \everydisplay
Here comes a not really complete list of these macros. \underline

\everymath puts the argument before any inlined math expression, e.g.,
\everymath{\small}.

\everydisplay puts the argument before any displayed math expression,

e.g., \everydisplay{\color{blue}}.

\underline underlines a math expression and has to be used inside the math
mode. ˆ
F (x) = f (x) dx

18
http://physics.nist.gov/Pubs/SP811/sec10.html

Mathmode.tex 49
Part II
AMSmath package
In general the AMS packages are at least a collection of three different ones:
1. amsmath.sty

2. amssymb.sty

3. amsfonts.sty
In the following only the first one is described in detail.
The AMSmath has the following options:

centertags (default) For a split equation, place equation numbers verti-

cally centered on the total height of the equation.
tbtags ‘Top-or-bottom tags’ For a split equation, place equation
numbers level with the last (resp. first) line, if numbers are
on the right (resp. left).
sumlimits (default) Place the subscripts and superscripts of summation
symbols above and below, in displayed equations. QThis `
option
N L also affects other symbols of the same type – , ,
, , and so forth – but excluding integrals (see below).
nosumlimits Always place the subscripts and superscripts of summation-
type symbols to the side, even in displayed equations.
intlimits Like sumlimits, but for integral symbols.
nointlimits (default) Opposite of intlimits.
namelimits (default) Like sumlimits, but for certain ‘operator names’
such as det, inf, lim, max, min, that traditionally have
subscripts placed underneath when they occur in a displayed
equation.
nonamelimits Opposite of namelimits.

To use one of these package options, put the option name in the optional
argument, e.g., \usepackage[intlimits]{amsmath}. The AMSmath also
recognises the following options which are normally selected (implicitly or
explicitly) through the documentclass command, and thus need not be
repeated in the option list of the \usepackage{amsmath} statement.

leqno Place equation numbers on the left.

reqno (default) Place equation numbers on the right.
fleqn Position equations at a fixed indent from the left margin rather
than centered in the text column. AMSmath defines the length
\mathindent and uses it when the equations have only one tabbing
character (&).

Mathmode.tex 50
 25 ALIGN ENVIRONMENTS

All math environments are displayed ones, so there is no special inline math.

25 align environments
There are four different align environments, described in the following subsec-
tions. Their behaviour is shown in table 15. The symbolic code for all align
environments is:
1 \ begin { < name >}
2 < name > &= x & x &= x \\
3 < name > &= x & x &= x
4 \ end { < name >}

Table 15: Comparison between the different align environments with the
same code, where the first three can have an equation number

align = x x = x
align = x x = x

alignat = x x = x
alignat = x x = x

flalign = x x = x
flalign = x x = x

xalignat = x x = x
xalignat = x x = x

xxalignat = x x = x
xxalignat = x x = x

In difference to the eqnarray environment from standard LATEX (sec-

tion 3.2), the “three” parts of one equation expr.-symbol-expr. are divided

Mathmode.tex 51
25.1 The default align environment 25 ALIGN ENVIRONMENTS

by only one ampersand in two parts. In general the ampersand should be be-
fore the symbol to get the right spacing, e.g., y &= x. Compare the following
three equations, the second one has a wrong spacing.
y=x 1 y &= x
y =x 2 y =& x
3 y ={}& x
y=x

25.1 The default align environment

The eqnarray environment has a not so good spacing between the cells.
Writing the equations no. 3 to 6 with the align environment gives:

y=d (44)
y = cx + d (45)
y12 = bx2 + cx + d (46)
3 2
y(x) = ax + bx + cx + d (47)

The code looks like:

1 \ begin { align }
2 y & = d \ label { eq : IntoSection }\\
3 y & = cx + d \\
4 y _{12} & = bx ^{2}+ cx + d \\
5 y ( x ) & = ax ^{3}+ bx ^{2}+ cx + d
6 \ end { align }

• The align environment has an implicit {rlrl...} horizontal alignment

with a vertical column-alignment, e.g.,
1 \begin{align*}
2 1 & 2 & 3
12 3 3 \end{align*}

• A nonumber-version \begin{align*}...\end{align*} exists.

• Unnumbered single rows are possible with \nonumber.

• The align environment takes the whole horizontal space if you have
more than two columns:

y=d z=1 (48)

y = cx + d z =x+1 (49)
2 2
y12 = bx + cx + d z =x +x+1
3 2
y(x) = ax + bx + cx + d z = x3 + x2 + x + 1 (50)

Mathmode.tex 52
 25.2 alignat environment 25 ALIGN ENVIRONMENTS

The code for this example looks like

1 \ begin { align }
2 y & = d & z & =1\\
3 y & = cx + d & z & = x +1\\
4 y _{12} & = bx ^{2}+ cx + d & z & = x ^{2}+ x +1\ nonumber \\
5 y ( x ) & = ax ^{3}+ bx ^{2}+ cx + d & z & = x ^{3}+ x ^{2}+ x +1
6 \ end { align }

25.2 alignat environment

\begin{align}
...
>From now the counting of the equation changes. It is introduced \end{align}
with a foregoing command, which doesn’t really make sense, it is only for
demonstration:
\renewcommand{\theequation}{\thepart-\arabic{equation}}.

This means “align at several places” and is something like more than two
align environment side by side. Parameter is the number of the align
environments, which is not important for the user. The above last align
example looks like:

y=d z=1 (II-51)

y = cx + d z =x+1 (II-52)
2 2
y12 = bx + cx + d z =x +x+1
y(x) = ax + bx + cx + d z = x3 + x2 + x + 1
3 2
(II-53)

The parameter was 2 and it is 3 for the following example:

i11 = 0.25 i12 = i21 i13 = i23

1
i21 = i11 i22 = 0.5i12 i23 = i31 (II-54)
3
i31 = 0.33i22 i32 = 0.15i32 i33 = i11 (II-55)

For this example the code is:

1 \ begin { alignat }{3}
2 i _{11} & =0.25 & i _{12} & = i _{21} & i _{13} & = i _{23}\ nonumber \\
3 i _{21} & =\ frac {1}{3} i _{11} & i _{22} & =0.5 i _{12}& i _{23} & = i _{31}\\
4 i _{31} & =0.33 i _{22}\ quad & i _{32} & =0.15 i _{32}\ quad & i _{33} & = i _{11}
5 \ end { alignat }

With the alignat environment one can easily align equations vertically
at more than one marker:

Mathmode.tex 53
 25.3 flalign environment 25 ALIGN ENVIRONMENTS

abc = xxx = xxxxxxxxxxxx = aaaaaaaaa (II-56)

ab = yyyyyyyyyyyyyyy = yyyy = ab (II-57)

1 \ begin { alignat }{3}

2 abc &= xxx &&= xxxxxxxxxxxx &&= aaaaaaaaa \\
3 ab &= yyyyyyyyyyyyyyy &&= yyyy &&= ab
4 \ end { alignat }

• The alignat environment has an implicit {rlrl...rlrl} horizontal align-

ment with a vertical column alignment.

• A nonumber-version \begin{alignat*}...\end{alignat*} exists.

• Unnumbered single rows are possible with \nonumber.

25.3 flalign environment

\begin{flalign}
This is the new replacement for the xalignat and xxalignat environments. ...
It is nearly the same as the xalignat environment, only more “out spaced” \end{flalign}
and “left aligned”.
1 \begin{flalign}
i11 = 0.25 2 i_{11} & =0.25\nonumber \\
1 3 i_{21} & =\frac{1}{3}i_{11}\\
i21 = i11 (II-58) i_{31} & =0.33i_{22}
3 4
5 \end{flalign}
i31 = 0.33i22 (II-59)

As seen, the equations are not really left aligned, when they have only
one ampersand. In this case flalign has the same behaviour as the align
environment.
When there are more than one tabbing characters (&), then the equations
are really left aligned. This is also an easy way to get an equation with only
one ampersand left aligned, see equation II-63 below.

i11 = 0.25 i12 = i21 i13 = i23

1
i21 = i11 i22 = 0.5i12 i23 = i31 (II-60)
3
i31 = 0.33i22 i32 = 0.15i32 i33 = i11 (II-61)

The code looks like:

1 \ begin { flalign }
2 i _{11} & =0.25 & i _{12} & = i _{21} & i _{13} & = i _{23}\ nonumber \\
3 i _{21} & =\ frac {1}{3} i _{11} & i _{22} & =0.5 i _{12}& i _{23} & = i _{31}\\
4 i _{31} & =0.33 i _{22}\ quad & i _{32} & =0.15 i _{32}\ quad & i _{33} & = i _{11}
5 \ end { flalign }

Mathmode.tex 54
 25.4 xalignat environment 25 ALIGN ENVIRONMENTS

This environment can be used to mix centered and left aligned equations
without using the document wide valid option fleqn.
ˆ
1
f (x) = dx (II-62)
x2

ˆ
1
f (x) = dx (II-63)
x2
Equation II-63 is left aligned in fact of the second tabbing character &.
1 \ begin { align }\ label { eq : centered }
2 f ( x ) & = \ int \ frac {1}{ x ^2}\ ,\ mathrm { d } x
3 \ end { align }
4
5 \ begin { flalign }\ label { eq : leftaligned }
6 f ( x ) & = \ int \ frac {1}{ x ^2}\ ,\ mathrm { d } x &
7 \ end { flalign }

Another case is placing text left aligned, whereas the formulas should be
right aligned.

12(x − 1) + 20(y − 3) + 14(z − 2) = 0

same as 6x + 10y + 7z = 0

1 \ begin { flalign *}
2 && 12( x -1) +20( y -3) +14( z -2) &= 0\\
3 \ text { same as } && 6 x +10 y +7 z &= 0
4 \ end { flalign *}

25.4 xalignat environment

\begin{xalignat}
This is an obsolete macro but still supported by the AMSmath package. ...
Same as alignat environment, only a little more “out spaced”. \end{xalignat}

i11 = 0.25 i12 = i21 i13 = i23

1
i21 = i11 i22 = 0.5i12 i23 = i31 (II-64)
3
i31 = 0.33i22 i32 = 0.15i32 i33 = i11 (II-65)

The same code looks like:

1 \ begin { xalignat }{3}
2 i _{11} & =0.25 & i _{12} & = i _{21} & i _{13} & = i _{23}\ nonumber \\
3 i _{21} & =\ frac {1}{3} i _{11} & i _{22} & =0.5 i _{12}& i _{23} & = i _{31}\\
4 i _{31} & =0.33 i _{22}\ quad & i _{32} & =0.15 i _{32}\ quad & i _{33} & = i _{11}
5 \ end { xalignat }

Mathmode.tex 55
 25.5 xxalignat environment 25 ALIGN ENVIRONMENTS

25.5 xxalignat environment

\begin{xxalignat}
Like xalignat an obsolete macro but still supported by the AMSmath ...
package. Same as align environment, only extremely “out spaced”, therefore \end{xxalignat}
no equation number!

i11 = 0.25 i12 = i21 i13 = i23

1
i21 = i11 i22 = 0.5i12 i23 = i31
3
i31 = 0.33i22 i32 = 0.15i32 i33 = i11

The same code looks like:

1 \ begin { xxalignat }{3}
2 i _{11} & =0.25 & i _{12} & = i _{21} & i _{13} & = i _{23}\ nonumber \\
3 i _{21} & =\ frac {1}{3} i _{11} & i _{22} & =0.5 i _{12}& i _{23} & = i _{31}\\
4 i _{31} & =0.33 i _{22} & i _{32} & =0.15 i _{32} & i _{33} & = i _{11}
5 \ end { xxalignat }

25.6 aligned environment

\begin{aligned}
In difference to the split environment (section 26.4 on page 61), the aligned ...
environment allows more than one horizontal alignment but has also only \end{aligned}
one equation number:

2x + 3 = 7 2x + 3 − 3 = 7 − 3
2x 4
2x = 4 = (II-66)
2 2
x=2

1 \ begin { equation }
2 \ begin { aligned }
3 2 x +3 &= 7 & 2 x +3 -3 &= 7 -3 \\
4 2x &= 4 & \ frac {2 x }2 &= \ frac 42\\
5 x &= 2
6 \ end { aligned }
7 \ end { equation }

The aligned environment is similar to the array environment, there

exists no starred version and it has only one equation number and has to be
part of another math environment, which should be equation environment.
The advantage of aligned is the much better horizontal and vertical spacing.

25.7 Problems
When using one of the align environments, there should be no \\ at the end
of the last line, otherwise you’ll get another equation number for this “empty”
line:

Mathmode.tex 56
 26 OTHER ENVIRONMENTS

1 \begin{align}
2x + 3 = 7 (II-67) 2 2x+3 &= 7\\
(II-68) 3 \end{align}

1 \begin{align}
2 2x+3 &= 7
2x + 3 = 7 (II-69) 3 \end{align}

26 Other environments
26.1 gather environment
\begin{gather}
This is like a multi line environment with no special horizontal alignment. ...
All rows are centered and can have an own equation number: \end{gather}

i11 = 0.25 (II-70)

1
i21 = i11
3
i31 = 0.33i22 (II-71)

For this example the code looks like:

1 \ begin { gather }
2 i _{11} = 0.25\\
3 i _{21} = \ frac {1}{3} i _{11}\ nonumber \\
4 i _{31} =0.33 i _{22}
5 \ end { gather }

• The gather environment has an implicit {c} horizontal alignment with

no vertical column alignment. It is just like an one column array/table.

• A nonumber-version \begin{gather*}...\end{gather*} exists. Look at

section 26.4 on page 61 for an example.

26.2 gathered environment

\begin{gathered}[c]
The gathered environment is like the aligned or alignedat environment. ...
They use only so much horizontal space as the widest line needs. In difference \end{gathered}
to the gather environment it must be itself inside the math mode.

i11 = 0.25
1
i21 = i11 (II-72)
3
i31 = 0.33i22

Mathmode.tex 57
26.2 gathered environment 26 OTHER ENVIRONMENTS

1 \ begin { align }
2 \ rule {2 cm }{1 pt }
3 \ begin { gathered }
4 \ quad i _{11}=0.25\\
5 \ quad i _{21}=\ frac {1}{3} i _{11}\\
6 \ quad i _{31}=0.33 i _{22}
7 \ end { gathered }
8 \ rule {2 cm }{1 pt }
9 \ end { align }

The optional argument can be used for setting the vertical alignment
which is by default c (centered). It can also be t for top or b for bottom.

A=a
A=a B=b
A=a B=b C=c (II-73)
B=b C=c
C=c
1 \ begin { align }
2 \ rule {1 cm }{1 pt }
3 \ begin { gathered }[ t ]
4 \ quad A = a \\
5 \ quad B = b \\
6 \ quad C = c
7 \ end { gathered }
8 %
9 \ begin { gathered }[ c ]
10 \ quad A = a \\
11 \ quad B = b \\
12 \ quad C = c
13 \ end { gathered }
14 %
15 \ begin { gathered }[ b ]
16 \ quad A = a \\
17 \ quad B = b \\
18 \ quad C = c
19 \ end { gathered }
20 \ \ rule {1 cm }{1 pt }
21 \ end { align }

When using a square bracket as first character inside the environment,

then everything is ignored by AMS until a following closing bracket, because
AMS takes this as an optional argument:
A=a
[B] B = b (II-74)
[C] C = c

Mathmode.tex 58
 26.3 multline environment 26 OTHER ENVIRONMENTS

1 \ begin { align }
2 \ begin { gathered }
3 [ A ]\ quad A = a \\
4 [ B ]\ quad B = b \\
5 [ C ]\ quad C = c
6 \ end { gathered }
7 \ end { align }

The [A] is completely ignored, which can be avoided by using the optional
argument [c] or at least an empty one directly after the \begin{gather}.
Another possibility is using the package empheq, which fixes this behaviour
by default.

[A] A = a
[B] B = b (II-75)
[C] C = c

1 \ begin { align }
2 \ begin { gathered }[]
3 [ A ]\ quad A = a \\
4 [ B ]\ quad B = b \\
5 [ C ]\ quad C = c
6 \ end { gathered }
7 \ end { align }

26.3 multline environment

\begin{multline}
This is also like a multi line19 environment with a special vertical alignment. ...
The first row is left aligned, the second and all following ones except the \end{multline}
last one are centered and the last line is right aligned. It is often used to
write extremely long formulas:
1 \ begin { multline }
2 A = \ lim _{ n \ rightarrow \ infty }\ Delta x \ left ( a ^{2}+\ left ( a ^{2}+2 a \ Delta
x
3 +\ left ( \ Delta x \ right ) ^{2}\ right ) \ right .\\
4 +\ left ( a ^{2}+2\ cdot 2 a \ Delta x +2^{2}\ left ( \ Delta x \ right ) ^{2}\ right ) \\
5 +\ left ( a ^{2}+2\ cdot 3 a \ Delta x +3^{2}\ left ( \ Delta x \ right ) ^{2}\ right ) \\
6 + \ ldots \\
7 \ left .+\ left ( a ^{2}+2\ cdot (n -1) a \ Delta x +( n -1) ^{2}\ left ( \ Delta x \ right )
^{2}\ right ) \ right ) \\
8 = \ frac {1}{3}\ left ( b ^{3} - a ^{3}\ right )
9 \ end { multline }

19
It is no typo, the name of the environment is multline, no missing i here!

Mathmode.tex 59
26.3 multline environment 26 OTHER ENVIRONMENTS

  
A = lim ∆x a2 + a2 + 2a∆x + (∆x)2
n→∞
 
+ a2 + 2 · 2a∆x + 22 (∆x)2
 
+ a2 + 2 · 3a∆x + 32 (∆x)2
+ ...
 
+ a2 + 2 · (n − 1)a∆x + (n − 1)2 (∆x)2
1 3
b − a3 (II-76)


=
3

x
x
x
x
x
x (II-77)

Figure 1: multline Alignment demo (the fourth row is shifted to the right
with \shoveright)

\multlinegap= \multlinegap=
10.0pt (II-78) 0.0pt (II-79)

Figure 2: Demonstration of \multlinegap (default is 0pt)

• A nonumber-version \begin{multline*}...\end{multline*} exists.

• By default only the last line (for right equation numbers) or the first
line (for left equation numbers) gets a number, the others can’t.
• The alignment of a single line can be changed with the command
\shoveright (figure 1)
• The first line and the last line have a small gap to the text border.20
See figure 2, where the length of \multlinegap is set to 0pt for the
right one.
20
When the first (numbers left) or last line (numbers right) has an equation number
then \multlinegap is not used for these ones, only for the line without a number.

Mathmode.tex 60
 26.4 split environment 26 OTHER ENVIRONMENTS

26.4 split environment

\begin{split}
...
>From now on the counting of the equations changes. It is introduced \end{split}
with a foregoing command, which doesn’t really make sense, it is only for
demonstration:
1 \ makeatletter
2 \ @removefromreset { equation }{ section }
3 \ makeatother

The split environment is like the multline or array environment for equa-
tions longer than the column width. Just like the array environment and
in contrast to multline, split can only be used as part of another en-
vironment. split itself has no own numbering, this is given by the other
environment. Without an ampersand all lines in the split environment are
right-aligned and can be aligned at a special point by using an ampersand. In
difference to the aligned environment (section 25.6 on page 56), the split
environment permits more than one horizontal alignment.
It is important that the split environment has another behaviour when
used inside one of the “old” LATEX environments \[...\] or \begin{equation}
... \end{equation}, in this case more than one horizontal alignment tabs
are possible.
\[
\begin{split}
x \framebox[0.35\columnwidth]{x}\\
x \framebox[0.75\columnwidth]{x}\\
\framebox[0.65\columnwidth]{x}\\
x \framebox[0.95\columnwidth]{x}
x \end{split}
\]
\[
\begin{split}
~a = x \vec{a} = {}&\framebox[0.35\columnwidth]{x}\\
x &\framebox[0.75\columnwidth]{x}\\
&\framebox[0.65\columnwidth]{x}\\
x &\framebox[0.95\columnwidth]{x}
x \end{split}
\]
The following example shows the split environment as part of the
equation environment:

Mathmode.tex 61
 26.4 split environment 26 OTHER ENVIRONMENTS

ˆ 1  ˆ 2 
   
A1 =  (f (x) − g(x)) dx +  (g(x) − h(x)) dx
  
0 1
ˆ 1  ˆ 2 
2 2
   
=  (x − 3x) dx +  (x − 5x + 6) dx
  
0 1
 3 1  3 2
x 3 2  x 5 2 
=  − x  +  − x + 6x
  (II-80)
3 2 0 3 2
    1 
 1 3   8 20 1 5 
=  −  +  − + 12 − − + 6 
3 2 3 2 3 2
   
 7   14 23  7 5
= −  +  −  = + = 2 FE
6 3 6 6 6

1 \ begin { equation }
2 \ begin { split }
3 A _{1} & = \ left | \ int _{0}^{1}( f ( x ) -g ( x ) ) \ ,\ mathrm { d } x \ right | +\ left |
4 \ int _{1}^{2}( g ( x ) -h ( x ) ) \ ,\ mathrm { d } x \ right | \\
5 & = \ left | \ int _{0}^{1}( x ^{2} -3 x ) \ ,\ mathrm { d } x \ right | +\ left |
6 \ int _{1}^{2}( x ^{2} -5 x +6) \ ,\ mathrm { d } x \ right | \\
7 & = \ left | \ frac { x ^{3}}{3} -\ frac {3}{2} x ^{2}\ right | _{0}^{1}+
8 \ left | \ frac { x ^{3}}{3} -
9 \ frac {5}{2} x ^{2}+6 x \ right | _{1}^{2}\\
10 & = \ left | \ frac {1}{3} -\ frac {3}{2}\ right | +\ left |
11 \ frac {8}{3} -\ frac {20}{2}+12 -
12 \ left ( \ frac {1}{3} -\ frac {5}{2}+6\ right ) \ right | \\
13 & = \ left | -\ frac {7}{6}\ right | +\ left | \ frac {14}{3} -\ frac {23}{6}
14 \ right | =\ frac {7}{6}+\ frac {5}{6}=2\ , \ textrm { FE }
15 \ end { split }
16 \ end { equation }

The same using the array environment with {rl}-alignment instead of

split gives same horizontal alignment, but another vertical spacing21 and
the symbols are only in scriptsize and not textsize:22

´  ´ 
 1   2
A1 =  0 (f (x) − g(x)) dx +  1 (g(x) − h(x)) dx

´  ´ 
 1   2
=  0 (x2 − 3x) dx +  1 (x2 − 5x + 6) dx

1  3 2
(II-81)
 3
=  x3 − 32 x2  +  x3 − 52 x2 + 6x
   
 1 3  08 20 1
=  3 − 2  +  3 − 2 + 12 − 13 − 52 + 6 



= − 76  +  14 23  7 5
3 − 6 = 6 + 6 = 2 FE

• There exists no starred version (\begin{split*}) of the split envi-

ronment.
21
Can be changed with \renewcommand\arraystretch{1.5}
22
See section 12 on page 38

Mathmode.tex 62
 26.5 Specials 26 OTHER ENVIRONMENTS

26.5 Specials for multline and split environments

With the multline environment the equation 28 on page 28 looks like:


1 X
∆(fij f ij ) = 2  χij (σi − σj )2 + f ij ∇j ∇i (∆f )+
2
i<j

+ ∇k fij ∇k f ij + f ij f k [2∇i Rjk − ∇k Rij ] (II-82)

which is again a bad typesetting because of the two unequal parentheses.

