Astronomy – General Relativity and Gravitation I

Notation
Norman Gray
Public version 1.1, June 2013

[Public version 1.1]

GRG I, notation

Context matters. . . .
Parts 2 and 3 introduce a great deal of sometimes confusing notation. The best way to get
used to this is: to get used to it, by working through examples, but while you’re slogging there,
this crib might be useful.
GRG I, notation Tensors

Tensors
A .MN / tensor is a linear function of M one-forms and N vectors, which turns them into a
number. A .10/ tensor is called a vector, and a .01/ tensor is a one-form. Vectors are written with
a bar over them, V , and one-forms with a tilde e p (§2-2.1). In my (informal) notation in the
lectures, T.e
;
/ is a .11/ tensor – a machine with a single one-form-shaped slot and a single
vector-shaped slot. Note that this is a different beast from T.
; e
/, which is also a .11/ tensor,
but with the slots differently arranged.
GRG I, notation Coordinates and components

Coordinates and components

In a space of dimension n, a set of n linearly independent vectors e i (i D 1; : : : ; n) forms a
basis for all vectors in the space; a set of n linearly independent one-forms e ! i forms a basis
for all one-forms in the space.
! j / D ıi j , e
e i .e ! j .e i / D ı j i : Choose basis vectors and one-forms (2.3a)
!i
to be dual (remember that e i and e
are functions)
V D V 0e0 C V 1e1 C


Vectors have components, written §2-2.5
with raised indexes
! 0 C p1 e
p D p0 e
e !1 C


. . . so do one-forms, but written with
lowered indexes
! i /;
V i D V .e p .e i /
pi D e Components of vectors and one-forms (2.4)
(a consequence of the above)
T i j D T.e
! i ; ej / Tensors have components, too (2.5)
j j
T i D T.e i ; e
! / A different beast (note the
arrangement of indexes)
The object T i j is a number – a component of a tensor in a particular basis. However
we also (loosely, wickedly) use this same notation to refer to the corresponding matrix of
numbers, and even to the corresponding .11/ tensor T.
The vector space in which these objects live is the tangent plane to the manifold M at
the point P , TP .M / (§3-1.1). In this space, the basis vectors are e i D @=@x i (§3-1.2), and
the basis one-forms e dx i , where x i is the i -th coordinate (more precisely, coordinate function;
note that x i is not a component of any vector, though the notation makes it look a bit like one).
These bases are dual: e i .e ! i / D @=@x i .e
dx j / D ı i j (cf, (3.6)).
GRG I, notation Contractions

Contractions
A contraction is a tensor with some or all of its arguments filled in.
p/ D e
V .e p .V / by choice §2-2.5
p .V /  he
e p; V i special notation for vectors contracted (2.1)
with one-forms
p .V / D pi V i
e basis independent §2-2.5
p ; e
; V /j D pi V k T ij k
T.e partially contracted .21/ tensor (a §2-2.5
vector)
p D g.e
p ; e
/ a vector, with components. . . §3-2.3
j i j j
g.e ! / D pi g.e
p; e ! /p
! ;e definition of vector p, with raised
indexes, dual to one-form ep , written
with lowered indexes
gij D g.e i ; e j /; components of the metric (2.8)
ij i j
g D g.e
! ;e
! / . . . up and down
gij g kl T j l D T i k the metric raises and lowers indexes;

;
/ and T.
; e
/ are distinct but
T.e
related
i i ij
gij ; g j D ı j; g different tensors in principle, but all (2.10)
referred to as ‘the metric’
GRG I, notation Differentiation

Differentiation
V i ;j  @V i =@x j ; pi;j  @pi =@x j (non-covariant) derivative of a
component (no surprises)
rV covariant derivative of V §3-2.2
(a .11/ tensor)
rU V covariant derivative of V in the cf. (3.33)
direction U (a vector)
rei V  ri V shorthand
.rV /i j D V i Ij components of the .11/ tensor rV (3.18)
p /ij D piIj
.re components of the .02/ tensor re
p
V iIj D g j k V i Ik it’s a tensor, so you can raise its
indexes, too
rU V D U ˛ V ˇ I˛ e ˇ . . . putting all that together
GRG I, notation Changing bases

Changing bases
We might change from a basis e i (for example e 0 ; : : : ; e 3 ) to a basis e {N (e 0N ; : : : ; e 3N ), noting
that the bar goes over the index and not, as might be more intuitive, the vector (that is, we
don’t write e i or e 0i ). The transformation is described using a matrix (it’s not a tensor) ƒi|N
(§2-2.7).
e {N D ƒi{N e i ; e
! {N D