Each one has a size which is correct for the line but not for the whole formula.
LATEX accepts only pairs of parentheses for one line and has an “empty”
parentheses, the dot “\left.” or “\right.” to get only one of the “pair”.
There are different solutions to get the right size of the parentheses. One of
them is to use the \vphantom command, which reserves the vertical space
without any horizontal one, like a vertical rule without any thickness. The
sum symbol from the first line is the biggest one and responsible for the
height, so this one is the argument of \vphantom which has to be placed
anywhere.


1 X
∆(fij f ij ) = 2  χij (σi − σj )2 + f ij ∇j ∇i (∆f )+
2
i<j


+∇k fij ∇k f ij + f ij f k [2∇i Rjk − ∇k Rij ]  (II-83)

1 \ begin { multline }
2 \ frac {1}{2}\ Delta ( f _{ ij } f ^{ ij }) =
3 2\ left (\ sum _{ i < j }\ chi _{ ij }(\ sigma _{ i } -
4 \ sigma _{ j }) ^{2}+ f ^{ ij }\ nabla _{ j }\ nabla _{ i }(\ Delta f ) +\ right .\\
5 \ left .+\ nabla _{ k } f _{ ij }\ nabla ^{ k } f ^{ ij }+
6 f ^{ ij } f ^{ k }\ left [2\ nabla _{ i } R _{ jk } -
7 \ nabla _{ k } R _{ ij }\ right ]\ vphantom {\ sum _{ i < j }}\ right )
8 \ end { multline }

Instead of using the \vphantom command it is also possible to use fixed-

width parentheses, which is described in section 8 on page 26.

26.6 cases environment

This gives support for an often used mathematical construct. You can also
choose the more than once described way to convert some text into math,
like

Mathmode.tex 63
26.7 Matrix environments 26 OTHER ENVIRONMENTS

$x=\begin{cases}
0 & \text{if A=...}\\
1 & \text{if B=...}\\
x & \textrm{this runs with as much text as you like,
but without an automatic linebreak, it runs out
of page....}
\end{cases}$

which gives equation II-84. It is obvious what the problem is.


0 if A=...

x = 1 if B=... (II-84)

x this runs with as much text as you like, but without a linebreak, it runs out of page....


In this case it is better to use a parbox for the text part with a flushleft
command for a better view.


 0 if A=...

1 if B=...






x= this runs with as much text as (II-85)

x you like, but without an




automatic linebreak, it runs




out of page....
1 \ begin { equation }
2 x =\ begin { cases }
3 0 & \ text { if A =...}\\
4 1 & \ text { if B =...}\\
5 x & \ parbox {5 cm }{ %
6 \ flushleft %
7 this runs with as much text as you like ,
8 but without an automatic linebreak ,
9 it runs out of page ....} %
10 \ end { cases }
11 \ end { equation }

>From now on the counting of the equations changes. It is introduced

with a foregoing command, which doesn’t really make sense, it is only for
demonstration:
1 \ renewcommand \ theequation {\ arabic { equation }}

26.7 Matrix environments

All matrix environments can be nested and an element may also contain
any other math environment, so that very complex structures are possible.

Mathmode.tex 64
 28 DOTS

 
a b a b a b
\Vmatrix
c
\Bmatrix \matrix
d c d c d
     
a b  a b a b
\vmatrix 
c \bmatrix \pmatrix
d c d c d
\smallmatrix a b
c d

Table 16: Matrix environments

By default all cells have a centered alignment, which is often not the best
when having different decimal numbers or plus/minus values. Changing the
alignment to right (not for the smallmatrix) is possible with matrix
vmatrix
1 \ makeatletter
Vmatrix
2 \ def \ env@matrix {\ hskip -\ arraycolsep
3 \ let \ @ifnextchar \ new@ifnextchar bmatrix
4 \ array {*\ c@MaxMatrixCols r }} Bmatrix
5 \ makeatother pmatrix
smallmatrix
The special matrix environment smallmatrix, which decreases horizontal
and vertical space is typeset in scriptstyle. The smallmatrix environment
makes some sense in the inline mode to decrease the line height. For dots
over several columns look for \hdotsfor in the following section.

27 Vertical whitespace
See section 11.5 on page 35 for the lengths which control the vertical whites-
pace. There is no difference to AMSmath.

28 Dots
In addition to section 13 on page 39 AMSmath has two more commands for
dots: \dddot{...}23 and \ddddot{...}
...
$\dddot{y}$: y
....
$\ddddot{y}$: y
Another interesting dot command is \hdotsfor with the syntax:
1 \ hdotsfor [ < spacing factor >]{ < number of columns >}

With the spacing factor the width of the dots can be stretched or shrinked.
The number of columns allows a continuing dotted line over more columns.
Equation 86 shows the definition of a tridiagonal matrix.

23
already mentioned in section 14

Mathmode.tex 65
 29 FRACTION COMMANDS

 
a11 a12 0 ... ... ... 0

 a21 a22 a23 0 ... ... 0 


 0 a32 a33 a34 0 ... 0 

 .. .. .. .. .. .. .. 

 . . . . . . . 

A=
 ..................................................... 
 (86)
 .. .. .. .. .. .. .. 

 . . . . . . . 

 0 ... 0 an−2,n−3 an−2,n−2 an−2,n−1 0 
 
 0 ... ... 0 qn−1,n−2 an−1,n−1 an−1,n 
0 ... ... ... 0 an,n−1 ann

1 \ begin { equation }
2 \ underline { A }=\ left [\ begin { array }{ ccccccc }
3 a _{11} & a _{12} & 0 & \ ldots & \ ldots & \ ldots & 0\\
4 a _{21} & a _{22} & a _{23} & 0 & \ ldots & \ ldots & 0\\
5 0 & a _{32} & a _{33} & a _{34} & 0 & \ ldots & 0\\
6 \ vdots & \ vdots & \ vdots & \ vdots & \ vdots & \ vdots & \ vdots \\
7 \ hdotsfor {7}\ cr \ vdots & \ vdots & \ vdots & \ vdots & \ vdots & \ vdots & \ vdots
\\
8 0 & \ ldots & 0 & a _{ n -2 ,n -3} & a _{ n -2 ,n -2} & a _{ n -2 ,n -1} & 0\\
9 0 & \ ldots & \ ldots & 0 & q _{ n -1 ,n -2} & a _{ n -1 ,n -1} & a _{ n -1 , n }\\
10 0 & \ ldots & \ ldots & \ ldots & 0 & a _{ n ,n -1} & a _{ nn }
11 \ end { array }\ right ]
12 \ end { equation }

29 fraction commands
29.1 Standard
Additional to the font size problem described in subsection 2.2 on page 11
AMSmath supports some more commands for fractions. The \frac command
described in [7], does no more exist in AMSmath.

• The global fraction definition has five parameters

1 \ genfrac { < left delim >}{ < right delim >}{ < thickness >}{ < mathstyle
>}{ < nominator >}{ < denominator >}

where thickness can have any length with a valid unit like
2 +x+1
genfrac{}{}{1pt}{}{x^2+x+1}{3x-2} → x 3x−2

• \cfrac (continued fraction) which is by default set in the display

Mathmode.tex 66
29.2 Binoms 29 FRACTION COMMANDS

mathstyle and useful for fractions like

1
(87)
√ 1
2+
√ 1
3+
√ 1
4+
...
which looks with the default \frac command like
1
√ 1
(88)
2+ √
3+ √1
1
4+ ...

where the mathstyle decreases for every new level in the fraction. The
\cfrac command can be called with an optional parameter which
defines the placing of the nominator, which can be [l]eft, [r]ight or
[c]enter (the default - see equation 87):

1 1
√ 1 √ 1
2+ 2+
√ 1 √ 1
3+ 3+
√ 1 √ 1
4+ 4+
... ...
• \dfrac which takes by default the displaystyle, so that fractions in
1
inline mode have the same size than in display mode.
2
• \tfrac (vice versa to \dfrac) which takes by default the scriptstyle, so
that fractions in display mode have the same size than in inline mode.

2
3 \tfrac{2}{3}
2
\frac{2}{3}
3

29.2 Binoms
\binom
They are like fractions without a rule and its syntax is different to the \choose \dbinom
command from standard LATEX (see section 2.2 on page 11). AMSmath \tbinom
provides three different commands for binoms just like the ones for fractions.

Mathmode.tex 67
 31 ACCENTS

Command Inlinemath Displaymath

 
m

m
\binom{m}{n} n n
   
m m
\dbinom{m}{n}
n n
m m



\tbinom{m}{n} n n

Table 17: binom commands

30 Roots
The typesetting for roots is sometimes not the best. Some solutions for
better typesetting are described in section 7 on page 25 for standard LATEX. \leftroot
AMSmath has some more commands for the n-th root: \uproot
1 \ sqrt [\ leftroot { < number >}\ uproot { < number >} < root >]{ < ... >}

<number> indicates a value for the points24 of which the root can be adjusted
√
to the left and/or to the top, e.g., kn a ($\sqrt[k_n]{a}$) has a too deep
√
exponent, whereas kn a $\sqrt[\uproot{2}k_n]{a}$ looks nicer.

30.1 Roots with \smash command

p \smash
The default for a root with λki as root argument looks like λki , which may
be not the best typesetting. It is possible to reduce the lowest point of the
p with \smash √
root to the baseline with the \smash command: λki −−−−−−−→ λki
The syntax of the \smash command25 renewed by the AMSmath package
is
1 \ smash [ < position >]{ < argument >}

The optional argument for the position can be:

t keeps the bottom and annihilates the top

b keeps the top and annihilates the bottom

tb annihilates top and bottom (the default)

31 Accents
With the macro \mathaccent it is easy to define new accent types, for example
1 \ def \ dotcup {$\ mathaccent \ cdot \ cup $}

24
In PostScript units (bp – Big Points).
25
In latex.ltx \smash is defined without an optional argument.

Mathmode.tex 68
 33 EQUATION NUMBERING

∪·
Overwriting of two symbols is also possible:
xyx
y
In this case the second symbol has to be shifted to the left for a length of
5mu (mu: math unit).
1 \ def \ c ur v ea rr o wl e ft ri g ht { %
2 \ ensuremath { %
3 \ mathaccent \ curvearrowright {\ mkern -5 mu \ curvearrowleft
}%
4 }%
5 }

For other possibilities to define new accents see section 47.1 on page 96.

32 \mod command
In standard LATEX the modulo command is not an operator, though it is
often used in formulas. AMSmath provides two (three) different commands
for modulo, which are listed in tabular 18.
• They all insert some useful space before and behind the mod-operator.

a\mod{nˆ2}=b → a mod n2 = b
a\pmod{nˆ2}=b → a (mod n2 ) = b
a\pod{nˆ2}=b → a (n2 ) = b

Table 18: The modulo commands and their meaning

33 Equation numbering
See section 3.3 on page 16 for equation numbering. It is mostly the same, \numberwithin
only one command is new to AMSmath. If you want a numbering like “44”
then write either in the preamble or like this example anywhere in your doc:
1 \ numberwithin { equation }{ section }

From now on the numbering looks like equation 44 on page 52. For
the book-class you can get the same for chapters.
If you want to get rid of the parentheses then write in the preamble:
1 \ makeatletter
2 \ def \ tagform@ #1{\ maketag@@@ {\ ignorespaces #1\ unskip \
@@italiccorr }}
3 \ makeatother

Now the following four subequation numbers have no parentheses.

Mathmode.tex 69
33.1 Subequations 33 EQUATION NUMBERING

33.1 Subequations
Amsmath supports this with the environment subequation. For example:

y=d 33.89a
y = cx + d 33.89b
2
y = bx + cx + d 33.89c
3 2
y = ax + bx + cx + d 33.89d

1 \ begin { subequations }
2 \ begin { align }
3 y & = d \\
4 y & = cx + d \\
5 y & = bx ^{2}+ cx + d \\
6 y & = ax ^{3}+ bx ^{2}+ cx + d
7 \ end { align }
8 \ end { subequations }

Inside of subequations only complete other environments (\begin{...}

... \end{...}) are possible.
1 \ renewcommand {\ theequation }{ %
2 \ theparentequ ation {} -\ arabic { equation } %
3 }

y=d (33.90-1)
y = cx + d (33.90-2)
2
y = bx + cx + d (33.90-3)
y = ax3 + bx2 + cx + d (33.90-4)

A ref to a subequation is possible like the one to equation 33.90-2. The

environment chooses the same counter “equation” but saves the old value
into “parentequation”.
It is also possible to place two equations side by side with counting as
subfigures:
y = f (x) (33.91a) y = f (z) (33.91b)
In this case, the AMSmath internal subfigure counter cannot be used
and an own counter has to be defined:
1 \ newcounter { mySubCounter }
2 \ newcommand {\ twocoleqn }[2]{
3 \ setcounter { mySubCounter }{0} %
4 \ let \ OldTheEquation \ theequation %
5 \ renewcommand {\ theequation }{\ OldTheEquation \ alph {
mySubCounter }} %
6 \ noindent %

Mathmode.tex 70
 35 LIMITS

7 \ begin { minipage }{.49\ textwidth }

8 \ begin { equation }\ refstepcounter { mySubCounter }
9 #1
10 \ end { equation }
11 \ end { minipage }\ hfill %
12 \ addtocounter { equation }{ -1} %
13 \ begin { minipage }{.49\ textwidth }
14 \ begin { equation }\ refstepcounter { mySubCounter }
15 #2
16 \ end { equation }
17 \ end { minipage } %
18 \ let \ theequation \ OldTheEquation
19 }
20 [ ... ]
21 \ twocoleqn { y = f ( x ) }{ y = f ( z ) }

34 Labels and tags

For the \label command see section 3.4 on page 18, it is just the same \tag
behaviour. AMSmath allows to define own single “equation numbers” with
the \tag command.
f (x) = a (linear)
2
g(x) = dx + cx + b (quadratic)
h(x) = sin x trigonometric

1 \ begin { align }
2 f ( x ) & = a \ tag { linear }\ label { eq : linear }\\
3 g ( x ) & =\ ,\ mathrm { d } x ^{2}+ cx + b \ tag { quadratic }\ label { eq :
quadratic }\\
4 h ( x ) & =\ sin x \ tag *{ trigonometric }
5 \ end { align }

• The \tag command is also possible for unnumbered equations, LATEX

changes the behaviour when a tag is detected.
• There exists a starred version \tag{*}{...}, which supresses any
annotations like parentheses for equation numbers.
• There exist two package options for tags, ctagsplit and righttag
(look at the beginning of this part on page 50).

35 Limits
By default the sum/prod has the limits above/below and the integral at the
side. To get the same behaviour for all symbols which can have limits load
the package AMSmath in the preamble as

Mathmode.tex 71
35.1 Multiple limits 35 LIMITS

1 \ usepackage [ sumlimits , intlimits ]{ amsmath }

There exist also options for the vice versa (see page 50). See also section 41
for the additional commands \underset and \overset.

35.1 Multiple limits

For general information about limits read section 2.1 on page 11. Standard
LATEX provides the \atop command for multiple limits (section 6.1 on page 24).
AMSmath has an additional command for that, which can have several lines \substack
with the following syntax: \begin{Sb}
...
1 \ substack {...\\...\\...}
\end{Sb}
The environments described in [7] \begin{Sp}
...
1 \ begin { Sb } ... \ end { Sb } \end{Sp}
2 \ begin { Sp } ... \ end { Sp }

are obsolete and no more part of AMSmath.

The example equation 21 on page 24 with the \substack command looks
like:
X
aij bjk cki (35.1)
1≤i≤p
1≤j≤q
1≤k≤r

Insert these limits in the following way:

1 \ begin { equation }
2 \ sum _{ %
3 \ substack {1\ le i \ le p \\
4 1\ le j \ le q \\
5 1\ le k \ le r }
6 }%
7 a _{ ij } b _{ jk } c _{ ki }
8 \ end { equation }

35.2 Problems
There are still some problems with limits and the following math expression.
For example:
X
X= Xij
1≤i≤j≤n

1 \[
2 X = \ sum _{1\ le i \ le j \ le n } X _{ ij }
3 \]

Mathmode.tex 72
35.3 \sideset 35 LIMITS

does not look nice because of the long limit. Using a \makebox also does not
really solve the problem, because \makebox is in TEX horizontal mode and
knows nothing about the appropriate math font size, because limits have a
smaller font size. It is better to define a \mathclap macro, similiar to the two
macros \llap and \rlap and uses the also new defined \mathclap macro:
1 \ def \ mathllap {\ mathpalette \ mathllapinternal }
2 \ def \ mathllapinternal #1#2{ %
3 \ llap {$\ mathsurround =0 pt #1{#2}$} % $
4 }
5 \ def \ clap #1{\ hbox to 0 pt {\ hss #1\ hss }}
6 \ def \ mathclap {\ mathpalette \ mathclapinternal }
7 \ def \ mathclapinternal #1#2{ %
8 \ clap {$\ mathsurround =0 pt #1{#2}$} %
9 }
10 \ def \ mathrlap {\ mathpalette \ mathrlapinternal }
11 \ def \ mathrlapinternal #1#2{ %
12 \ rlap {$\ mathsurround =0 pt #1{#2}$} % $
13 }

Now we can write limits which have a boxwidth of 0pt and the right font
size and the following math expression appears just behind the symbol:
X
X= Xij
1≤i≤j≤n

1 \[
2 X = \ sum _{\ mathclap {1\ le i \ le j \ le n }} X _{ ij }
3 \]

Another problem occurs when having operators with stacked limits in

braces:
 
X 

 . . . 
 (35.2)
i,j
i>j

This case is not easy to handle when some other math expressions are
around the braces which should be on the same baseline. However, the
following may help in some cases to get better looking braces.
1 \begin{align}
"X # 2 foo \left[\begin{array}{@{}c@{}}
... 3 \displaystyle\sum_{\substack{i,j\\i>j}} \dots
f oo i,j
bar (35.3)
4 \end{array}\right] bar
i>j
5 \end{align}

35.3 \sideset
This is a command for a very special purpose, to combine over/under limits
with superscript/subscripts for the sum-symbol. For example: it is not \sideset

Mathmode.tex 73
 36 OPERATOR NAMES

possible to place the prime for the equation 35.4 near to the sum symbol,
because it becomes an upper limit when writing without an preceeding {}.
X
0
nEn (35.4)
n<k
n odd

The command \sideset has the syntax

1 \ sideset { < before >}{ < behind >}

It can place characters on all four corners of the sum-symbol:

T
U pperLef t XU pperRight
LowerLef t LowerRight
B

1 \[
2 \ sideset {_{ LowerLeft }^{ UpperLeft }}{_{ LowerRight }^{ UpperRight
}}\ sum _{ B }^{ T }
3 \]

Now it is possible to write the equation 35.4 in a proper way with the
command \sideset{}{’} before the sum symbol:
X0
nEn (35.5)
n<k
n odd

36 Operator names
\operatorname
By default variables are written in italic and operator names in upright mode,
like y = sin(x).26 This happens only for the known operator names, but
creating a new one is very easy with:
1 \ newcommand {\ mysin }{\ operatorname { mysin }}

Now \mysin is also written in upright mode y = mysin(x) and with some
additional space before and behind.
It is obvious, that only those names can be defined as new operator names
which are not commands in another way. Instead of using the new definition
as an operator, it is also possible to use the text mode. But it is better to
have all operators of the same type, so that changing the style will have an
effect for all operators. \operatornamewithlimi
The new defined operator names cannot have limits, only superscript/sub-
script is possible. amsopn.sty has an additional command \operatornamewithlimits,
which supports over/under limits like the one from \int or \sum. \mathop
26
See section 16 on page 43, where all the standard LATEX known operator names are
listed. Package AMSmath has some more (see documentation).

Mathmode.tex 74
 37 TEXT IN MATH MODE

It is also possible to use the macro \mathop to declare anything as operator,

like
1B

1 \[ \ sideset {_1}{}{\ mathop {\ mathrm { B }}} \]

With this definition it is possible to use \sideset for a forgoing index, which
is only possible for an operator.
For a real LATEX definition have a look at section 16 on page 43.

37 Text in math mode

If you need complex structures between formulas, look also at section 65.

37.1 \text command

\text
This is the equivalent command to \mathrm or \mbox from the standard \mbox
LATEX (section 9 on page 30) with the exception, that \mathrm always uses \textnormal
the roman font and \text the actual one and that the font size is different \mathrm
when used in super- and subscript.
For example: f (x) = x this was math .

Atext
text A text
text Atext
text A text
text
1 $\ boxed { f ( x ) = x \ quad \ text { this was math }}$
2
3 {\ sffamily \ huge
4 $ A ^{\ mbox { text }}_{\ mbox { text }}$\ quad
5 $ A ^{\ text { text }}_{\ text { text }}$\ quad
6 $ A ^{\ textnormal { text }}_{\ textnormal { text }}$\ quad
7 $ A ^{\ mathrm { text }}_{\ mathrm { text }}$
8 }

The \text macro can be used at any place and can be in some cases a
better solution as \intertext (see section 37.2).

12(x − 1) + 20(y − 3) + 14(z − 2) = 0

and 6x + 10y + 7z = 0

12(x − 1) + 20(y − 3) + 14(z − 2) = 0 (37.1)

and 6x + 10y + 7z = 0 (37.2)

Mathmode.tex 75
 37.2 \intertext command 37 TEXT IN MATH MODE

1 \ begin { flalign *}
2 && 12( x -1) + 20( y -3) + 14( z -2) & = 0 &&\\
3 \ text { and } && 6 x + 10 y + 7 z & = 0 &&
4 \ end { flalign *}
5
6 \ begin { align }
7 && 12( x -1) + 20( y -3) + 14( z -2) & = 0\\
8 \ text { and } && 6 x + 10 y + 7 z & = 0
9 \ end { align }

37.2 \intertext command

This is useful when you want to place some text between two parts of math
stuff without leaving the math mode, like the name “intertext” says. For
example we write the equation II-80 on page 62 with an additional command
after the second line.

ˆ 1  ˆ 2 
   
A1 =  (f (x) − g(x)) dx +  (g(x) − h(x)) dx
0 1
ˆ 1  ˆ 2 
2 2
   
=  (x − 3x) dx +  (x − 5x + 6) dx
  
0 1

Now the limits of the integrals are used

 3 1  3 2
x 3 2  x 5 2 
=  − x  +  − x + 6x
 
3 2 0 3 2
    1 
 1 3   8 20 1 5 
= − + −
   + 12 − − + 6 
3 2 3 2 3 2
   
 7   14 23  7 5
= −  +  −  = + = 2 FE
6 3 6 6 6

The code looks like:

1 \ begin { equation }
2 \ begin { split }
3 A _{1} & = \ left | \ int _{0}^{1}( f ( x ) -g ( x ) ) \ ,\ mathrm { d } x \ right | +\ left | \
int _{1}^{2}( g ( x ) -h ( x ) ) \ ,\ mathrm { d } x \ right | \\
4 & = \ left | \ int _{0}^{1}( x ^{2} -3 x ) \ ,\ mathrm { d } x \ right | +\ left | \
int _{1}^{2}( x ^{2} -5 x +6) \ ,\ mathrm { d } x \ right | \\
5 \ intertext { Now the limits of the integrals are used }
6 & = \ left | \ frac { x ^{3}}{3} -\ frac {3}{2} x ^{2}\ right | _{0}^{1}+\ left |
\ frac { x ^{3}}{3} -
7 \ frac {5}{2} x ^{2}+6 x \ right | _{1}^{2}\\
8 & = \ left | \ frac {1}{3} -\ frac {3}{2}\ right | +\ left | \ frac {8}{3} -\
frac {20}{2}+12 -

Mathmode.tex 76
 38 EXTENSIBLE ARROWS

9 \ left ( \ frac {1}{3} -\ frac {5}{2}+6\ right ) \ right | \\

10 & = \ left | -\ frac {7}{6}\ right | +\ left | \ frac {14}{3} -\ frac {23}{6}\
right | =\ frac {7}{6}+
11 \ frac {5}{6}=2\ , \ textrm { FE }
12 \ end { split }
13 \ end { equation }

Writing very long text is possible by using a parbox, see section 9 on

page 30 for an example with \textrm, which behaves in the same way as
\text.

38 Extensible arrows
\xrightarrow
above the arrow
To write something like −−−−−−−−−−→ ubelow you can use the following macro \xleftarrow
\xmapsto
$\xrightarrow[\text{below}]{\text{above the arrow}}$

and the same with \xleftarrow. You can define your own extensible arrow
macros if you need other than these two predefined ones. To get a doublelined
extensible arrow like $\Longleftrightarrow$ (⇐⇒) but with the same
behaviour as an extensible one, write in the preamble
1 \ newcommand \ x Lo ng L ef t RightArrow [2][]{ %
2 \ ext@arrow 0055{\ L o n gL ef tR ig ht Ar ro wf il l@ }{#1}{#2}}
3 \ def \ L o n g L e f t R i g h t A r r o wf il l@ { %
4 \ arrowfill@ \ Leftarrow \ Relbar \ Rightarrow }
5 \ newcommand \ x lo ng l ef t rightarrow [2][]{ %
6 \ ext@arrow 0055{\ l o n gl ef tr ig ht ar ro wf il l@ }{#1}{#2}}
7 \ def \ l o n g l e f t r i g h t a r r o wf il l@ { %
8 \ arrowfill@ \ leftarrow \ relbar \ rightarrow }

The three parts \Leftarrow\Relbar\Rightarrow define left|middle|right

of the arrow, where the middle part would be stretched in a way that the
arrow is at least as long as the text above and/or below it. This macro has
one optional and one standard parameter. The optional one is written below
and the standard one above this arrow. Now we can write

$\xLongLeftRightArrow[\text{below}]{\text{above the arrow}}$

$\xlongleftrightarrow[\text{below}]{\text{above the arrow}}$

above the arrow above the arrow

to get ⇐========⇒ ubelow or ←−−−−−−−−→ ubelow . The “number” 0055 after
\ext@arrow defines the position relative to the extended error and is not a
number but four parameters for additional space in the math unit mu.
1 \ def \ mapstofill@ { %
2 \ arrowfill@ {\ mapstochar \ relbar }\ relbar \ rightarrow }
3 \ newcommand *\ xmapsto [2][]{ %
4 \ ext@arrow < four digits >\ mapstofill@ {#1}{#2}}

Mathmode.tex 77
 39 FRAMES

over
$\ext@arrow 0000$ 7−−−→ uunder
over
$\ext@arrow 9000$ 7−−−→ u under
over
$\ext@arrow 0900$ 7−−−→ uunder
over
$\ext@arrow 0009$ 7−−−−→ uunder
over
$\ext@arrow 0090$ 7−−−−→ uunder
over
$\ext@arrow 0099$ 7−−−−→ uunder
over
$\ext@arrow 9999$ 7−−−−→ u under
• 1st digit: space left

• 2nd digit: space right

• 3rd digit: space left and right

• 4th digit: space relativ to the tip of the “arrow”

The two macros \xrightarrow and \xleftarrow are defined as:

1 \ newcommand {\ xrightarrow }[2][]{\ ext@arrow 0359\
rightarrowfill@ {#1}{#2}}
2 \ newcommand {\ xleftarrow }[2][]{\ ext@arrow 3095\ leftarrowfill@
{#1}{#2}}

39 Frames
\boxed
AMSmath knows the macro \boxed which can be used for inline a b + c
and displayed math expressions:

ˆ ∞
1
f (x) = dt = 1 (39.1)
1 x2

1 \ begin { align }
2 \ boxed { f ( x ) =\ int _1^{\ infty }\ frac {1}{ x ^2}\ , dt =1}
3 \ end { align }

For coloured boxes use package empheq. For an example see section 47.10
on page 103.

Mathmode.tex 78
 42 PROBLEMS WITH AMSMATH

40 Greek letters
\pmb
The AMSmath package simulates a bold font for the greek letters by writing
a greek character twice with a small kerning. This is done with the macro
\pmb{<letter>}. The \mathbf{<character>} doesn’t work with lower greek
character.
α α
β β
γ γ
δ δ
 
... ...

41 Miscellaneous commands
There are several commands which can be used in math mode: \overset
Some examples are shown in table 19. \underset
\boxed

$\underset{under}{baseline}$ baseline
under
over
$\overset{over}{baseline}$ baseline

Table 19: Different mathcommands

\underset is a useful macro for having limits under non-operators (see

page 97).

42 Problems with amsmath

AMSmath is an excellent package with some “funny features”. When using
an align environment inside a gather environment, it should be centered
just like the other lines. This is only true, when there is a number/tag or an
additional ampersand:

Mathmode.tex 79
 42 PROBLEMS WITH AMSMATH

m2 = m02 + m002
V20 V200
= + 00
v20 v2
v20
⇒ m2 v20 = V − V200 + V200
v200

m2 = m02 + m002
V20 V200
= + 00
v20 v2
v20
⇒ m2 v20 = V − V200 + V200
v200

1 \begin{gather*}
2 \begin{align*}
3 m_2 &= m_2’ + m_2’’\\
4 &= \frac{V_2’}{v_2’} + \frac{V_2’’}{v_2’’}
5 \end{align*}\\
6 \Rightarrow m_2 v_2’ = V - V_2’’ + V_2’’\frac{v_2’}{v_2’’}\\
7 \end{gather*}
8 \begin{gather*}
9 \begin{align*}
10 m_2 &= m_2’ + m_2’’\\
11 &= \frac{V_2’}{v_2’} + \frac{V_2’’}{v_2’’} & %<<<====
12 \end{align*}\\
13 \Rightarrow m_2 v_2’ = V - V_2’’ + V_2’’\frac{v_2’}{v_2’’}\\
14 \end{gather*}

This effect depends to the horizontal width, which is wrong in the first
example, in fact of a missing tag or number the right whitespace is cut, but
the left one is still there. The additional ampersand prevents AMSmath to
change the right margin.
Another kind of curiousity is the following example, which depends to
the same problem of cutting whitespace only on one side.

Mathmode.tex 80
 42 PROBLEMS WITH AMSMATH

a=b
c=d
a=b
c=d

1 \fbox{%
2 \begin{minipage}{10cm}
3 \begin{align*}
4 a&=b \\ c&=d
5 \end{align*}
6 \end{minipage}}
7
8 \fbox{%
9 \begin{minipage}{10cm}
10 \noindent\begin{align*}
11 a&=b \\ c&=d
12 \end{align*}
13 \end{minipage}}

Mathmode.tex 81
 43 LENGTH REGISTERS

Part III
TEX and math
There is in general no need to use the TEX macros, because the ones defined
with LATEX or with AMSmath are much more useful. Nevertheless there may
be situations, where someone has to use one of the TEX macros or special
TEX math length. One can not expect, that all macros work in the usual way,
a lot of them are redefined by LATEX or AMSmath. On the other hand some
of these basic macros or length definitions are used in the TEX way, so it
might be interesting to have all declared in a short way for some information.

43 Length registers
43.1 \abovedisplayshortskip
A length with glue, see section 11.5.1 for an example.

43.2 \abovedisplayskip
A length with glue, see section 11.5.1 for an example.

43.3 \belowdisplayshortskip
A length with glue, see section 11.5.1 for an example.

43.4 \belowdisplayskip
A length with glue, see section 11.5.1 for an example.

43.5 \delimiterfactor
The height of a delimiter is often not optimally calculated by TEX. In some
cases it is too short. With \delimiterfactor one can correct this height.
The delimiterheight is < calculated height > · < #1 > /1000 where #1 is the
parameter of \delimiterfactor. The default value is 901.
1 \[
2 y = \left\{%
 2 3 \begin{array}{ll}
 x + 2x if x < 0, 4 x^2+2x &\textrm{if }x<0,\\
 3

x^3 &\textrm{if }0\le x<1,\\
x if 0 ≤ x < 1, 5
y= 6 x^2+x &\textrm{if }1\le x<2,\\

 x2 + x if 1 ≤ x < 2, 7 x^3-x^2 &\textrm{if }2\le x.
 3
x − x2 if 2 ≤ x. 8 \end{array}%
9 \right.
10 \]

Mathmode.tex 82
43.6 \delimitershortfall 43 LENGTH REGISTERS

1 \[
 2 \delimiterfactor=1500
 3 y = \left\{%

4 \begin{array}{ll}
x2 + 2x



 if x < 0, 5 x^2+2x &\textrm{if }x<0,\\
 3

x if 0 ≤ x < 1, 6 x^3 &\textrm{if }0\le x<1,\\
y=

 x2 + x if 1 ≤ x < 2, 7 x^2+x &\textrm{if }1\le x<2,\\
x3 − x2 if 2 ≤ x. x^3-x^2 &\textrm{if }2\le x.

 8



 9 \end{array}%
10 \right.
11 \]

43.6 \delimitershortfall
Additionally to the forgoing \delimiterfactor one can modify the height
of the delimiter with another value. TEX makes the delimiter larger than
the values of < calculated height > · < delimiterfactor > /1000 and <
calculated height > − < delimitershortfall >. This makes it possible to
always get different heights of a sequence of delimiters.
1 $x\cdot\left(\left(x^2-y^2\right)-3\right)$\\[7
x · x2 − y 2 − 3



pt]
2
 
 3 $
x· 2 2
x −y −3 4 \delimitershortfall-1pt
5 x\cdot\left(\left(x^2-y^2\right)-3\right)$

1 $\left(\left(\left(A\right)\right)\right)$\\[7pt]
(((A))) 2

 3 $\delimitershortfall-1pt
(A) 4 \left(\left(\left(A\right)\right)\right)$

43.7 \displayindent
This is the left shift amount of a line holding displayed equation. By default
it is 0pt but gets the value of an indented paragraph when there is an
environment like the quotation one.
The following formula is typeset in the usual way without modifying
anything.
ˆ
sin x
f (x) = dx
x
Now we start a quotation environment which sets \labelwidth to new
values for a greater left margin.

• The following formula is typeset in the usual way without modifying

anything.
ˆ
sin x
f (x) = dx
x

Mathmode.tex 83
43.8 \displaywidth 43 LENGTH REGISTERS

• Now we write the same equation, but now with modifying displayindent,
it is set to the negative \labelwidth:
ˆ
sin x
f (x) = dx
x

1 \[
2 \ displayindent = -\ leftskip
3 f ( x ) = \ int \ frac {\ sin x }{ x }\ ,\ mathrm { d } x
4 \]

43.8 \displaywidth
The width of the line holding a displayed equation, which is by default
\linewidth. In the second example the formula is centered for a display
width of 0.5\linewidth.
ˆ
sin x
f (x) = dx
x
ˆ
sin x
f (x) = dx
x
1 \[ f ( x ) = \ int \ frac {\ sin x }{ x }\ ,\ mathrm { d } x \]
2 \[
3 \ displaywidth =0.5\ linewidth
4 f ( x ) = \ int \ frac {\ sin x }{ x }\ ,\ mathrm { d } x
5 \]

43.9 \mathsurround
Extra space added when switching in and out of the inline math mode (see
section 11.5).

43.10 \medmuskip
See section 11.1 for an example.

43.11 \mkern
Similiar to \kern, but adds a math kern item to the current math list. Length
must be a math unit.

43.12 \mskip
Similiar to \skip, but adds math glue to the current math list. Length must
be a math unit.

Mathmode.tex 84
43.13 \muskip 43 LENGTH REGISTERS

43.13 \muskip
Assigns a length with a math unit to one of the 256 \muskip register.

43.14 \muskipdef
Defines a symbolic name for a \muskip register.

43.15 \nonscript
Ignores immediately following glue or kern in script and scriptscript styles,
which makes a redefinition of \mathchoice superfluous.

43.16 \nulldelimiterspace
This is the width of a null or missing delimiter, e.g., \right. or for the left
one.

43.17 \predisplaysize
Is the effective width of the line preceeding a displayed equation, whether
\abovedisplayskip or abovedisplayshortskip is used for the vertical skip.

43.18 \scriptspace
The space inserted after an exponent or index, predefined as \scriptspace=0.5pt

43.19 \thickmuskip
See section 11.1.

43.20 \thinmuskip
The short version for positive skip is defined as \def\,{\mskip\thinmuskip}
and the one for a negative skip as \def\!{\mskip-\thinmuskip} (see also
section 11.1).
√ √
√ 2x – √2 x
1 $\sqrt 2 x$ -- $\sqrt 2\,x$\\
log x – log x 2 $\sqrt{\log x}$ -- $\sqrt{\,\log x}$\\
√ √ 3 $P\left({1/\sqrt n}\right)$ -- $P\left({1/ \
P (1/ n) – P (1/ n ) sqrt n}\,\right)$\\[8pt]
4 $[0,1)$ -- $[\,0,1)$\\
[0, 1) – [ 0, 1) 5 $x^2/2$ -- $x^2\!/2$\\
x2 /2 – x2/2

Mathmode.tex 85
43.21 \medmuskip 44 MATH FONT MACROS

ˆ ˆ ˆ ˆ 1 \[\int\int_D \mathrm{d}x\mathrm{d}y \quad

dxdy dx dy 2 \int\!\int_D \mathrm{d}x\,\mathrm{d}y\]
D D 3 \[\int\!\!\int_D \mathrm{d}x\,\,\mathrm{d}y
ˆˆ ˆˆ \quad
4 \int\!\!\!\int_D \mathrm{d}x\,\,\,\mathrm{
dx dy dx dy d}y\]
D D
ˆˆ ˆˆ 5 \[\int\!\!\!\!\int_D \mathrm{d}x\,\,\,\,\
mathrm{d}y \quad
dx dy dx dy 6 \int\!\!\!\!\!\int_D \mathrm{d}x\,\,\,\,\,\
D D
ˆˆ mathrm{d}y\]
7 \[\int\!\!\!\int_D \mathrm{d}x\,\mathrm{d}y
dx dy
D
\]

43.21 \medmuskip
See section 11.1.

44 Math font macros

44.1 \delcode
Each character has not only a \catcode and \mathcode but also a \delcode
which defines for a single chracter how it should look when used as a math
delimiter.

44.2 \delimiter
Every character can be declared as a delimiter, but TEX must know which
characters should be used for the default and the big size. For LATEX the
macro \DeclareMathDelimiter should be used (see section 8.2 on page 29).
In the following example \tdela is the character 0x22 (↑) from font
number 2 (csmy) and character 0x78 from font number 3 (cmex) for the big
version. \tdelb is the same vice versa (↓).
1 \def\tdela{\delimiter"4222378\relax}
↑x − y↓(x + y) = x2 − y 2 2 \def\tdelb{\delimiter"5223379\relax}
3
∞
X 1 2 4 $\tdela x-y\tdelb(x+y)=x^2-y^2$
↑ ↓ =4
2n 5
n=0 6 \[\tdela\sum_{n=0}^\infty {1\over2^n}\tdelb
2 ^2 = 4\]
∞
x
7
X 1
 =4
  8 \[\left\tdela\sum_{n=0}^\infty {1\over2^n}\
 n
2 y

n=0 right\tdelb^2 = 4\]

44.3 \displaystyle
See section 12 for an example.

Mathmode.tex 86
44.4 \fam 44 MATH FONT MACROS

44.4 \fam
When TEX switches into the math mode, it typesets everything using one
of the 16 possible families of fonts. \fam in an internal register where other
macros can check which font is the actual one. At the beginning TEX starts
with \fam=-1.

\fam=-1 123abcABCαβγ
\fam=0 123abcABCfffifl
\fam=1 abcABCαβγ
\fam=2 ∞∈3abcABC
\fam=3 abc

\fam=4 123abcABC
\fam=5 123aABC
1 $\mathrm{123abcABC\alpha\beta\gamma (\the\fam)
}$\\[5pt]
123abcABCfffifl(0) 2 $\mathbf{123abcABC\alpha\beta\gamma (\the\fam)

123abcABCfffifl(13) }$\\[5pt]
3 $\mathit{123abcABC\alpha\beta\gamma (\the\fam)
123abcABCfffifl(14) }$\\[5pt]
4 $\mathtt{123abcABC\alpha\beta\gamma (\the\fam)
123abcABC↑↓'(15) }$\\[5pt]
5 $\mathsf{123abcABC\alpha\beta\gamma (\the\fam)
123abcABCαβγ(−1)
}$\\[5pt]
abcABCαβγ() 6 $\mathnormal{123abcABC\alpha\beta\gamma (\the\
fam)}$

44.5 \mathaccent
Requires three parameter as one number, the class, the font family and the
character.
1 \def\dA{\mathaccent"7015\relax}
Ă 2 {\Large $\dA{A}$}

44.6 \mathbin
Declares a following character as a binary symbol with another spacing before
and behind such a symbol.
1 {\Large
a|b a | b 2 $a|b \quad a\mathbin| b$}

44.7 \mathchar
Declares a math character by three integer numbers as Parameters, giving
its class, font family, and font position. In the following example \mathchar

Mathmode.tex 87
44.8 \mathchardef 44 MATH FONT MACROS

defines a character of class 1 (big operators), font family 3 (math extension

font) and number 58 (big sum character).

∞ ∞
X 1 {\Large
P
a b a b 2 $a\sum\limits_{i=1}^{\infty} b \quad
3 a\mathchar"1358\limits_{i=1}^{\infty} b$}
i=1 i=1

44.8 \mathchardef
This is in principle the same as \mathchar, it only allows to make such
definitions permanent.
∞ 1 \bgroup
X √ 2 \mathchardef\sum="1358
a i+1
3 $a\sum\limits_{i=1}^{\infty}\sqrt{i+1}$\\[5pt]
i=1
4 \egroup
∞ √
P 5
a i+1 6 $a\sum\limits_{i=1}^{\infty}\sqrt{i+1}$
i=1

44.9 \mathchoice
Specifies specific subformula sizes for the 4 main styles: \displaystyle –
\textstyle – \scriptstyle – \scriptscriptstyle.
1 \Large
2 \def\myRule{{%
3 \color{red}%
∞ √ 4 \mathchoice{\rule{2pt}{20pt}}{\rule{1pt}{10pt}}%
P i+1
i2 5 {\rule{0.5pt}{5pt}}{\rule{0.25pt}{2.5pt}}%
i=1 6 \mkern2mu}}
7 $\myRule\sum\limits_{\myRule i=1}^{\myRule\infty}%
8 \myRule\frac{\myRule\sqrt{\myRule i+1}}{\myRule i^2}$

44.10 \mathclose
Assigns class 5 (closing character) to the following parameter, which can hold
a single character or a subformula.
B 1 {\large
A: C
:D
2 $A:\frac{B}{C}:D$\\[5pt]
A: B
C
:D 3 $A\mathopen:\frac{B}{C}\mathclose: D $}

44.11 \mathcode
A math font is far different from a text font. A lot of the characters has
to be defined with \mathcode, which defines the character with its class,
font family and character number, e.g., \mathcode‘\<="313C. It defines the
character “<” as a realtion symbol (class 3) from the font family 1 and the
character number 0x3C, which is 60 decimal.

Mathmode.tex 88
44.12 \mathop 44 MATH FONT MACROS

44.12 \mathop
Assigns class 1 (large operator) to the parameter, which can be a single
character or a subformula.

A∞
i=1 1 \[ A_{i=1}^{\infty} \]
∞ 2 \[ \mathop{A}_{i=1}^{\infty} \]
A
i=1

44.13 \mathopen
Vice versa to \mathclose (see section 44.10).

44.14 \mathord
Assigns class 0 (ordinary character) to the following parameter, which can be
a single character or a subformula.
1 {\large
y = f (x) 2 $y = f(x)$\\[5pt]
y=f (x) 3 $y \mathord= f(x)$}

44.15 \mathpunct
Assigns class 6 (punctuation) to the following parameter, which can be a
single character or a subformula (see section 11.4 for an example).

44.16 \mathrel
Assigns class 3 (relation) to the following parameter, which can be a single
character or a subformula.

x1 ox2 ox3 1 {\large

2 $x_1 o x_2 o x_3$\\[5pt]
x1 o x2 o x3 3 $x_1\mathrel o x_2\mathrel o x_3$}

44.17 \scriptfont
Specifies the scriptstyle font (used for super/subscript) for a family.
1 $A_1$
2 \font\tenxii=cmr12
A1 A1 3 \scriptfont0=\tenxii
4 $A_1$

44.18 \scriptscriptfont
Specifies the scriptscriptstyle font for a family.

Mathmode.tex 89
44.19 \scriptscriptstyle 45 MATH MACROS

44.19 \scriptscriptstyle
Selects scriptscript style for the following characters.

44.20 \scriptstyle
Selects script style for the following characters.

44.21 \skew
Especially for italic characters double accents are often misplaced. \skew has
three arguments

horizontal shift: A value in math units for the additional shift of the accent.

the accent: The symbol which is placed above the character.

the character: This is in general a single character, but can also include
itself an accent.

AMSmath redefines the setting of double accents. This is the reason why
there are only a few cases where someone has to use \skew when the package
amsmath is loaded, like in this document.
1 \large
ĩ Ã 2 $\tilde i$ \qquad $\tilde{A}$\\[5pt]
3 $\skew{3}{\tilde}{i}$ \qquad $\skew{7}{\tilde}{A}$
ĩ Ã

44.22 \skewchar
Is -1 or the character (reference symbol) used to fine-tune the positioning of
math accents.

44.23 \textfont
Specifies the text font for a family.

44.24 \textstyle
Selects the text style for the following characters.

45 Math macros
45.1 \above

Mathmode.tex 90
45.2 \abovewithdelims 45 MATH MACROS

a
b 1 $a\above0pt b$\\[8pt]
2
a
b 3 ${a\above1pt b}$\\[8pt]
4
a
5 ${a\above2.5pt b}$\\[8pt]
b
6
a 7 $\displaystyle{a\above0pt b}$
b

45.2 \abovewithdelims

1 $a\abovewithdelims()0pt b$\\[8pt]
a


2
b
 a
3 \def\fdelimA{\abovewithdelims\{)1.0pt}
4 ${a\fdelimA b}$\\[8pt]
b
5
a
6 \def\fdelimB{\abovewithdelims[]2.0pt}
b 7 ${a\fdelimB b}$\\[8pt]

a 8
9 \def\fdelimC{\abovewithdelims\{.0pt}
b 10 $\displaystyle{a\fdelimC b}$

45.3 \atop

a
b 1 $a\atop b$\\[8pt]
2
( nk ) = n!
k!(n−k)! 3 $({n \atop k}) = {n!\above1pt k!(n-k)!}$\\[8pt]
4
a 5 $\displaystyle{a\atop b}$
b

45.4 \atopwithdelims

a


1 $a\atopwithdelims() b$\\[8pt]
b
2
n n!


k = k!(n−k)! 3 ${n \atopwithdelims() k} = {n!\above1pt k!(n-k)!}$\\[8pt
]
 4
a
5 $\displaystyle{a\atopwithdelims\{. b}$
b

45.5 \displaylimits
Resets the conventions for using limits with operators to the standard for the
used environment.

Mathmode.tex 91
45.6 \eqno 45 MATH MACROS

45.6 \eqno
Puts an equation number at the right margin, the parameter can hold anything.
\eqno places only the parameter, but doesn’t increase any equation counter.
1 \[ y=f(x) \eqno{(A12)} \]
y = f (x) (A12)

45.7 \everydisplay
Inserts the parameter at the start of every switch to display math mode.
ˆ
sin x 1 \everydisplay{\color{red}
f (x) = dx
x 2 }
3 \[ f(x) = \int \frac{\sin x}{x}\,\mathrm{
ˆ 2 d}x \]
sin x 4 \[ g(x) = \int \frac{\sin^2 x}{x^2}\,\
g(x) = dx
x2 mathrm{d}x \]

45.8 \everymath
Same as \everydisplay, but now for the inline mode. In the following
example the displaystyle is used (besides using color red) for every inline
math expression.
ˆ 1 \everymath{\color{red}%
sin x
f (x) = dx 2 \displaystyle}
x 3 \[ f(x) = \int \frac{\sin x}{x}\,\
mathrm{d}x \]
sin x cos x 4 Instead of $\frac{\sin x}{x}$
Instead of now with :
xˆ x 5 now with $\frac{\cos x}{x}$:
cos x 6 \[ g(x) = \int \frac{\cos x}{x}\,\
g(x) = dx mathrm{d}x \]
x

Pay attention for side effects on footnotes and other macros which use
the math mode for superscript and other math related modes. In this case
you’ll get the footnotes also in red.

45.9 \left
TEX calculates the size of the following delimiter needed at the left side of a
formula. Requires an additional right.

45.10 \leqno
Vice versa to \eqno (see section 45.6).

45.11 \limits
Typesets limits above and/or below operators (see section 6 on page 24).

Mathmode.tex 92
45.12 \mathinner 45 MATH MACROS

45.12 \mathinner
Defines the following parameter as subformula.

45.13 \nolimits
The opposit of \limits, instead of above/below limits are placed to the right
of large operators (class 1).

45.14 \over
Is equivalent to the fraction macro of LATEX and equivalent to the \overwithdelims,
see section 45.16.
m
a n
b a+b 1 $ {a\over b} \qquad {{m\over n}\over{a+b}} $
m
n
2 \[ {m\over n}\over{a+b} \]
a+b

45.15 \overline
Puts a line over the following character or subformula and has the same
problems with different heights as underlines (see section 45.19).
1 $\overline{x}+\overline{y}=\overline{z}$\\
x+y =z 2 \let\ol\overline
x+A=z 3 $ \ol{x} + \ol{A} = \ol{z} $\\[5pt]
4 \def\yPh{\vphantom{A}}
x+A=z 5 $ \ol{x\yPh} + \ol{A} = \ol{z\yPh} $

45.16 \overwithdelims
Is a generalized fraction command with preset fraction bar thickness.
m
a

 
n 1 $ {a\overwithdelims() b} \qquad {{m\over n}\
b a+b
 m overwithdelims[]{a+b}} $
n 2 \[ {m\over n}\overwithdelims\{.{a+b} \]
a+b

45.17 \radical
Makes a radical atom from the delimiter (27-bit number) and the math field.

Mathmode.tex 93
45.18 \right 46 MATH PENALTIES

q 1 \def\mySqrt{\radical"0270371\relax}
1
7 2 $ \mySqrt{\frac{1}{7}} $\\[5pt]
r 3
1 4 \def\mySqrt{\radical"0270372\relax}
7 5 $ \mySqrt{\frac{1}{7}} $\\[5pt]
s 6

1 7 \def\mySqrt{\radical"0270373\relax}
7 8 $ \mySqrt{\frac{1}{7}} $\\[5pt]
9
v
t 10 \def\mySqrt{\radical"0270374\relax}
1
7 11 $ \mySqrt{\frac{1}{7}} $\\[5pt]

45.18 \right
Opposite to \left, makes TEX calculate the size of the delimiter needed at
the right of a formula.

45.19 \underline
When there is a combination of variables with and without an index, the
underlines are typeset with a different depth. Using \vphantom in this case
is a good choice.
1 $\underline{x}+\underline{y}=\underline{z}$\\
2
x+y =z 3 \let\ul\underline
x+y =z 4 \def\yPh{\vphantom{y}}
5 $ \ul{x\yPh} + \ul{y} = \ul{z\yPh} $\\
x1 + y2 = z3
6
7 $ \ul{x_1} + \ul{y_2} = \ul{z_3} $

45.20 \vcenter
Centers vertical material with respect to the axis.

46 Math penalties
46.1 \binoppenalty
A penalty for breaking math expressions between lines in a paragraph. TeX
breaks lines only when the binary symbol is not the last one and when the
penalty is below 10,000.

46.2 \displaywidowpenalty
The penalty which is added after the penultimate line immediately preceeding
a display math formula.

Mathmode.tex 94
46.3 \postdisplaypenalty 46 MATH PENALTIES

46.3 \postdisplaypenalty
Is added immediately after a math display ends.

46.4 \predisplaypenalty
Is added immediately before a math display starts.

46.5 \relpenalty
The penalty for a line break after a relation symbol (if a break is possible).

Mathmode.tex 95
 47 LIST OF AVAILABLE MATH PACKAGES

Part IV
Other packages
The following sections are not a replacement for the package documentation!

47 List of available math packages

accents alphalph amsart amsbook

amsbsy amscd amscls amsfonts
amslatex amsltx11 amsmath amsppt
amsppt1 amsproc amssym (plain TeX) amssymb (LaTeX)
amstex (Plain TeX) amstext amsthm bez123
bitfield brclc breqn cancel
cases comma datenumber diagxy
doublestroke easyeqn easybmat easymat
eqnarray esvect fixmath ftlpoint
icomma leftidx mathdots mathtools
mathematica mil3 mtbe Nath
numprint random romannum TeXaide

The following examples depend on the listed versions of the packages:

amsopn.sty 1999/12/14 v2.01 operator names

bm.sty 1999/07/05 v1.0g Bold Symbol Support (DPC/FMi)
empheq.sty 2004/08/03 v2.11 Emphasizing equations (MH)
amscd.sty 1999/11/29 v2.0
accents.sty 2000/08/06 v1.2 Math Accent Tools
framed.sty 2002/12/29 v 0.5: framed or shaded text with page breaks
pstcol.sty 2001/06/20 v1.1 PSTricks color colompatibility (DPC)
pstricks.sty 2004/05/06 v0.2k LaTeX wrapper for ‘PSTricks’ (RN,HV)
pstricks.tex 2003/03/07 v97 patch 15 ‘PSTricks’ (tvz)
pst-node.sty 1997/03/25 package wrapper for PSTricks pst-node.tex
delarray.sty 1994/03/14 v1.01 array delimiter package (DPC)
xypic.sty 1999/02/16 Xy-pic version 3.7
exscale.eps Graphic file (type veps)

47.1 accents
If you want to write for example an underlined M, then you can do it by
\underline{$M$} M
\underbar{$M$} M
\underaccent{\bar}{M} M
¯

Mathmode.tex 96
47.2 amscd – commutative47diagrams
LIST OF AVAILABLE MATH PACKAGES

As seen, there is no difference between \underline and \underbar.

For some reasons it may be better to use the accent package with the
\underaccents macro.

47.2 amscd – commutative diagrams

The amscd package is part of the AMSmath bundle or available at CTAN27
and has no options for the \usepackage command. amscd does not support
diagonal arrows but is much easier to handle than the complex pstricks
package or the xypic package. On the other hand simple diagrams can be
written with the array environment or look at [23].
restriction
R × S × T −−−−−−→ S × T
 

proj y
proj
y
R×S ←−−−−− S
inclusion

1 \[
2 \ begin { CD }
3 R \ times S \ times T @ >\ text { restriction } > > S \ times T \\
4 @VprojVV @VVprojV \\
5 R \ times S @ < <\ text { inclusion } < S
6 \ end { CD }
7 \]

47.3 amsopn
With the amsopn package it is very easy to declare new math operators, which
are written in upright mode:
Res versus Res
s=p s=p

1 \ documentclass [10 pt ]{ article }

2 \ usepackage { amsmath }
3 \ usepackage { amsopn }
4 \ D ec l ar eM a th O pe ra t or {\ Res }{ Res }
5 \ begin { document }
6 $\ underset { s = p }{ Res }\ quad \ underset { s = p }{\ Res }$
7 \ end { document }

Table 20 shows the predefined operatornames of amsopn.

47.4 bigdel
This is a very useful package together with the multirow package. In the
following example we need additional parentheses for a different number of
27
CTAN://macros/latex/required/amslatex/math/amscd.dtx

Mathmode.tex 97
 47.4 bigdel 47 LIST OF AVAILABLE MATH PACKAGES

\arccos arccos \arcsin arcsin \arctan arctan

\arg arg \cos cos \cosh cosh
\cot cot \coth coth \csc csc
\deg deg \det det \dim dim
\exp exp \gcd gcd \hom hom
\inf inf \injlim inj lim \ker ker
\lg lg \lim lim \liminf lim inf
\limsup lim sup \ln ln \log log
\max max \min min \Pr Pr
\projlim proj lim \sec sec \sin sin
\sinh sinh \sup sup \tan tan
\tanh tanh

Table 20: The predefined operators of amsopn.sty

rows. This is also possible with the array environment, but not as easy as
with the bigdelim package. The trick is that you need one separate column
for a big delimiter, but with empty cells in all rows, which the delimiter
spans.

  
x11 x12 ... x1p 


x21 x22 ... x2p
  

  .. some text 

 
 text . 

 


 
 xn1 1 xn1 2 ... xn1 p 



 xn1 +1,1 xn1 +1,2 . . . xn1 +1,p  

 .. some more text 

 .  


 xn1 +n2 ,1 xn1 +n2 ,2 . . . xn1 +n2 ,p 

..
.

1 \[
2 \ begin { pmatrix }
3 & x _{11} & x _{12} & \ dots & x _{1 p } & \ rdelim \}{4}{3 cm }[ some text ]\\
4 \ ldelim [{5}{1 cm }[ text ] & x _{21} & x _{22} & \ dots & x _{2 p } \\
5 & \ vdots \\
6 & x _{ n _1 1}& x _{ n _1 2} & \ dots & x _{ n _1 p }\\
7 & x _{ n _1+1 ,1}& x _{ n _1+1 ,2} & \ dots & x _{ n _1+1 , p } &
8 \ rdelim \}{3}{3 cm }[ some more text ]\\
9 & \ vdots \\
10 & x _{ n _1+ n _2 , 1} & x _{ n _1+ n _2 ,2} & \ dots & x _{ n _1+ n _2 , p }\\
11 & \ vdots \\
12 \ end { pmatrix }
13 \]

As seen in the above listing the left big delimiter is placed in the first
column, all other rows start with second column. It is possible to use all

Mathmode.tex 98
47.5 bm 47 LIST OF AVAILABLE MATH PACKAGES

columns above and below the delimiter. For the array environment there
must be two more columns defined, in case of a big delimiter left and right.
The syntax of \ldelim and \rdelim is:

\ldelim<delimiter>{<n rows>}{<added horizontal space>}[<text>]

\rdelim<delimiter>{<n rows>}{<added horizontal space>}[<text>]

Any delimiter which is possible for the \left or \right command is

allowed, e.g., “()[]{}|”. The text is an optional argument and always
typeset in text mode.

47.5 bm
By default the math macro \mathbf writes everything in bold and in upright
mode y = f (x) ($\mathbf{y=f(x)}$), but it should be in italic mode espe-
cially for variables y = f (x) ($\bm{y=f(x)}$), which is possible with the
package bm. For writing a whole formula in bold have a look at section 22 on
page 47.

47.6 braket
It is available at CTAN://macros/latex/contrib/other/misc/braket.sty and
provides several styles for writing math expressions inside brakets. For
example:
 
5
x ∈ R|0 < |x| <
3
1 \[ \ left \{ x \ in \ mathbf { R } | 0 <{| x |} <\ frac {5}{3}\ right \} \]

looks not quit right and it is not really easy to get the first vertical line in
the same size as the outer braces. Some solution may be using \vphantom:
  
 5
x ∈ R 0 < |x| <

3
1 \[
2 \ left \{\ vphantom {\ frac {5}{3}} x \ in \ mathbf { R } \ right |\ left .
0 <{| x |} <\ frac {5}{3}\ right \}
3 \]

The package braket has the macros

1 \ Bra { < math expression >}
2 \ Ket { < math expression >}
3 \ Braket { < math expression >}
4 \ Set { < math expression >}

Mathmode.tex 99
47.7 cancel 47 LIST OF AVAILABLE MATH PACKAGES

and the same with a leading lower letter, which are not really interesting.
 
5 
x ∈ R|0 < |x| < 
3
 
x ∈ R|0 < |x| < 5

 3
    
   5
x ∈ R  0 <  x  <
3
  
 5
x ∈ R  0 < |x| <

3
  
 5
x ∈ R  0 < |x| <
3
1 \[ \ Bra { x \ in \ mathbf { R } | 0 <| x | <\ frac {5}{3}} \]
2 \[ \ Ket { x \ in \ mathbf { R } | 0 <| x | <\ frac {5}{3}} \]
3 \[ \ Braket { x \ in \ mathbf { R } | 0 <| x | <\ frac {5}{3}} \]
4 \[ \ Braket { x \ in \ mathbf { R } | 0 <\ vert x \ vert <\ frac {5}{3}} \]
5 \[ \ Set { x \ in \ mathbf { R } | 0 <| x | <\ frac {5}{3}} \]

The difference between the \Set and the \Braket macro is the handling
of the vertical lines. In \Set only the first one gets the same size as the
braces and in \Braket all.
  2  
 ∂ 
φ  2  ψ
∂t
  2 
 ∂
φ  2 |ψ

∂t
1 \[\ Braket {\ phi | \ frac {\ partial ^2}{\ partial t ^2} | \ psi }\]
2 \[\ Set {\ phi | \ frac {\ partial ^2}{\ partial t ^2} | \ psi }\]

\Bra and \Ket do nothing with the inner vertical lines.

47.7 cancel
This is a nice package for canceling anything in mathmode with a slash,
backslash or a X. To get a horizontal line we can define an additional macro
called \hcancel with an optional argument for the line color (requires package
color):
1 \ newcommand \ hcancel [2][ black ]{\ setbox 0=\ hbox {#2} %
2 \ rlap {\ raisebox {.45\ ht 0}{\ textcolor {#1}{\ rule {\ wd 0}{1 pt
}}}}#2}

Mathmode.tex 100
47.8 cool 47 LIST OF AVAILABLE MATH PACKAGES

It is no problem to redefine the \cancel macros to get also colored lines.

A horizontal line for single characters is also decribed in section 14 on page 39.
x2 + 1 


(x−1)

\cancel: f (x) =
(x
 −1)(x
 + 1)

\bcancel: 3A 1234567
hhhh

\xcancel: 3
A 1234567

hh
( h(
(( h

\hcancel: 3 1234567

1 $ f ( x ) =\ dfrac {\ left ( x ^2+1\ right ) \ cancel {( x -1) }}{\ cancel {( x -1)

}( x +1) }$\\[0.5 cm ]
2 $\ bcancel {3}\ qquad \ bcancel {1234567}$\\[0.5 cm ]
3 $\ xcancel {3}\ qquad \ xcancel {1234567}$\\[0.5 cm ]
4 $\ hcancel {3}\ qquad \ hcancel [ red ]{1234567}$

47.8 cool
The cool package defines a lot of special mathematical expressions to use
them by the macro name. The following list shows only some of them, for
more informations look at the example file, which comes with the package.

\Sin{x} sin(x)
\Cos{x} cos(x)
\Tan{x} tan(x)
\Csc{x} csc(x)
\Sec{x} sec(x)
\Cot{x} cot(x)

\Style{ArcTrig=inverse} (default)

\ArcSin{x} sin−1 (x)

\ArcCos{x} cos−1 (x)
\ArcTan{x} tan−1 (x)

\Style{ArcTrig=arc}

\ArcSin{x} arcsin(x)
\ArcCos{x} arccos(x)
\ArcTan{x} arctan(x)
\ArcCsc{x} csc−1 (x)
\ArcSec{x} sec−1 (x)
\ArcCot{x} cot−1 (x)

Mathmode.tex 101
47.8 cool 47 LIST OF AVAILABLE MATH PACKAGES

\Factorial{n} n!
\DblFactorial{n} n!!
 
n
\Binomial{n}{k}
k
\Multinomial{1,2,3,4} (10; 1, 2, 3, 4)

\GammaFunc{x} Γ(x)
\IncGamma{a}{x} Γ(a, x)
\GenIncGamma{a}{x}{y} Γ(a, x, y)
\RegIncGamma{a}{x} Q(a, x)
\RegIncGammaInv{a}{x} Q−1 (a, x)
\GenRegIncGamma{a}{x}{y} Q(a, x, y)
\GenRegIncGammaInv{a}{x}{y} Q−1 (a, x, y)
\Pochhammer{a}{n} (a)n
\LogGamma{x} logΓ(x)

\Hypergeometric{0}{0}{}{}{x} 0 F0 (; ; x)
\Hypergeometric{0}{1}{}{b}{x} 0 F1 (; b; x)

\RegHypergeometric{0}{0}{}{}{x} 0 F̃ 0 (; ; x)
\RegHypergeometric{0}{1}{}{b}{x} 0 F̃ 1 (; b; x)

\MeijerG[a,b]{n}{p}{m}{q}{x}
  
m,n
 a1 , . . . , an , an+1 , . . . , ap
Gp,q x  
b1 , . . . , bm , bm+1 , . . . , bq
\MeijerG{1,2,3,4}{5,6}{3,6,9}{12,15,18,21,24}{x}
  
3,4  1, 2, 3, 4, 5, 6
G6,8 x 

3, 6, 9, 12, 15, 18, 21, 24

\RiemannZeta{s} ζ(s)
\Zeta{s} ζ(s)
\HurwitzZeta{s}{a} ζ(s, a)
\Zeta{s,a} ζ(s, a)
\RiemannSiegelTheta{x} ϑ(x)
\RiemannSiegelZ{x} Z(x)
\StieltjesGamma{n} γn

\MathieuC{a}{q}{z} Ce(a, q, z)
\MathieuS{a}{q}{z} Se(a, q, z)

\MathieuCharacteristicA{r}{q} ar (q)
\MathieuCharisticA{r}{q} ar (q)
\MathieuCharacteristicB{r}{q} br (q)
\MathieuCharisticB{r}{q} br (q)
\MathieuCharacteristicExponent{a}{q} r(a, q)
\MathieuCharisticExp{a}{q} r(a, q)

Mathmode.tex 102
47.9 delarray 47 LIST OF AVAILABLE MATH PACKAGES

47.9 delarray
Package delarray28 supports different delimiters which are defined together
with the beginning of an array:
1 \ begin { array } < delLeft >{ cc } < delRight >
2 ...

defines an array with two centered columns and the delimiters

“<delLeft><delRight>”, e.g., “()”.
1 \[
2 A =\ begin { array }({ cc })
 
3 a & b \\ a b
4 c & d A=
5 \ end { array }
c d
6 \]

The delarray package expects a pair of delimiters. If you need only one
(like the cases structure) then use the dot for an “empty” delimiter, e.g.,
1 \[
2 A =\ begin { array }\{{ cc }.

3 a & b \\ a b
4 c & d A=
5 \ end { array }
c d
6 \]

which is a useful command for a cases structure without the AMSmath

package, which is described in the AMSmath part.

47.10 empheq
This package supports different frames for math environments of the AMSmath
package. It doesn’t support all the environments from standard LATEX which
are not modified by AMSmath, e.g., eqnarray environment.
With the optional argument of the empheq environment the preferred box
type can be specified. A simple one is \fbox
ˆ ∞
1
f (x) = dt = 1 (47.1)
1 x2

1 \ begin { empheq }[ box =\ fbox ]{ align }

2 f ( x ) =\ int _1^{\ infty }\ frac {1}{ x ^2}\ , dt =1
3 \ end { empheq }

28
CTAN://macros/latex/required/tools/delarray.dtx

Mathmode.tex 103
47.11 esint 47 LIST OF AVAILABLE MATH PACKAGES

The same is possible with the macro \colorbox:

ˆ ∞
1
f (x) = dt = 1 (47.2)
1 x2

1 \ begin { empheq }[ box ={\ fboxsep =10 pt \ colorbox { yellow }}]{ align }
2 f ( x ) =\ int _1^{\ infty }\ frac {1}{ x ^2}\ , dt =1
3 \ end { empheq }

The key box can hold any possible LATEX command sequence. Boxing
subequations is also no problem, the empheq environment works in the same
way:

ˆ ∞
1
f (x) = dt = 1 (47.3a)
x1
ˆ 1
∞
1
f (x) = dt = 0.25 (47.3b)
2 x2

1 \ begin { subequations }
2 \ begin { empheq }[ box ={\ fboxsep =10 pt \ colorbox { cyan }}]{ align }
3 f ( x ) & =\ int _1^{\ infty }\ frac {1}{ x ^2}\ , dt =1\\
4 f ( x ) & =\ int _2^{\ infty }\ frac {1}{ x ^2}\ , dt =0.25
5 \ end { empheq }
6 \ end { subequations }

For more information on empheq package have a look at the documentation

of the package which is available at any CTAN server.

47.11 esint
This is a very useful package when you want nice double or triple integral or
curve integral symbols. The ones from the wasysym package29 are not the
best. esint30 supports the following symbols:

ˆ
\int : (47.4)
¨
\iint : (47.5)
˚
\iiintop : (47.6)
˘
\iiiintop : (47.7)

29
CTAN://macros/latex/contrib/wasysym/
30
CTAN://macros/latex/contrib/esint/ CTAN://fonts/ps-type1/esint/

Mathmode.tex 104
47.12 eucal and euscript
47 LIST OF AVAILABLE MATH PACKAGES

˙
\dotsintop : (47.8)
˛
\ointop : (47.9)
‹
\oiint : (47.10)
“
\sqint : (47.11)
„
\sqiint : (47.12)
‰
\ointctrclockwise : (47.13)

\ointclockwise : (47.14)
fi
\varointclockwise : (47.15)
ffi
\varointctrclockwise : (47.16)

\fint : (47.17)
"
\varoiint : (47.18)
$
\landupint : (47.19)
&
\landdownint : (47.20)

47.12 eucal and euscript

These packages should be part of your local TEX installation, because they
come with the AMSmath packages. Otherwise get them from CTAN31 . They
support a scriptwriting of only uppercase letters:

\mathscr{...} ABCDEFGHIJKLMNOPQRSTUVWXYZ

Read the documentation for the interdependence to the \mathcal com-

mand. For the above example the package eucal was loaded with the option
mathscr.

47.13 exscale
The following formula is written with the default fontsize where everything
looks more or less well:
31
CTAN://fonts/amsfonts/latex/euscript.sty

Mathmode.tex 105
47.14 mathtools 47 LIST OF AVAILABLE MATH PACKAGES

ˆ +1 n   
f (x) πX 2i − 1
√ dx ≈ f cos
−1 1 − x2 n 2n
i=1
Writing the same with the fontsize \huge gives a surprising result, which
belongs to the historical development of LATEX, the \int and \sum symbols
are not stretched. This extreme fontsize is often needed for slides and not
only written “just for fun”.

+1 √f (x) dx ≈ π X  2i − 1 
  
n
cos
Z
f

  
−1 2
  

1−x n i=1 2n
  

Using the exscale package32 package, which should be part of any local
TEX installation, all symbols get the right size.
ˆ +1 n   
f (x) π X 2i − 1
√ dx ≈ f cos
−1 1 − x2 n i=1 2n
47.14 mathtools
This package comes with a lot of additional features for typesetting math
code. Sometimes it is useful when only such equations are numbered which
are referenced in the text. This is possible with the switch \showonlyrefs.
Matrices are set by default with a centered horizontal alignment, which is
often not the best way. The mathtools package provides a starred version of
the matrix environments which allow an optional argument for the horizontal
alignment:
 
1 −1 0
 −1 1 −1
 
 1 −1 0
−11 11 −11
1 \[
2 \ begin { pmatrix *}[ r ]
3 1 & -1 & 0 \\
4 -1 & 1 & -1 \\
5 1 & -1 & 0 \\
6 -11 & 11 & -11 \\
32
CTAN://macros/latex/base/

Mathmode.tex 106
47.15 nicefrac 47 LIST OF AVAILABLE MATH PACKAGES

7 \ end { pmatrix *}
8 \]

mathtools also provides some more environments for setting equations.

Very interesting is the lgathered environment, which allows to typeset a
formula in the following way:
1 \begin{align}
2 x &=
3 \begin{lgathered}[t]
4 a + b + c \\
5 d + e +
x= a+b+c (47.21) 6 \!\begin{gathered}[t
d+e+f +g+h ]
7 f + g + h \\
i+j+k 8 i + j + k
9 \end{gathered}
10 \end{lgathered}
11 \end{align}

The \! revokes the internal horizontal space in front of the gathered

environment.

47.15 nicefrac
Typesetting fractions in the inline mode is often a bad choice, the vertical
spacing increases in fact of the fraction. The nicefrac package defines the
macro \nicefrac, which is used in the same way as the \frac command, but
it typesets the fraction with a less height: 2/3 \nicefrac{2}{3}. The package is
part of the units package bundle and can be found in the directory of units.

47.16 relsize
Often consecutives math operators are used, like two sum symbols, e.g.,
n
XX
i2
i=1

As seen the sums are of the same size. To increase the first operator size,
someone can use the \scalebox macro from package graphicx environment
and write an own macro \Sum, e.g.,
1 \ def \ Sum {\ ensuremath \ mathop {\ scalebox {1.2}{$\ displaystyle \ sum
$}}}
2 \[ \ Sum _{ j =1}\ sum _{ i =1}^\ infty i \]

∞
XX
i
j=1 i=1

Mathmode.tex 107
47.17 xypic 47 LIST OF AVAILABLE MATH PACKAGES

Another solution is to use the relsize package33 together with the

exscale one. relsize defines a useful macro \mathlarger:
n
XX 1 \[ \mathlarger{\sum}\sum_{i=1}^n
i2 i^2 \]
i=1

47.17 xypic
The \xymatrix macro is part of the xypic package34 which can be loaded
with several options which are not so important here.35 .

AA B C (47.22)
O AA
O AA
O AA
D E /o o/ o/ F
> ~ ~> ~>
>~
G H I
This matrix was created with
1 \[
2 \ xymatrix { A \ POS [];[ d ]**\ dir {~} ,[];[ dr ]**\ dir { -} & B & C \\
3 D & E \ POS [];[ l ]**\ dir {.} ,[];[ r ]**\ dir {~} & F \ POS [];[ dl
]**\ dir {~}\\
4 G & H & I}
5 \]

33
CTAN://macros/latex/ltxmisc/
34
CTAN://macros/generic/diagrams/xypic/xy-3.7/
35
For more information look at the package documentation or the package xy itself,
which is often saved in /usr/share/texmf/tex/generic

Mathmode.tex 108
 49 LATIN MODERN

Part V
Math fonts
Typesetting text and math is far different. There exist a lot of free text fonts
without additional math characters. This is the reason why we have to buy a
commercial math font, e. g. Palatino (pamath) or Helvetica (hvmath), or to
combine the free text font with another free math font.

48 Computer modern
This is the default font, designed by Knuth. For the PDF output the Type 1
fonts cm-super and BlueSkye were used.

Theorem 1 (Residuum). Für eine in einer punktierten Kreisscheibe D\{a}

analytische Funktion f definiert man das Residuum im Punkt a als
Z
1
Resf (z) = Resf = f (z) dz ,
z=a a 2πi
C

wobei C ⊂ D\{a} ein geschlossener Weg mit n(C, a) = 1 ist (z.B. ein entgegen
dem Uhrzeigersinn durchlaufener Kreis).
AΛ∆∇BCDΣEFΓGHIJKLMNOΘΩfPΦΠΞQRSTUVWXYΥΨZ
aαbβc∂dδeεf ζξgγh~}ιiıjkκκl`λmnηθϑoσςφϕ℘pρ%qrstτ πuµνvυwω$
xχyψz∞ ∝ ∅∅dð 

49 Latin modern
This is the new designed font which comes with an own Type 1 version.lm

Theorem 1 (Residuum). Für eine in einer punktierten Kreisscheibe D\{a}

analytische Funktion f definiert man das Residuum im Punkt a als
Z
1
Resf (z) = Resf = f (z) dz ,
z=a a 2πi
C

wobei C ⊂ D\{a} ein geschlossener Weg mit n(C, a) = 1 ist (z.B. ein entgegen
dem Uhrzeigersinn durchlaufener Kreis).
AΛ∆∇BCDΣEFΓGHIJKLMNOΘΩfPΦΠΞQRSTUVWXYΥΨZ
aαbβc∂dδeεf ζξgγh~}ιiıjkκκl`λmnηθϑoσςφϕ℘pρ%qrstτ πuµνvυwω$
xχyψz∞ ∝ ∅∅dð 

Mathmode.tex 109
 51 PALATINO – MICROIMP

50 Palatino
There is a free package mathpazo.mathpazo

Theorem 1 (Residuum). Für eine in einer punktierten Kreisscheibe D \{ a} ana-

lytische Funktion f definiert man das Residuum im Punkt a als

1
Z
Res f (z) = Res f = f (z) dz ,
z= a a 2πi
C

wobei C ⊂ D \{ a} ein geschlossener Weg mit n(C, a) = 1 ist (z.B. ein entgegen
dem Uhrzeigersinn durchlaufener Kreis).
AΛ∆∇BCDΣEFΓGHIJKLMNOΘΩfPΦΠΞQRSTUVWXYΥΨZ
aαbβc∂dδeeε f ζξgγhh̄}ιiıjkκ κ l `λmnηθϑoσςφϕ℘ pρ$qrstτπuµνvυwωv
xχyψz∞ ∝ ∅∅dð 

51 Palatino – microimp
There is the package pamath for the nonfree palatino font.mathpazo

Theorem 1 (Residuum). Für eine in einer punktierten Kreisscheibe D\{a} ana-

lytische Funktion f definiert man das Residuum im Punkt a als
Z
1
Resf(z) = Resf = f(z) dz ,
z=a a 2πi
C

wobei C ⊂ D\{a} ein geschlossener Weg mit n(C, a) = 1 ist (z.B. ein entgegen
dem Uhrzeigersinn durchlaufener Kreis).
AΛ∆∇BCDΣEFΓ GHIJKLMNOΘΩfPΦΠΞQRSTUVWXYΥ ΨZ
aαbβc∂dδeεfζξgγhħ}ιiıjkκκl`λmnηθϑoσςφϕ℘pρ%qrstτπuµνvυwω$
xχyψz∞ ∝ ∅∅dð 

Mathmode.tex 110
 53 MINION

52 cmbright

Theorem 1 (Residuum). Für eine in einer punktierten Kreisscheibe D\{a} analytis-

che Funktion f definiert man das Residuum im Punkt a als
Z
1
Resf (z) = Resf = f (z) dz ,
z=a a 2πi
C

wobei C ⊂ D\{a} ein geschlossener Weg mit n(C, a) = 1 ist (z.B. ein entgegen dem
Uhrzeigersinn durchlaufener Kreis).
AΛ∆∇BCDΣEFΓ GHIJKLMNOΘΩfPΦΠΞQRSTUVWXYΥΨZ
aαbβc∂dδeεf ζξgγh~}ιi ıjkκκl`λmnηθϑoσςφϕ℘pρ%qr stτ πuµνv υw ω$
xχy ψz∞ ∝ ∅∅dð 

53 minion

Theorem 1 (Residuum). Für eine in einer punktierten Kreisscheibe Dƒ˜a analytische

Funktion f definiert man das Residuum im Punkt a als

∫
1
Res f(z) = Res f = f(z) dz ,
z=a a 2πi
C

wobei C ⊂ Dƒ˜a ein geschlossener Weg mit n(C, a) = 1 ist (z.B. ein entgegen dem
Uhrzeigersinn durchlaufener Kreis).
AΛ∆∇BCDΣEFΓGHIJKLMNOΘΩΩPΦΠΞQRSTUVWXYΥΨZ
aαbβc∂dδeєε fζξgγhħħιiı jȷkκϰlℓλmnηθϑoσςϕφ³pρρqrstτπuµνvυwωϖ
xχyψzª Œ ggdðэ

Mathmode.tex 111
 54 INTEGRAL SYMBOLS

Part VI
Special symbols
In this section only those symbols are defined, which are not part of the list of
all available symbols: CTAN://info/symbols/comprehensive/symbols-a4.pdf.
With fontmath.ltx LATEX itself defines the following special symbols for
using inside math:

Name Meaning
\mathparagraph ¶
\mathsection §
\mathdollar $
\mathsterling £
\mathunderscore
\mathellipsis ...

Table 21: Predefined math symbols from fontmath.ltx

54 Integral symbols
Name Symbol
´
\dashint −
´
\ddashint =
´
\clockint 
´
\counterint

For all new integral symbols limits can be used in the usual way:
ˆ ˆ ˛∞ ˆ ˆ
=1=−0< = (54.1)
0 1 A
−∞

1 \ ddashint _01=\ dashint _10 <\ oint \ limits _{ -\ infty }^\ infty = \
clockint \ counterint _ A

Put the following definitions into the preamble to use one or all of these
new integral symbols.
1 \ def \ Xint #1{\ mathchoice
2 {\ XXint \ displaystyle \ textstyle {#1}} %
3 {\ XXint \ textstyle \ scriptstyle {#1}} %
4 {\ XXint \ scriptstyle \ scriptscriptstyle {#1}} %
5 {\ XXint \ scriptscriptstyle \ scriptscriptstyle {#1}} %
6 \!\ int }
7 \ def \ XXint #1#2#3{{\ setbox 0=\ hbox {$#1{#2#3}{\ int }$}
8 \ vcenter {\ hbox {$#2#3$}}\ kern -.5\ wd 0}}

Mathmode.tex 112
 55 HARPOONS

9 \ def \ ddashint {\ Xint =}

10 \ def \ dashint {\ Xint -}
11 \ def \ clockint {\ Xint \ circlearrowright } % GOOD !
12 \ def \ counterint {\ Xint \ rotcirclearrowleft } % Good for Computer
Modern !
13 \ def \ ro tci rcl ear ro wle ft {\ mathpalette {\ RotLSymbol { -30}}\
circlearrowleft }
14 \ def \ RotLSymbol #1#2#3{\ rotatebox [ origin = c ]{#1}{$#2#3$}}

55 Harpoons
LATEX knows no stretchable harpoon symbols, like \xrightarrow. The
following code defines several harpoon symbols. \xrightharpoondown
\xrightharpoonup
1 \ def \ r i g h t h a r p o o n d o w n f ill@ { %
\xleftharpoondown
2 \ arrowfill@ \ relbar \ relbar \ rightharpoondown }
3 \ def \ r ig h th ar p oo n up fi l l@ { % \xleftharpoonup
4 \ arrowfill@ \ relbar \ relbar \ rightharpoonup } \xleftrightharpoons
5 \ def \ l e f t ha r p o o n d ow n f i ll@ { % \xrightleftharpoons
6 \ arrowfill@ \ leftharpoondown \ relbar \ relbar }
7 \ def \ le fth arp oon up fil l@ { %
8 \ arrowfill@ \ leftharpoonup \ relbar \ relbar }
9 \ newcommand {\ xrightharpoondown }[2][]{ %
10 \ ext@arrow 0359\ r i ght harpoondow nfill@ {#1}{#2}}
11 \ newcommand {\ xrightharpoonup }[2][]{ %
12 \ ext@arrow 0359\ r ightharpoonupfill@ {#1}{#2}}
13 \ newcommand {\ xleftharpoondown }[2][]{ %
14 \ ext@arrow 3095\ l e ftharpoondownfill@ {#1}{#2}}
15 \ newcommand {\ xleftharpoonup }[2][]{ %
16 \ ext@arrow 3095\ le ftharpoonupfill@ {#1}{#2}}
17 \ newcommand {\ xl eft rig htharpoons }[2][]{\ mathrel { %
18 \ raise .22 ex \ hbox { %
19 $\ ext@arrow 3095\ leftharpoonupfill@ {\ phantom
{#1}}{#2}$} %
20 \ setbox 0=\ hbox { %
21 $\ ext@arrow 0359\ right harpoondo wnfill@ {#1}{\ phantom
{#2}}$} %
22 \ kern -\ wd 0 \ lower .22 ex \ box 0} %
23 }
24 \ newcommand {\ xr igh tle ftharpoons }[2][]{\ mathrel { %
25 \ raise .22 ex \ hbox { %
26 $\ ext@arrow 3095\ rightharpoonupfill@ {\ phantom
{#1}}{#2}$} %
27 \ setbox 0=\ hbox { %
28 $\ ext@arrow 0359\ leftharpoondownfill@ {#1}{\ phantom
{#2}}$} %
29 \ kern -\ wd 0 \ lower .22 ex \ box 0} %
30 }

Mathmode.tex 113
 57 STACKED EQUAL SIGN

over
\xrightharpoondown[under]{over} −−−+ uunder
over
\xrightharpoonup[under]{over} −−−* uunder
over
\xleftharpoondown[under]{over} )−−− u under
over
\xleftharpoonup[under]{over} (−−− u under
over
\xleftrightharpoons[under]{over} −−
(−
− −+
− uunder
over
\xrightleftharpoons[under]{over} −
)−
−−−*
− uunder

56 Bijective mapping arrow

→ we can define:
To get something like 
1 \ def \ bijmap { %
2 \ ensuremath { %
3 \ mathrlap {\ rightarrowtail }\ rightarrow %
4 }%
5 }

This uses the \mathrlap definition from section 35.2 on page 72. With
this definition a huge symbol is also possible: {\Huge\bijmap} →.

57 Stacked equal sign
There are several symbols stacked with an equal sign, e.g., \doteq, \equiv
.
or \cong (=, ≡ , ∼
= ). But there are still some missing, which are shown in
table 22 and the following definitions.
def
\eqdef =
!
\eqexcl =
\eqhat =
b

Table 22: New symbols in combination with the equal sign

1 \ newcommand {\ eqdef }{\ ensuremath {\ stackrel {\ mathrm { def }}{=}}}

2 \ newcommand {\ eqexcl }{\ ensuremath {\ stackrel {\ mathrm {!}}{=}}}
3 \ newcommand {\ eqhat }{\ ensuremath {\ widehat {=}}}

Mathmode.tex 114
 58 OTHER SYMBOLS

58 Other symbols
1 \ newcommand *{\ threesim }{ %
2 \ mathrel {\ vcenter {\ offinterlineskip
3 \ hbox {$\ sim $}\ vskip -.35 ex \ hbox {$\ sim $}\ ∼
∼
∼ ABC
vskip -.35 ex \ hbox {$\ sim $}}}}
4 $\ threesim ABC $

Mathmode.tex 115
 59 TUNING MATH TYPESETTING

Part VII
Examples
59 Tuning math typesetting
Chapter 18 of the TEXbook is named "‘Fine Points of Mathematics Typ-
ing"’ [12] and it shows on 20 pages some more or less important facts when
typesetting mathematical expressions. Often inline formulas contain a punc-
tuation character like a dot, comma, colon, etc.. It is a general rule to write
those characters outside the math mode. Compare
a, b, c, d, e, and f 1 $a, b, c, d, e, \textrm{and }f$ \\[5pt]
2 $a$, $b$, $c$, $d$, $e$, and $f$
a, b, c, d, e, and f
Having such math as single expressions enables TEX to insert a linebreak
at several places (see Section 2.6 on page 12).
Writing an ellipses as three single dots, doesn’t look very nice, one should
always use the \ldots command:
1, ..., 10 1 $1,...,10$\\[5pt]
2 $1,\ldots,10$
1, . . . , 10
This is correct as long as on the left and right are a comma as a separator.
For sums the \cdot command should be used instead:
1 + 2 + · · · + 10 1 $1+2+\cdots+10$\\[5pt]
2 $x_n=x_{n-1}=\cdots=n_0=1$
xn = xn−1 = · · · = n0 = 1
For a multiplication it is important which character is used, in european
countries often a centered dot. In such a case it is appropriate not to use the
\cdots command for a ellipsis.
For typesetting integrals or differential equations it makes sense to define
the following short macros:
1 \ newcommand *\ dy {\ ,\ mathrm { d } y }
2 \ newcommand *\ dx {\ ,\ mathrm { d } x }
3 \ newcommand *\ dyx {\ ,\ frac {\ mathrm { d } y }{\ mathrm { d } x }}
4 \ newcommand *\ ds {\ ,\ mathrm { d } s }
5 \ newcommand *\ dt {\ ,\ mathrm { d } t }
6 \ newcommand *\ dst {\ ,\ frac {\ mathrm { d } s }{\ mathrm { d } t }}

ˆ
1 \begin{align*}
F (x) = f (x) dx
2 F(x) &= \int\!f(x)\dx\\
ds 3 v(t) &= \dst\\
v(t) = 4 a(t) &= \frac{\mathrm{d}^2s}{\mathrm{d}t
dt ^2}
d2 s 5 \end{align*}
a(t) = 2
dt

Mathmode.tex 116
 60 MATRIX

ˆ ˆ
G(t) = · · · dx dy . . . 1 \begin{align*}
2 G(t) &= \underbrace{\,\int\cdots\!\int
| {z } \!\!}_D\;\dx\dy\ldots\\
ˆ D 3 u_C(t) &= \int\!\,i_C(t)\dt
uC (t) = iC (t) dt 4 \end{align*}

60 Matrix
60.1 Identity matrix
There are several possibilities to write this matrix. Here is a solution with
the default array environment.
1 \[
2 \left(
  3 \begin{array}{ccccc}
1 4 1\\

 1 0 
 5
6
& 1 & & \text{{\huge{0}}}\\
& & 1\\
1
 
 
7 & \text{{\huge{0}}} & & 1\\
0
 
 1  8 & & & & 1
1 9 \end{array}
10 \right)
11 \]

60.2 System of linear equations

y1 = x11 + x12 + x13 + . . . + x1(n−1) + x1n

y2 = x21 + x22 + x23 + . . . + x2(n−1) + x2n
.. . . . . . .
. = .. + .. + .. + .. + .. + ..
yn−1 = x(n−1)1 + x(n−1)2 + x(n−1)3 + . . . + x(n−1)3 + x(n−1)n
yn = xn1 + xn2 + xn3 + . . . + x(n−1)(n−1) + xnn
1 \[
2 \begin{array}{l@{\:=\:}*{5}{l@{\:+\:}}l}
3 y_1 & x_{11} & x_{12} & x_{13} & \dots & x_{1(n-1)} & x_{1n} \\
4 y_2 & x_{21} & x_{22} & x_{23} & \dots & x_{2(n-1)} & x_{2n} \\
5 \ \vdots &\ \vdots &\ \vdots &\ \vdots &\ \vdots &\ \vdots &\ \vdots\\
6 y_{n-1} & x_{(n-1)1} & x_{(n-1)2} & x_{(n-1)3} & \dots & x_{(n-1)3} & x
_{(n-1)n}\\
7 y_n & x_{n1} & x_{n2} & x_{n3} & \dots & x_{(n-1)(n-1)} & x_{nn}
8 \end{array}
9 \]

Mathmode.tex 117
60.3 Matrix with comments on top 61 CASES STRUCTURE

60.3 Matrix with comments on top

1 \def\rb#1{\rotatebox{90}{$\xleftarrow{#1}$}}
2 \begin{tabular}{c}
3 $\begin{matrix}
←−−−

←−−−

←−−−

←−−−
\rb{text1}&\rb{text1}&\rb{text1}&\rb{text
text1

text1

text1

text1
4
1}\\
  5 \end{matrix}$\\
Xx Yx Zx Tx 6 $\begin{bmatrix}
Xy Yy Zy Ty  7 X_x & Y_x & Z_x & T_x \\
 
Xz Yz Zz Tz  8 X_y & Y_y & Z_y & T_y \\
9 X_z & Y_z & Z_z & T_z \\
0 0 0 1 10 0 & 0 & 0 & 1
11 \end{bmatrix}$
12 \end{tabular}

61 Cases structure
Sometimes it is better to use the array environment instead of amsmath’s
cases environment. To get optimal horizontal spacing for the conditions, there
are two matrixes in series, one 3 × 1 followed by 3 × 3 matrix. To minimize
the horizontal space around the variable z a
1 \ addtolength {\ arraycolsep }{ -3 pt }

is a useful command.

D +z 
  −D ≤ z ≤ −p
z2
I(z) = δ0 D − 21 p − p −p ≤ z ≤ p (61.1)

D−z p ≤z≤ D


1 \ addtolength {\ arraycolsep }{ -3 pt }
2 I ( z ) =\ delta _{0}\ left \{ %
3 \ begin { array }{ lcrcl }
4 D + z & \ quad & -D & \ le z \ le & -p \\
5 D -\ frac {1}{2}\ left (p -\ frac { z ^{2}}{ p }\ right ) %
6 & \ quad & -p & \ le z \ le & \ phantom { -} p \\
7 D - z & \ quad & p & \ le z \ le & \ phantom { -} D
8 \ end { array }\ right .
9 \ end { equation }

The \phantom command replaces exactly that place with whitespace

which the argument needs.

61.1 Cases with numbered lines

This is not possible in an easy way, because cases uses the array environment
for typesetting which has by default no numbering. However, there are some
tricky ways to get numbered lines. The following three examples use the
tabular, the tabularx and the array environment.

Mathmode.tex 118
 62 ARRAYS

 x = 2 if y > 2 (61.2)
some text here
x = 3 if y ≤ 2 (61.3)

1 \ begin { tabular }{ rc }
2 \ ldelim \{{2}{2.75 cm }[ some text here ] &
3 \ parbox {{\ linewidth -3 cm -4\ tabcolsep }}{
4 \ vspace *{1 ex }
5 \ begin { flalign }
6 x & = 2\ quad \ text { if } y >2 &\\
7 x & = 3\ quad \ text { if } y \ le 2&
8 \ end { flalign }}
9 \ end { tabular }


x = 2 if y > 2 (61.4)
some text here
x = 3 if y ≤ 2 (61.5)

1 \ begin { tabularx }{\ linewidth }{ rXc }

2 \ ldelim \{{2}{2.75 cm }[ some text here ]
3 & $ x =2\ quad \ text { if }y >2$ &\ refstepcounter { equation }(\ theequation ) \\
4 & $ x =3\ quad \ text { if } y \ le 2$&\ refstepcounter { equation }(\ theequation )
5 \ end { tabularx }


x = 2 if y > 2 (61.6)
some text here
x = 3 if y ≤ 2 (61.7)

1 \[
2 \ begin { array }{ rc@ {\ qquad } c }
3 \ ldelim \{{2}{2.75 cm }[ some text here ]
4 & x = 2\ quad \ text { if } y > 2 & \ refstepcounter { equation }(\
theequation ) \\
5 & x = 3\ quad \ text { if } y \ le 2& \ refstepcounter { equation }(\
theequation )
6 \ end { array }
7 \]

62 Arrays
There is a general rule that a lot of mathematical stuff should be divided in
smaller pieces. But sometimes it is difficult to get a nice horizontal alignment
when splitting a formula. The following ones uses the array environment to
get a proper alignment.

Mathmode.tex 119
62.1 Quadratic equation 62 ARRAYS

62.1 Quadratic equation

y = x2 + bx + c
b
= x2 + 2 · x + c
2  2  2
2 b b b
= x +2· x+ − +c
2 2 2
|  {z 2 }
b
x+
2   
b 2
 2
b 2
 
b 
= x+ − +c + −c

2 2  2
 2
b 2
 
b
y+ −c = x+ |(Scheitelpunktform)
2 2
y − yS = (x − xS )2 !
 2
b b
S(xS ; yS ) bzw. S − ; −c
2 2
(62.1)
1 \ begin { equation }
2 \ begin { array }{ rcll }
3 y & = & x ^{2}+ bx + c \\
4 & = & x ^{2}+2\ cdot {\ displaystyle \ frac { b }{2} x + c }\\
5 & = & \ underbrace { x ^{2}+2\ cdot \ frac { b }{2} x +\ left (\ frac { b }{2}\ right )
^{2}} -{\ displaystyle %
6 \ left (\ frac { b }{2}\ right ) ^{2}+ c }\\
7 & & \ qquad \ left ( x +{\ displaystyle \ frac { b }{2}}\ right ) ^{2}\\
8 & = & \ left ( x +{\ displaystyle \ frac { b }{2}}\ right ) ^{2} -\ left ({\
displaystyle %
9 \ frac { b }{2}}\ right ) ^{2}+ c & \ left |+\ left ({\ displaystyle %
10 \ frac { b }{2}}\ right ) ^{2} - c \ right .\\
11 y +\ left ({\ displaystyle \ frac { b }{2}}\ right ) ^{2} - c & = & \ left ( x +{\
displaystyle %
12 \ frac { b }{2}}\ right ) ^{2} & \ left |(\ textrm { Scheitelpunktform }) \ right .\\
13 y - y _{ S } & = & (x - x _{ S }) ^{2}\\
14 S ( x _{ S }; y _{ S }) & \ ,\ textrm { bzw .}\ , & S \ left ( -{\ displaystyle %
15 \ frac { b }{2};\ ,\ left ({\ displaystyle \ frac { b }{2}}\ right ) ^{2} - c }\ right )
16 \ end { array }
17 \ end { equation }

Mathmode.tex 120
62.2 Vectors and matrices 62 ARRAYS

62.2 Vectors and matrices

 
01 a4 55 87 5a 58 db 9e
 a4 56 82 f3 1e c6 68 e5 
RS = 
 02

a1 fc c1 47 ae 3d 19 
a4 55 87 5a 58 db 9e 03
 
  m8i+0
si,0
 si,1 
 m8i+1 
  (62.2)
  = RS · 
 si,2   ··· 

 m8i+6 
si,3
m8i+7
P3
Si = j=0 si,j · 28j i = 0, 1, ..., k − 1

S = (Sk−1 , Sk−2 , ..., S1 , S0 )

1 \ begin { equation }
2 \ begin { array }{ rcl }
3 \ underline { RS } & = & \ left (\ begin { array }{ cccccccc }
4 01 & a 4 & 55 & 87 & 5 a & 58 & db & 9 e \\
5 a 4 & 56 & 82 & f 3 & 1 e & c 6 & 68 & e 5\\
6 02 & a 1 & fc & c 1 & 47 & ae & 3 d & 19\\
7 a 4 & 55 & 87 & 5 a & 58 & db & 9 e & 03\ end { array }\ right ) \\
8 \\
9 \ left (\ begin { array }{ c }
10 s _{ i ,0}\\
11 s _{ i ,1}\\
12 s _{ i ,2}\\
13 s _{ i ,3}
14 \ end { array }\ right ) & = & \ underline { RS }\ cdot %
15 \ left (\ begin { array }{ c }
16 m _{8 i +0}\\
17 m _{8 i +1}\\
18 \ cdots \\
19 m _{8 i +6}\\
20 m _{8 i +7}
21 \ end { array }\ right ) \\
22 \\
23 S _{ i } & = & \ sum _{ j =0}^{3} s _{ i , j }\ cdot 2^{8 j }\ qquad i =0 ,1 ,... , k -1\\
24 \\
25 S & = & \ left ( S _{ k -1} , S _{ k -2} ,... , S _{1} , S _{0}\ right )
26 \ end { array }
27 \ end { equation }

62.3 Cases with (eqn)array environment

This solution is important when AMSmath can’t be used.

Mathmode.tex 121
62.4 Arrays inside arrays 62 ARRAYS



 divergent q ≤ −1
0 |q| < 1

lim q n =
n−>∞ 
 1 q = 1
∞ q > 1


1 $\ lim \ limits _{ n - >\ infty } q ^{ n }=\ left \{ %

2 \ begin { array }{ lc@ {\ kern 2 pt } c@ {\ kern 2 pt } r }
3 \ textrm { divergent }\ & q & \ le & -1\\
4 0 & | q | & < & 1\\
5 1 & q & = & 1\\
6 \ infty & q & > & 1
7 \ end { array }\ right .$

62.4 Arrays inside arrays

The array environment is a powerful one because it can be nested in several
ways:
 
a11 a12
 a 0 0
 21 a22


b11 b12 b13
 
 
0 b21 b22 b23 0
 
 
b31 b32 b33
 
 
 
 c11 c12 
0 0
c21 c22

1 \[
2 \ left (
3 \ begin { array }{ c@ {} c@ {} c }
4 \ begin { array }{| cc |}\ hline
5 a _{11} & a _{12} \\
6 a _{21} & a _{22} \\\ hline
7 \ end { array } & \ mathbf {0} & \ mathbf {0} \\
8 \ mathbf {0} &
9 \ begin { array }{| ccc |}\ hline
10 b _{11} & b _{12} & b _{13}\\
11 b _{21} & b _{22} & b _{23}\\
12 b _{31} & b _{32} & b _{33}\\\ hline
13 \ end { array } & \ mathbf {0} \\
14 \ mathbf {0} & \ mathbf {0} &
15 \ begin { array }{| cc |}\ hline
16 c _{11} & c _{12} \\
17 c _{21} & c _{22} \\\ hline
18 \ end { array } \\
19 \ end { array }
20 \ right )
21 \]

Mathmode.tex 122
62.5 Colored cells 62 ARRAYS

 
0 0 1 0
Y1 =  1 0 1 0 
1 1 1 1
2 1 3 1

1 \[
2 Y ^1=
3 \ begin { array }{ c }
4 \ null \\[1 ex ] % only vor vertical alignment
5 \ left [\ begin { array }{ rrrr }
6 0 & 0 & 1 & 0\\
7 1 & 0 & 1 & 0\\
8 1 & 1 & 1 & 1
9 \ end { array }\ right ]\\[3 ex ]\ hline
10 \ begin { array }{ rrrr }
11 % \ hdotsfor {4}\\%( needs \ AmSmath ) instead of \\[3 ex ]\ hline
12 2 & 1 &3 & 1
13 \ end { array }
14 \ end { array }
15 \]

62.5 Colored cells

In general there is no difference in coloring tabular or array cells. The
following example shows how one can put colors in rows, columns and cells.

 
hk,1,0 (n) hk,1,1 (n) hk,1,2 (n) 0 0
hk,2,0 (n) hk,2,1 (n) hk,2,2 (n) 0 0
 
 
 

 hk,3,0 (n) hk,3,1 (n) hk,3,2 (n) 0 0 


 hk,4,0 (n) hk,4,1 (n) hk,4,2 (n) 0 0 

0 hk,1,0 (n − 1) hk,1,1 (n − 1) hk,1,2 (n − 1) 0
 
 
 

 0 hk,2,0 (n − 1) hk,2,1 (n − 1) hk,2,2 (n − 1) 0 

0 hk,3,0 (n − 1) hk,3,1 (n − 1) hk,3,2 (n − 1) 0
 
 
 

 0 hk,4,0 (n − 1) hk,4,1 (n − 1) hk,4,2 (n − 1) 0 

0 0 hk,1,0 (n − 2) hk,1,1 (n − 2) hk,1,2 (n − 2)
 
 
 

 0 0 hk,2,0 (n − 2) hk,2,1 (n − 2) hk,2,2 (n − 2) 


 0 0 hk,3,0 (n − 2) hk,3,1 (n − 2) hk,3,2 (n − 2) 

0 0 hk,4,0 (n − 2) hk,4,1 (n − 2) hk,4,2 (n − 2) 12×5

1 ...
2 \ usepackage { array }
3 \ usepackage { colortbl }
4 \ definecolor { umbra }{ rgb }{0.8 ,0.8 ,0.5}
5 \ def \ zero {\ multicolumn {1}{ >{\ columncolor { white }} c }{0}}
6 \ def \ colCell #1#2{\ multicolumn {1}{ >{\ columncolor {#1}} c }{#2}}

Mathmode.tex 123
62.6 Boxed rows and columns 62 ARRAYS

7 \ begin { document }
8 \[\ left [\ ,
9 \ begin { array }{*{5}{ >{\ columncolor [ gray ]{0.95}} c }}
10 h _{ k ,1 ,0}( n ) & h _{ k ,1 ,1}( n ) & h _{ k ,1 ,2}( n ) & \ zero & \ zero \\
11 h _{ k ,2 ,0}( n ) & h _{ k ,2 ,1}( n ) & h _{ k ,2 ,2}( n ) & \ zero & \ zero \\
12 h _{ k ,3 ,0}( n ) & h _{ k ,3 ,1}( n ) & h _{ k ,3 ,2}( n ) & \ zero & \ zero \\
13 h _{ k ,4 ,0}( n ) } & \ colCell { umbra }{ h _{ k ,4 ,1}( n ) } & h _{ k ,4 ,2}( n ) & \ zero
& \ zero \\
14 \ zero & h _{ k ,1 ,0}( n -1) & h _{ k ,1 ,1}( n -1) & h _{ k ,1 ,2}( n -1) & \ zero \\
15 \ zero & h _{ k ,2 ,0}( n -1) & h _{ k ,2 ,1}( n -1) & h _{ k ,2 ,2}( n -1) & \ zero \\
16 \ zero & h _{ k ,3 ,0}( n -1) & h _{ k ,3 ,1}( n -1) & h _{ k ,3 ,2}( n -1) & \ zero \\
17 \ zero & \ colCell { umbra }{ h _{ k ,4 ,0}( n -1) } & h _{ k ,4 ,1}( n -1) & h _{ k
,4 ,2}( n -1) & \ zero \\
18 \ zero & \ zero & h _{ k ,1 ,0}( n -2) & h _{ k ,1 ,1}( n -2) & h _{ k ,1 ,2}( n -2) \\
19 \ zero & \ zero & h _{ k ,2 ,0}( n -2) & h _{ k ,2 ,1}( n -2) & h _{ k ,2 ,2}( n -2) \\
20 \ zero & \ zero & h _{ k ,3 ,0}( n -2) & h _{ k ,3 ,1}( n -2) & h _{ k ,3 ,2}( n -2) \\
21 \ zero & \ zero & h _{ k ,4 ,0}( n -2) & h _{ k ,4 ,1}( n -2) & h _{ k ,4 ,2}( n -2)
22 \ end { array } \ ,\ right ]_{12\ times 5}\]
23 ...

62.6 Boxed rows and columns

1 \[
2 \overrightarrow{A}=\left[
  3 \begin{array}{cccc}
1 2 3 4 4 1 & 2 & 3 & 4\\
→ 
− 1 2 3 4  5 1 & 2 & 3 & 4\\\hline
A =
 1
 6 \multicolumn{1}{|c}{1} & 2 & 3 &
2 3 4  7 \multicolumn{1}{c|}{4}\\\hline
1 2 3 4 8 1 & 2 & 3 & 4
9 \end{array}\right]
10 \]

1 \[
2 \overrightarrow{A}=\left[
  3 \begin{array}{cc|c|c}\cline{3-3}
1 2 3 4 4 1 & 2 & 3 & 4\\
→ 
− 1 2 3 4  5 1 & 2 & 3 & 4\\
A =
 1

2 3 4  6 1 & 2 & 3 & 4\\
1 2 3 4 7 1 & 2 & 3 & 4\\\cline{3-3}
8 \end{array}\right]
9 \]

1 \[
2 \overrightarrow{A}=\left[
  3 \begin{array}{cc|c|c}\cline{3-3}
1 2 3 4 4 1 & 2 & 3 & 4\\
→ 
− 1 2 3 4  5 1 & 2 & 3 & 4\\\hline
A =
 1
 6 \multicolumn{1}{|c}{1} & 2 & 3 &
2 3 4  7 \multicolumn{1}{c|}{4}\\\hline
1 2 3 4 8 1 & 2 & 3 & 4\\\cline{3-3}
9 \end{array}\right]
10 \]

Mathmode.tex 124
 63 OVER- AND UNDERBRACES

63 Over- and underbraces

63.1 Braces and roots
To put an underbrace in a root without enlarging the root symbol is possible
with the \makebox macro:
p
z= x2 + y 2
| {z }
=z 2

1 \[
2 z =\;\;\ underbrace { %
3 \ makebox [\ widthof {~$ x ^2+ y ^2$}][ r ]{ %
4 $\ sqrt { x ^2+ y ^2}$}}_{= z ^2}
5 \]

63.2 Overlapping braces

o
Overlapping under- and overbraces like | needs some
z }| {
{z } | {z }
u1 u2

tricky code, because we cannot have parts of the argument inside overbrace
and also underbrace. The following equation 63.1 is an example for such a
construction:

y = 2x2 − 3x + 5
=0
 z }| { 
 2   2
3 3 3 5
= 2 x2 − x + − + (63.1)
2 4 4 2
| {z }| {z }
 2 !
3 31
=2 x− +
4 16
3 2
 
31
y− = 2 x−
8 4

1 \ begin { align }\ label { eq : pqFormel }

2 y &= 2 x ^2 -3 x +5\ nonumber \\
3 & \ hphantom {= \ 2\ left ( x ^2 -\ frac {3}{2}\ , x \ right . } %
4 \ textcolor { blue }{ %
5 \ overbrace {\ hphantom {+\ left (\ frac {3}{4}\ right ) ^2 - %
6 \ left (\ frac {3}{4}\ right ) ^2}}^{=0}}\ nonumber \\[ -11 pt ]
7 &= 2\ left (\ textcolor { red }{ %
8 \ underbrace { %
9 x ^2 -\ frac {3}{2}\ , x + \ left (\ frac {3}{4}\ right ) ^2} %

Mathmode.tex 125
63.3 Vertical alignment 63 OVER- AND UNDERBRACES

10 }%
11 \ underbrace { %
12 - \ left (\ frac {3}{4}\ right ) ^2 + \ frac {5}{2}} %
13 \ right ) \\
14 &= 2\ left (\ qquad \ textcolor { red }{\ left (x -\ frac {3}{4}\ right ) ^2}
15 \ qquad + \ \ frac {31}{16}\ qquad \ right ) \ nonumber \\
16 y \ textcolor { blue }{ -\ frac {31}{8}}
17 &= 2\ left ( x \ textcolor { cyan }{ -\ frac {3}{4}}\ right ) ^2\ nonumber
18 \ end { align }

63.3 Vertical alignment of different braces

When having several braces in one formula line, then it looks better when all
braces are also on the same line, e.g.,
      
xR sin γ − cos γ xK tx
= r · + (63.2)
yR cos γ sin γ yK ty
|{z} | {z } | {z }
Scaling Rotation Translation

1 \ begin { equation }
2 \ binom { x _ R }{ y _ R } = \ underbrace { r \ vphantom {\ binom { A }{ B
}}}_{\ text { Skaling }}\ cdot %
3 \ underbrace { %
4 \ begin { pmatrix }
5 \ sin \ gamma & -\ cos \ gamma \\
6 \ cos \ gamma & \ sin \ gamma \\
7 \ end { pmatrix } %
8 }_{\ text { Rotation }}
9 \ binom { x _ K }{ y _ K } +
10 \ underbrace {\ binom { t _ x }{ t _ y }}_{\ text { Translation }}
11 \ end { equation }

It is again the \vphantom macro which reserves the needed vertical space.
Nevertheless the horizontal space around the r of the first underbrace and
the last + should be decreased to get a better typesetting. This is possible
with \hspace or simply \kern:
      
xR sin γ − cos γ xK tx
= r· +
yR cos γ sin γ yK ty
|{z}| {z } | {z }
Skaling Rotation Translation

1 \[ \ binom { x _ R }{ y _ R } = %
2 \ kern -10 pt \ underbrace { r \ vphantom {\ binom { A }{ B }}}_{\ text { Skaling }}\
kern -10 pt %
3 \ cdot \ underbrace { %
4 \ begin { pmatrix }
5 \ sin \ gamma & -\ cos \ gamma \\
6 \ cos \ gamma & \ sin \ gamma \\
7 \ end { pmatrix } %
8 }_{\ text { Rotation }}

Mathmode.tex 126
63.4 Alignment 63 OVER- AND UNDERBRACES

9 \ binom { x _ K }{ y _ K } +\ kern -5 pt %
10 \ underbrace {\ binom { t _ x }{ t _ y }}_{\ text { Translation }} \]

63.4 Vertical and horizontal alignment

The forgoing example simply uses \hspace to decrease the horizontal width
between two underbraces. This may be okay for a single solution, but in
general it is better to have some code which works in any case.
The following example looks simple but it needs some tricky code to get
vertical and horizontal alignment.

300 29 19 9 8 1 1
7−→ 7−→ 7 →
− 7−→ 7 → . . . 7−→
− 7−→ . . . 7−→
5069 |{z} 490 | 321
{z } 152 | 135 {z } 16 | {z }1
∆a=271 ∆a=10 =h271i29 ∆a=1 =h10i9 ∆a=0=h1i1
∆b=4579 ∆b=169=h4579i490 ∆b=17=h169i152 ∆b=1=h17i16
1 iteration 2 iterations 8 iterations 8 iterations

It uses the macro \mathclap defined in section 35.2 on page 72 , which

gives a better result. It is also possible to use \makebox[0pt]{...} but it
works only in text mode and this needs some more $...$.
1 \ def \ num #1{\ hphantom {#1}}
2 \ def \ vsp {\ vphantom {\ rangle _1}}
3
4 \ begin { equation *}
5 \ frac {300}{5069} %
6 \ underbrace {\ longmapsto \ vphantom {\ frac {1}{1}}}_{ %
7 \ mathclap {\ substack { %
8 \ Delta a =271\ num 9\ vsp \\[2 pt ]
9 \ Delta b =4579\ vsp \\[2 pt ]
10 \ text {$1$ iteration } %
11 }}} \ frac {29}{490} %
12 \ underbrace {\ longmapsto \ frac {19}{321}\ longmapsto }_{ %
13 \ mathclap {\ substack { %
14 \ Delta a =10\ num {9}=\ langle 271\ rangle _{29}\ num {20}\\[2 pt ]
15 \ Delta b =169=\ langle 4579\ rangle _{490}\\[2 pt ]
16 \ text {$2$ iterations }
17 }}} \ frac {9}{152}
18 \ underbrace {\ longmapsto \ frac {8}{135}\ longmapsto \ dots \ longmapsto
}_{ %
19 \ substack { %
20 \ Delta a =1\ num {7}=\ langle 10\ rangle _{9}\ num {119}\\[2 pt ]
21 \ Delta b =17=\ langle 169\ rangle _{152}\\[2 pt ]
22 \ text {$8$ iterations }
23 }} \ frac {1}{16}
24 \ underbrace {\ longmapsto \ dots \ longmapsto \ vphantom {\ frac {8}{135}}}_{
%
25 \ substack { %
26 \ Delta a =0=\ langle 1\ rangle _{1}\ num {76} \\[2 pt ]
27 \ Delta b =1=\ langle 17\ rangle _{16} \\[2 pt ]

Mathmode.tex 127
 65 VERTICAL ALIGNMENT

28 \ text {$8$ iterations }

29 }} \ frac {1}{1}
30 \ end { equation *}

64 Integrals
The first theorem of Green is:
˚ ‹
 2 ∂v 2
u∇ v + (∇u, ∇v) d3 V =

u d A
∂n
G S

The second theorem of Green is:

˚ ‹  
 2 2
 3 ∂v ∂u
u∇ v − v∇ u d V = u −v d2 A
∂n ∂n
G S

They are both written with the esint package36 , which gives nice integral
symbols. The LATEX code for the first equation is:
1 \[
2 \ underset {\ mathcal { G }\ quad }\ iiint \! %
3 \ left [ u \ nabla ^{2} v +\ left (\ nabla u ,\ nabla v \ right ) \ right ]\ mathrm { d
}^{3} V %
4 =\ underset {\ mathcal { S }\ quad }\ oiint u \ ,\ Q { v }{ n }\ ,\ ,\ mathrm { d }^{2} A
5 \]

with the following definition in the preamble for the partial derivation:
1 \ def \ Q #1#2{\ frac {\ partial #1}{\ partial #2}}

which makes things easier to write.

65 Vertical alignment
65.1 Example 1
Sometimes it may be useful to have a vertical alignment over the whole page
with a mix of formulas and text. Section 37 shows the use of \intertext.
There is another trick to get all formulas vertical aligned. Let’s have the
following formulas distributed over the whole page:

f (x) = a
g(x) = x2 − 4x
f (x) − g(x) = x2 + x3 + x
g = x2 + x3 + x4 + x5 + b
36
See section 64.

Mathmode.tex 128
65.1 Example 1 65 VERTICAL ALIGNMENT

They all have a different length of the left and right side. Now we want
to write some text and other objects between them, but let the alignment
untouched. We choose the longest left and the longest right side and take
them for scaling with the \hphantom command:

\hphantom{\mbox{$f(x)-g(x)$}} & \hphantom{\mbox{$= x^2+x^3+x^4+x^5+b$}}

This is the first (empty) line in every equation where now all other lines
are aligned to this one. For example:

blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah

f (x) = a (65.1)
2
g(x) = x − 4x (65.2)

blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah

f (x) − g(x) = x2 + x3 + x (65.3)

blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah

g(x) = x2 + x3 + x4 + x5 + b (65.4)

blah blah blah blah blah blah blah blah blah blah blah blah blah blah blah
blah blah blah blah blah blah blah blah blah

The phantom line is empty but leaves the vertical space for a line. This
could be corrected with decreasing the \abovedisplayshortskip length and
done all inside a group.
1 \ newcommand {\ x }{ blah blah blah blah blah blah blah blah }
2 \ bgroup
3 \ addtolength \ a b o v e d i s p laysh ortskip { -0.5 cm } % decrease the skip
4 \ addtolength \ abovedisplayskip { -0.5 cm }
5 \x\x\x
6 \ begin { align }
7 \ hphantom {\ mbox {$ f ( x ) -g ( x ) $}} & \ hphantom {\ mbox {$= x ^2+ x ^3+ x ^4+ x ^5+ b
$}}\ nonumber \\
8 f ( x ) &= a \\
9 g ( x ) &= x ^2 -4 x
10 \ end { align }
11 %
12 \x\x\x

Mathmode.tex 129
65.2 Example 2 65 VERTICAL ALIGNMENT

13 \ begin { align }
14 \ hphantom {\ mbox {$ f ( x ) -g ( x ) $}} & \ hphantom {\ mbox {$= x ^2+ x ^3+ x ^4+ x ^5+ b
$}}\ nonumber \\
15 f ( x ) -g ( x ) &= x ^2+ x ^3+ x
16 \ end { align }
17 \x\x\x
18 %
19 \ begin { align }
20 \ hphantom {\ mbox {$ f ( x ) -g ( x ) $}} & \ hphantom {\ mbox {$= x ^2+ x ^3+ x ^4+ x ^5+ b
$}}\ nonumber \\
21 g ( x ) &= x ^2+ x ^3+ x ^4+ x ^5+ b
22 \ end { align }
23 \x\x\x
24 \ egroup

Another case of aligning equations inside an itemize environment is the

following one. With the \makebox macro one can have the same size on the
left side of the equal sign to get a vertical alignment.

• first function X
P1 = ∈A
a

• but another one

sin (P1 ) = blabla

• or perhaps
P3 + P2 − P1 = blablub

1 \ newsavebox \ lW
2 \ sbox \ lW {$ P _{3}+ P _{2} - P _{1}$}
3
4 \ begin { itemize }
5 \ item first function \\
6 $\ displaystyle \ makebox [\ wd \ lW ][ r ]{$ P _1$}=\ sum _ a \ in A $
7 \ item but another one \\
8 $\ makebox [\ wd \ lW ][ r ]{$\ sin \ left ( P _1\ right ) $}= blabla $
9 \ item or perhaps \\
10 $ P _{3}+ P _{2} - P _{1}= blablub $
11 \ end { itemize }

65.2 Example 2
This one comes from Hartmut Henkel and offers a special form of placing
additional text between the equation and the equation number. This makes
only sense when you load the documentclass with the option fleqn. The
example places the additional text at 0.5\textwidth, changing this value is
no problem.

Mathmode.tex 130
65.2 Example 2 65 VERTICAL ALIGNMENT

text text text text text text text text text text text text text text text text text
text text text text text text text text text text text text text text text text text
text text text text text text text text text text text text text text text text text
text text text text text

a0 Bohrscher Radius (= 0,53 Å)

 2 2
− 21 e Elementarladung
E · 4 · π · ε0 · a0 · Zi + ZSi
3 3
Nsi Anzahl der Siliziumatome
ε=   ; (65.5)
pro Einheitsvolumen
Zi · ZSi · e2 · 1 + mmSii
m Atomgewicht
Z Kernladungszahl

a2 + b2 = c2 abc (65.6)

z=9 (65.7)

text text text text text text text text text text text text text text text text text
text text text text text text text text text text text text text text text text text
text text text text text text text text text text text text text text text text text
text text text text text

This solution works only with AMSmath, without you have to redefine
the LATEX macro, which creates the equation number.
1 \ newsavebox {\ myendhook } % for the tabulars
2 \ def \ tagform@ #1{{(\ maketag@@@ {\ ignorespaces #1\ unskip \ @@italiccorr ) }
3 \ makebox [0 pt ][ r ]{ % after the equation number
4 \ makebox [0.4\ textwidth ][ l ]{\ usebox {\ myendhook }} %
5 }%
6 \ global \ sbox {\ myendhook }{} % clear box content
7 }}
8 [ ... ]
9 \ sbox {\ myendhook }{ %
10 \ begin { footnotesize } %
11 \ begin { tabular }{ @ {} ll }
12 $ a _0$ & Bohrscher Radius ($\ mathrm {= 0{ ,}53\ ,\ mbox {\ AA }}$) \\
13 $ e $ & Elementarladung \\
14 $ N _{ si }$ & Anzahl der Siliziumatome \\
15 & pro Einheitsvolumen \\
16 $ m $ & Atomgewicht \\
17 $ Z $ & Kernladungszahl
18 \ end { tabular }
19 \ end { footnotesize }}
20 %
21 \ begin { equation }
22 \ varepsilon = \ frac { E \ cdot 4 \ cdot \ pi \ cdot \ varepsilon _{0}
23 \ cdot a _0 \ cdot \ left ( Z _ i ^{\ frac {2}{3}} + Z _{ Si }^{\ frac {2}{3}}
24 \ right ) ^{ -\ frac {1}{2}}} { Z _ i \ cdot Z _{ Si } \ cdot e 2 \ cdot \ left ( 1

Mathmode.tex 131
 66 NODE CONNECTIONS

25 + \ frac { m _ i }{ m _{ Si }} \ right ) }\ ,;
26 \ end { equation }
27 %
28 \ sbox {\ myendhook }{ abc }
29 %
30 \ begin { equation } a 2+ b 2= c 2 \ end { equation }
31 %
32 \ begin { equation } z = 9 \ end { equation }

66 Node connections
This is a typical application for PSTricks and it needs the package pst-node
and doesn’t work with pdflatex. Use vlatex, ps4pdf or ps2pdf.
Die Bindungsenergie im Tröpfchenmodell setzt sich aus folgenden Teilen zu-
sammen:

• dem Oberflächenanteil

• dem Volumenanteil,

2
E = av A + − af A2/3 + − ac Z(Z−1)
A1/3
+ − as (A−2Z)
A
+ Ep (1)

• dem Coulomb-Anteil

• der Symmetrieenergie

• sowie einem Paarbildungsbeitrag.

1 \ psset { nodesep =3 pt }
2 \ definecolor { lila }{ rgb }{0.6 ,0.2 ,0.5}
3 \ definecolor { darkyellow }{ rgb }{1 ,0.9 ,0}
4 Die Bindungsenergie im Tr \" opfchenmodell setzt sich aus
5 folgenden Teilen zusammen :
6 \ begin { itemize }
7 \ item dem \ rnode { b }{ Oberfl \" achenanteil }
8 \ item dem \ rnode { a }{ Volumenanteil } ,\\[1 cm ]
9 \ def \ xstrut {\ vphantom {\ frac {( A ) ^1}{( B ) ^1}}}
10 \ begin { equation }
11 E =
12 \ rnode [ t ]{ ae }{\ psframebox *[ fillcolor = darkyellow ,
13 linestyle = none ]{\ xstrut a _ vA }} +
14 \ rnode [ t ]{ be }{\ psframebox *[ fillcolor = lightgray ,
15 linestyle = none ]{\ xstrut -a _ fA ^{2/3}}} +
16 \ rnode [ t ]{ ce }{\ psframebox *[ fillcolor = green ,
17 linestyle = none ]{\ xstrut -a _ c \ frac { Z (Z -1) }{ A ^{1/3}}}} +

Mathmode.tex 132
 67 SPECIAL PLACEMENT

18 \ rnode [ t ]{ de }{\ psframebox *[ fillcolor = cyan ,

19 linestyle = none ]{\ xstrut -a _ s \ frac {( A -2 Z ) ^2}{ A }}} +
20 \ rnode [ t ]{ ee }{\ psframebox *[ fillcolor = yellow ,
21 linestyle = none ]{\ xstrut E _ p }}
22 \ end { equation }\\[0.25 cm ]
23 \ item dem \ rnode { c }{ Coulomb - Anteil }
24 \ item der \ rnode { d }{ Symmetrieenergie }
25 \ item sowie einem \ rnode { e }{ Paarbildungsbeitrag }.
26 \ end { itemize }
27 \ nccurve [ angleA = -90 , angleB =90]{ - >}{ a }{ ae }
28 \ nccurve [ angleB =45]{ - >}{ b }{ be } \ nccurve [ angleB = -90]{ - >}{ c }{ ce }
29 \ nccurve [ angleB = -90]{ - >}{ d }{ de } \ nccurve [ angleB = -90]{ - >}{ e }{ ee }

67 Special placement of displayed equations

67.1 Formulas side by side
Sometimes it may be useful to have numbered formulas side by side like the
following ones:

˛
Eds = 0 (67.1.a) ∇·B =0 (67.1.b)
c
a= (67.2.a) b=1 (67.2.b)
d
ˆ
c=1 (67.3.a) 2x dx = x2 + C (67.3.b)

And again a default display equation:

ˆ ∞
1
F (x) = dx (67.4)
0 x

1 \ begin { mtabular }{*{2}{ m {0.35\ linewidth } m {0.15\ linewidth }}}

2 \ begin { align *} \ oint E ds =0 \ end { align *} & \ eqnCnt %
3 & \ begin { align *} \ nabla \ cdot B =0 \ end { align *} & \ eqnCnt [\ label {
blah }]\\
4 \ begin { align *} a =\ frac { c }{ d } \ end { align *} & \ eqnCnt %
5 & \ begin { align *} b = 1 \ end { align *} & \ eqnCnt \\
6 \ begin { align *} c =1 \ end { align *} & \ eqnCnt [\ label { blub }]
7 & \ begin { align *} \ int 2 x \ ,\ mathrm { d } x = x ^2+ C \ end { align *} & \
eqnCnt
8 \ end { mtabular }

The new environment mtabular has two arguments, one optional and one
which is the same as the one from the tabular environment. With the option
long it is possible to have all the formulas in a longtable environment,
which allows a pagebreak. The new macro \eqnCnt controls the counting of

Mathmode.tex 133
67.1 Formulas side by side 67 SPECIAL PLACEMENT

these equations as subequations for one tabular line. This macro can have
an optional argument for a label. At least it counts the equations. If the
equation number is not centered to the foregoing equation, then it needs
some more horizontal space in the tabular column.

\eqnCnt[<optional label>]

The vertical space is controlled by the length mtabskip, which is by

default -1.25cm and can be modified in the usual way. To define all these
macros write into the preamble:
1 \ usepackage { amsmath }
2 \ newcounter { subequation }
3 \ newlength \ mtabskip \ mtabskip = -1.25 cm
4 \ newcommand \ eqnCnt [1][]{ %
5 \ refstepcounter { subequation } %
6 \ begin { align }#1\ end { align } %
7 \ addtocounter { equation }{ -1}}
8 \ def \ mtabLong { long }
9 \ makeatletter
10 \ newenvironment { mtabular }[2][\ empty ]{ %
11 \ def \ @xarraycr { %
12 \ stepcounter { equation } %
13 \ setcounter { subequation }{0} %
14 \ @ifnextchar [\ @argarraycr {\ @argarraycr [\ mtabskip ]}}
15 \ let \ theoldequation \ theequation %
16 \ renewcommand \ theequation {\ theoldequation .\ alph { subequation }}
17 \ edef \ mtabOption {#1}
18 \ setcounter { subequation }{0} %
19 \ tabcolsep =0 pt
20 \ ifx \ mtabOption \ mtabLong \ longtable {#2}\ else \ tabular {#2}\ fi %
21 }{ %
22 \ ifx \ mtabOption \ mtabLong \ endlongtable \ else \ endtabular \ fi %
23 \ let \ theequation \ theoldequation %
24 \ stepcounter { equation }}
25 \ makeatother

As seen in equation 67.3.a and equation 67.1.b, everything of the table

contents is nonsense . . . And the following tabular is defined as a longtable to
enable pagebreaks.

˛
Eds = 0 (67.5.a) ∇·B =0 (67.5.b)
c
a= (67.6.a) b=1 (67.6.b)
d
ˆ
c=1 (67.7.a) 2x dx = x2 + C (67.7.b)

Mathmode.tex 134
67.2 Itemize environment 67 SPECIAL PLACEMENT

˛
Eds = 0 (67.8.a) ∇·B =0 (67.8.b)
c
a= (67.9.a) b=1 (67.9.b)
d
ˆ
c=1 (67.10.a) 2x dx = x2 + C (67.10.b)
˛
Eds = 0 (67.11.a) ∇·B =0 (67.11.b)
c
a= (67.12.a) b=1 (67.12.b)
d
ˆ
c=1 (67.13.a) 2x dx = x2 + C (67.13.b)
˛
Eds = 0 (67.14.a) ∇·B =0 (67.14.b)
c
a= (67.15.a) b=1 (67.15.b)
d
ˆ
c=1 (67.16.a) 2x dx = x2 + C (67.16.b)

As seen in equation 67.13.a and equation 67.11.b, everything is nonsense

...
And again a default display equation:
ˆ ∞
1
F (x) = dx (67.17)
0 x

1 \ begin { mtabular }[ long ]{*{2}{ m {0.375\ linewidth } m {0.125\

linewidth }}}
2 \ begin { align *} \ oint E ds =0 \ end { align *} & \ eqnCnt %
3 & \ begin { align *} \ nabla \ cdot B =0 \ end { align *} & \ eqnCnt \\
4 \ begin { align *} a =\ frac { c }{ d } \ end { align *} & \ eqnCnt %
5 & \ begin { align *} b = 1 \ end { align *} & \ eqnCnt \\
6 \ begin { align *} c =1 \ end { align *} & \ eqnCnt
7 & \ begin { align *} \ int 2 x \ ,\ mathrm { d } x = x ^2+ C \ end { align
*} & \ eqnCnt \\
8
9 [ ... ]

67.2 Formulas inside an itemize enviroment

Without any modification it is not possible to get a numbered equation at
the same height as the symbol of the itemize environment. This depends on

Mathmode.tex 135
67.2 Itemize environment 67 SPECIAL PLACEMENT

the \abovedisplayskip. The formula has to be raised up for exactly this

length.
1 \ def \ itemMath #1{ %
2 \ raisebox { -\ a b o v e d isplaysho rtskip }{ %
3 \ parbox {0.75\ linewidth }{ %
4 \ begin { equation }#1\ end { equation }}}}
5 %
6 \ begin { itemize }
7 \ item \ itemMath { f = l }
8 \ item \ itemMath { g ( x ) = \ int f ( x ) \ ,\ mathrm { d } x }
9 \ end { itemize }

• f =l (67.18)
ˆ
• g(x) = f (x) dx (67.19)

Mathmode.tex 136
Part VIII
Lists, bibliography and index

Mathmode.tex 137
List of Figures
Figure Page
1 multline Alignment demo (the fourth row is shifted to the
right with \shoveright) . . . . . . . . . . . . . . . . . . . . . 60
2 Demonstration of \multlinegap (default is 0pt) . . . . . . . 60

Mathmode.tex 138
List of Tables
Table Page
1 Meaning of \mathsurround . . . . . . . . . . . . . . . . . . . 13
2 Difference between the default \bigg and the \biggm command 27
3 Use of the different parentheses for the “big” commands . . . 28
4 Old font style commands . . . . . . . . . . . . . . . . . . . . . 31
5 Fonts in math mode . . . . . . . . . . . . . . . . . . . . . . . 32
6 The meaning of the math spaces . . . . . . . . . . . . . . . . 32
7 Spaces in math mode . . . . . . . . . . . . . . . . . . . . . . . 33
8 Math styles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
9 Dots in math mode . . . . . . . . . . . . . . . . . . . . . . . . 39
10 Accents in math mode . . . . . . . . . . . . . . . . . . . . . . 39
11 Vectors with package esvec (in the right column the default
one from LATEX) . . . . . . . . . . . . . . . . . . . . . . . . . 42
12 The predefined operators of fontmath.ltx . . . . . . . . . . . 43
13 The predefined operators of latex.ltx . . . . . . . . . . . . . 43
14 The greek letters . . . . . . . . . . . . . . . . . . . . . . . . . 45
15 Comparison between the different align environments with the
same code, where the first three can have an equation number 51
16 Matrix environments . . . . . . . . . . . . . . . . . . . . . . . 65
17 binom commands . . . . . . . . . . . . . . . . . . . . . . . . . 68
18 The modulo commands and their meaning . . . . . . . . . . . 69
19 Different mathcommands . . . . . . . . . . . . . . . . . . . . 79
20 The predefined operators of amsopn.sty . . . . . . . . . . . . 98
21 Predefined math symbols from fontmath.ltx . . . . . . . . . 112
22 New symbols in combination with the equal sign . . . . . . . 114

Mathmode.tex 139
References
[1] Paul W. Abrahams, Karl Berry, and Kathryn Hargreaves. TEX for the
Impatient. http://tug.org/ftp/tex/impatient/book.pdf, 2003.

[2] Claudio Beccari. Typesetting mathematics for science and technology

according to iso 31/xi. TUGboat Journal, 18(1):39–47, 1997.

[3] Thierry Bouche. Diversity in math fonts. TUGboat Journal,

19(2):121–135, 1998.

[4] David Cobac. Atelier documents mathématiques.

http://dcobac.free.fr/latex/Presentation4.pdf, 2004.

[5] David Cobac. Ecrire des mathématiques avec LATEX.

http://dcobac.free.fr/latex/prepDocMaths.pdf, 2004.

[6] Michael Downes. Technical Notes on the amsmath package. American

Mathematical Society,
ftp://ftp.ams.org/pub/tex/doc/amsmath/technote.pdf, 1999.

[7] Michael Downes. Short Math Guide for LATEX. American Mathematical
Society, http://www.ams.org/tex/short-math-guide.html, 2002.

[8] Victor Eijkhout. TEX by Topic. http://www.eijkhout.net/tbt/,

1992.

[9] J. Anthony Fitzgerald. Web Math Formulas Using TEX.

http://www.unb.ca/web/Sample/math/, 1997.

[10] Michel Goosens, Frank Mittelbach, and Alexander Samarin. The LATEX
Companion. Addison Wesley, 13 edition, 1994.

[11] George Grätzer. Math into LATEX. Birkhäuser Boston, third edition,
2000.

[12] Donald E. Knuth. The TEXbook. Addison Wesley Professional, 21

edition, 1986.

[13] Donald E. Knuth, Tracy Larrabee, and Paul M. Roberts. Mathematical

Writing. Stanford University, Computer Science Department, http:
//sunburn.stanford.edu/~knuth/papers/mathwriting.tex.gz,
1987.

[14] R. Kuhn, R. Scott, and L. Andreev. An Introduction to using LATEX in

the Harvard Mathematics Department. Harvard University, Department
of Mathematics, http:
//abel.math.harvard.edu/computing/latex/manual/texman.html.

Mathmode.tex 140
[15] Johannes Küster. Designing Math Fonts.
http://www.typoma.com/publ/20040430-bachotex.pdf, apr 2004.
Vortrag auf der polnischen TeX-Konferenz »BachoTeX«.

[16] Johannes Küster. Fonts for Mathematics.

http://www.typoma.com/publ/20041002-atypi.pdf, oct 2004.
Vortrag auf der ATypI-Konferenz in Prag.

[17] Richard Lawrence. Math=Typography? TUGboat Journal,

24(2):165–168, 2003.

[18] NIST. Typefaces for Symbols in Scientific Manuscripts.

http://physics.nist.gov/Document/typefaces.pdf, 2004.

[19] Luca Padovani. Mathml formatting with tex rules and tex fonts.
TUGboat Journal, 24(1):53–61, 2003.

[20] Sebastian Rahtz and Leonor Barroca. A style option for rotated objects
in LATEX. TUGboat Journal, 13(2):156–180, July 1992.

[21] Steve Seiden. Math cheat sheet. TUG,

http://www.tug.org/texshowcase/#math, 2000.

[22] Carole Siegfried and Herbert Voß. Mathematik im Inline-modus. Die

TEXnische Komödie, 3/04:25–32, November 2004.

[23] Paul Taylor. Commutative Diagrams in TEX. Department of Computer

Science, Queen Mary and Westfield College,
http://www.dcs.qmw.ac.uk/~pt/diagrams/, 2000.

[24] Herbert Voß. Farbige Mathematik. Die TEXnische Komödie,

2/04:81–87, March 2004.

[25] Herbert Voß. LATEX in Naturwissenschaften & Mathematik. Franzis

Verlag, München, first edition, 2006.

Mathmode.tex 141
Index
Symbols \bar, 39
$, 10–12 \belowdisplayshortskip, 35
\!, 107 \belowdisplayshortskip, 82
\(, 10 \belowdisplayskip, 35
\), 10 \belowdisplayskip, 36, 82
\„ 33 \bf, 31
\:, 33 \Big, 26
\;, 33 \big, 26
\[, 14, 36 Bigg, 28
\], 14, 36 \Bigg, 26
\bigg, 26
A \Biggm, 27
\above, 90 \biggm, 27
\abovedisplayshortskip, 35 \Bigl, 26
\abovedisplayshortskip, 82 \bigl, 26
\abovedisplayskip, 35 \Bigm, 27
\abovedisplayskip, 36, 82, 136 \bigm, 27
\abovewithdelims, 91 \bigr, 26
\acute, 39 Binom, 46
\acute, 39 \Binomial, 102
aligned, 57 \binoppenalty, 94
alignedat, 57 \Bmatrix, 65
Alignment \bmatrix, 65
– left, 54 Bold greek letters, 79
\allowdisplaybreaks, 45 \boldmath, 47
\ArcCos, 101 \boldmath, 47
\ArcCot, 101 \bordermatrix, 22
\ArcCsc, 101 \bordermatrix, 22
\ArcSec, 101 \boxed, 78
\ArcSin, 101 boxed inline math, 12
\ArcTan, 101 Braces, 100
array, 56 – Problems, 73
array, 22, 37, 97–99 \Braket, 100
\arraycolsep, 20, 21 \breve, 39
\arraystretch, 37 \breve, 39
Arrows, 77
\atop, 72 C
\atop, 24, 46, 91 \cal, 31
\atopwithdelims, 91 \cancel, 101
Cases
B
– numbered lines, 118
\bar, 39

142
\cases, 20, 63 \dfrac, 11
\catcode, 86 differential equation, 116
\cdot, 116 Display math mode, 10
\cdots, 39 \displaybreak, 45
\cdots, 116 displayindent, 84
centertags, 50 \displayindent, 83
\cfrac, 66 \displaylimits, 91
\chapter, 11 displaymath, 20
\check, 39 \displaystyle, 67
\check, 39 \displaystyle, 11, 38, 86, 88
\choose, 46 \displaywidowpenalty, 94
\clap, 73 \displaywidth, 84
cmex10, 29 dot, 34, 116
cmr, 109 \dot, 39
colon, 116 \dot, 39
Color, 46 \doteq, 114
color, 123 \dotsb, 39
\colorbox, 12 \dotsc, 39
\columncolor, 123 \dotsi, 39
comma, 34, 116 \dotsm, 39
Computer modern, 109 \dotso, 39
\cong, 114 double stroke, 31
\Cos, 101 dsfont, 31
\Cot, 101
\Csc, 101 E
ctagsplit, 71 ellipses, 116
empheq, 103, 104
D \ensuremath, 46
\DblFactorial, 102 Environment
\ddddot, 65 – array, 22, 37, 97–99
\dddot, 39, 65 – description, 48
\dddot, 39 – displaymath, 20
\ddot, 39 – empheq, 103, 104
\ddot, 39 – eqnarray, 15, 20, 36, 51, 103
\ddots, 39 – equation, 20, 36
decimal point, 34 – gathered, 107
\DeclareMathDelimiter, 86 – graphicx, 107
\delcode, 86 – lgathered, 107
Delimiter, 14, 29 – matrix, 24
\delimiter, 86 – tabular, 20
\delimiterfactor, 30, 82 \eqdef, 46
\delimitershortfall, 30, 83 eqnarray, 15, 20, 36, 51, 103
description, 48 \eqno, 92
\dfrac, 67 Equation

Mathmode.tex 143
– number, 71 greek, 45
– numbering, 69 – bold, 45
equation, 20, 36 – upright, 45
Equation number, 71
equation number, 130 H
\equiv, 114 Harpoon, 113
\everydisplay, 49 \hat, 39
\everydisplay, 92 \hat, 39
\everymath, 49 \hcancel, 100
\everymath, 92 \hdotsfor, 65
Exponent, 42 Helvetica, 109
Extensible arrows, 77 \hphantom, 33, 127
\hphantom, 33, 34
F \hspace, 33
\Factorial, 102 \Huge, 38
\fam, 87 \huge, 106
\fbox, 12, 18, 103 \HurwitzZeta, 102
\fboxrule, 12 \Hypergeometric, 102
\fboxsep, 12
fleqn, 50 I
fleqn, 36 \imath, 40
Font \IncGamma, 102
– cmex10, 29 Indentation, 54, 83
– lm, 109 Indice, 42
fontmath.ltx, 29 \int, 11, 43, 106
Fontsize, 38 Integral, 116
\frac, 46, 107 Integral symbols, 112
Fraction, 11, 66, 107 \intertext, 76
\frac, 66 intlimits, 50
Framed inline math, 12 \intop, 43
\it, 31
G Italic, 30, 74
\GammaFunc, 102 \item, 48
gather, 57 itemize, 135
gathered, 57
gathered, 107 J
\genfrac, 66 \jmath, 40
\GenIncGamma, 102 \jot, 36, 37
\GenRegIncGamma, 102
K
\GenRegIncGammaInv, 102
\kern, 33
graphicx, 107
\grave, 39 L
\grave, 39 Label, 18
Greek, 44 \label, 71

Mathmode.tex 144
\labelwidth, 83 – \Cot, 101
\Large, 38 – \Csc, 101
large, 38 – \DblFactorial, 102
Latin modern, 109 – \DeclareMathDelimiter, 86
\ldelim, 99 – \Factorial, 102
\ldots, 39 – \GammaFunc, 102
\ldots, 116 – \GenIncGamma, 102
\left, 26, 92, 99 – \GenRegIncGammaInv, 102
Left aligned, 55 – \GenRegIncGamma, 102
Length – \Huge, 38
– \abovedisplayskip, 36, 136 – \HurwitzZeta, 102
– \arraycolsep, 20, 21 – \Hypergeometric, 102
– \belowdisplayskip, 36 – \IncGamma, 102
– \delimitershortfall, 30 – \Large, 38
– \fboxrule, 12 – \LogGamma, 102
– \fboxsep, 12 – \MathieuCharacteristicA, 102
– \jot, 36, 37 – \MathieuCharacteristicB, 102
– \mathsurround, 13 – \MathieuCharacteristicExponent,
– \topsep, 36 102
leqno, 50 – \MathieuCharisticA, 102
\leqno, 92 – \MathieuCharisticB, 102
lgathered, 107 – \MathieuCharisticExp, 102
\lim, 11 – \MathieuC, 102
Limits, 24, 44, 72, 74 – \MathieuS, 102
limits, 11 – \MeijerG, 102
\limits, 11, 43, 92, 93 – \Multinomial, 102
lm, 109 – \Pochhammer, 102
\LogGamma, 102 – \RegHypergeometric, 102
longtable, 134 – \RegIncGammaInv, 102
– \RegIncGamma, 102
M – \RiemannSiegelTheta, 102
Macro – \RiemannSiegelZ, 102
– \ArcCos, 101 – \RiemannZeta, 102
– \ArcCot, 101 – \Sec, 101
– \ArcCsc, 101 – \Set, 100
– \ArcSec, 101 – \Sin, 101
– \ArcSin, 101 – \StieltjesGamma, 102
– \ArcTan, 101 – \Tan, 101
– \Biggm, 27 – \Zeta, 102
– \Bigl, 26 – \[, 14, 36
– \Bigm, 27 – \], 14, 36
– \Binomial, 102 – \acute, 39
– \Braket, 100 – \allowdisplaybreaks, 45
– \Cos, 101 – \arraystretch, 37

Mathmode.tex 145
– \atop, 24, 46 – \ldots, 116
– \bar, 39 – \left, 26, 99
– \biggm, 27 – \limits, 11, 43, 93
– \bigl, 26 – \lim, 11
– \bigm, 27 – \makebox, 25, 130
– \bigr, 26 – \mathbf, 48, 99
– \boldmath, 47 – \mathchar, 88
– \bordermatrix, 22 – \mathcode, 86
– \breve, 39 – \mathlarger, 108
– \cancel, 101 – \mathord, 35
– \catcode, 86 – \mathpunct, 35
– \cdots, 116 – \mathring, 39
– \cdot, 116 – \mathrlap, 114
– \chapter, 11 – \mathversion, 47
– \check, 39 – \mbox, 30
– \choose, 46 – \medmuskip, 32, 34
– \colorbox, 12 – \nicefrac, 107
– \cong, 114 – \nolimits, 43
– \dddot, 39 – \nonumber, 14–16
– \ddot, 39 – \ointop, 43
– \delimiterfactor, 30 – \oint, 43
– \dfrac, 11 – \overbrace, 39, 41
– \displaybreak, 45 – \overbracket, 40
– \displaystyle, 11, 38, 88 – \overleftarrow, 39
– \doteq, 114 – \overleftrightarrow, 39
– \dot, 39 – \overline, 39
– \ensuremath, 46 – \overrightarrow, 39
– \eqdef, 46 – \overset, 25
– \equiv, 114 – \parbox, 30
– \everydisplay, 92 – \part, 11
– \fbox, 12, 18, 103 – \pmb, 44
– \frac, 46, 107 – \prod, 11, 24
– \grave, 39 – \rdelim, 99
– \hat, 39 – \right, 26, 99
– \hcancel, 100 – \scalebox, 107
– \hphantom, 33, 34 – \scriptscriptstyle, 38, 88
– \huge, 106 – \scriptstyle, 38, 88
– \imath, 40 – \section, 11, 48
– \intop, 43 – \show, 34
– \int, 11, 43, 106 – \smash, 26
– \item, 48 – \sqrt, 25
– \jmath, 40 – \stackrel, 45
– \labelwidth, 83 – \sum, 11, 24, 106
– \ldelim, 99 – \texorpdfstring, 12

Mathmode.tex 146
– \textstyle, 30, 38, 88 \MathieuCharacteristicExponent, 102
– \thickmuskip, 33 \MathieuCharisticA, 102
– \thinmuskip, 32, 34 \MathieuCharisticB, 102
– \tilde, 39 \MathieuCharisticExp, 102
– \underbar, 39 \MathieuS, 102
– \underbrace, 39, 41 \mathindent, 50
– \underbracket, 40 \mathinner, 93
– \underleftarrow, 39 \mathit, 31
– \underleftrightarrow, 39 \mathlarger, 108
– \underline, 39 \mathop, 89
– \underrightarrow, 39 \mathopen, 89
– \underset, 25 mathopen symbol, 27
– \vec, 39 \mathord, 35, 89
– \vert, 29 \mathpunct, 35, 89
– \vphantom, 25, 99 \mathrel, 89
– \vskip, 37 \mathring, 39
– \widehat, 39 \mathring, 39
– \widetilde, 39 \mathrlap, 114
– \xymatrix, 108 \mathrm, 31, 75
\makebox, 25, 130 \mathsf, 31
\mapstofill, 77 \mathsurround, 13, 84
Math operator, 11 \mathtt, 31
Math unit, 69 \mathversion, 47
math unit, 32 \mathversion, 47
\mathaccent, 87 \matrix, 65
\mathbb, 31 matrix, 24
\mathbf, 31 \mbox, 75
\mathbf, 48, 99 \mbox, 30
\mathbin, 87 \medmuskip, 32, 34, 84
\mathcal, 31 \medspace, 33
\mathchar, 87, 88 \MeijerG, 102
\mathchardef, 88 \mkern, 84
\mathchoice, 88 \mskip, 84
\mathclap, 73, 127 \Multinomial, 102
\mathclose, 88 Multiple exponents, 42
\mathcode, 86, 88 multline, 61
\mathds, 31 \multlinegap, 60
\mathfrak, 31 \muskip, 85
Mathieu \muskipdef, 85
– Characteristics, 102
– Functions, 102 N
\MathieuC, 102 namelimits, 50
\MathieuCharacteristicA, 102 \negmedspace, 33
\MathieuCharacteristicB, 102 \negthickspace, 33

Mathmode.tex 147
\negthinspace, 33 – color, 100
\nicefrac, 107 – cool, 101
nointlimits, 50 – delarray, 103
\nolimits, 43, 93 – empheq, 104
nonamelimits, 50 – esint, 128
\nonscript, 85 – esvect, 42
\nonumber, 14–16 – eucal, 105
nosumlimits, 50 – exscale, 39, 106, 108
\nulldelimiterspace, 85 – hvmath, 109
– hyperref, 12
O – icomma, 35
\oint, 43 – leqno, 17
\ointop, 43 – mathpazo, 110
Operator, 43 – mathtools, 106, 107
– names, 74 – multirow, 97
– size, 107 – nicefrac, 107
\operatornamewithlimits, 74 – numprint, 34
\over, 93 – pamath, 109, 110
\overbrace, 39, 125 – pstricks, 97
\overbrace, 39, 41 – relsize, 108
\overbracket, 40 – remreset, 16
\overleftarrow, 39 – setspace, 37
\overleftarrow, 39 – units, 107
\overleftrightarrow, 39 – upgreek, 45
\overleftrightarrow, 39 – wasysym, 104
\overline, 39 – xypic, 97, 108
\overline, 39, 93 – xy, 108
\overrightarrow, 39, 42 Pagebreak, 45
\overrightarrow, 39 Palatino, 109, 110
\overset, 79 \parbox, 64
\overset, 25 \parbox, 30
\overwithdelims, 93 \part, 11
\phantom, 118
P
\pmatrix, 65
Package
\pmb, 79
– accent, 97
\pmb, 44
– amscd, 97
\Pochhammer, 102
– amsmath, 90
Polylogarithm, 102
– amsopn, 97
\postdisplaypenalty, 95
– amssymb, 34, 39
\predisplaypenalty, 95
– bigdelim, 98
\predisplaystyle, 85
– bm, 45, 48, 99
\prod, 11, 24
– braket, 29, 99
punctuation, 116
– calc, 18

Mathmode.tex 148
Q \smallmatrix, 65
\qquad, 33 \smash, 26
\quad, 33 Spacing
– vertical, 37
R Split equation, 56
\radical, 93 \sqrt, 25
\rdelim, 99 Stacked limits, 73
Reference, 18 \stackrel, 45
\reflectbox, 39 \StieltjesGamma, 102
\RegHypergeometric, 102 Style, 38
\RegIncGamma, 102 Subequations, 70
\RegIncGammaInv, 102 Subscript, 11
\relpenalty, 95 \substack, 72, 73, 127
reqno, 50 \sum, 74
\RiemannSiegelTheta, 102 \sum, 11, 24, 106
\RiemannSiegelZ, 102 sumlimits, 50
\RiemannZeta, 102 Superscript, 11
\right, 26, 94, 99
righttag, 71 T
\rm, 31 tabular, 20
Root, 25, 68 \tag, 18
\rowcolor, 123 \Tan, 101
tbtags, 50
S \texorpdfstring, 12
\scalebox, 107 Text, 30
\scriptfont, 89 – \parbox, 30
\scriptscriptfont, 89 \textfont, 90
\scriptscriptstyle, 38, 88, 90 \textstyle, 30, 38, 88, 90
\scriptspace, 85 \tfrac, 67
scriptstyle, 11 \thickmuskip, 33, 85
\scriptstyle, 11, 67 \thickspace, 33
\scriptstyle, 38, 88, 90 \thinmuskip, 32, 34, 85
\Sec, 101 \thinspace, 33
\section, 11, 48 \tilde, 39
\Set, 100 \tilde, 39
Set symbol, 31 \topsep, 36
\shoveright, 60 \tt, 31
\show, 34
\sideset, 74 U
\Sin, 101 \unboldmath, 47
Size \underbar, 39
– Operator, 107 \underbar, 39
\skew, 90 \underbrace, 39, 125, 127
\skewchar, 90 \underbrace, 39, 41

Mathmode.tex 149
\underbracket, 40 – Functions, 102
\underleftarrow, 39 – Hurwitz, 102
\underleftarrow, 39 – Riemann, 102
\underleftrightarrow, 39 \Zeta, 102
\underleftrightarrow, 39
\underline, 39, 49
\underline, 39, 94
\underrightarrow, 39
\underrightarrow, 39
\underset, 79, 128
\underset, 25
Upright letters, 42
\uproot, 68

V
\vcenter, 94
\vdots, 39
\vec, 39
\vec, 39
Vector, 42, 121
\vert, 29
Vertical spacing, 37
\Vmatrix, 65
\vmatrix, 65
\vphantom, 127
\vphantom, 25, 99
\vskip, 37

W
\widehat, 39
\widehat, 39
\widetilde, 39
\widetilde, 39

X
\xleftharpoondown, 113
\xleftharpoonup, 113
\xleftrightharpoons, 113
\xrightharpoondown, 113
\xrightharpoonup, 113
\xrightleftharpoons, 113
\xymatrix, 108

Z
Zeta, 102

Mathmode.tex 150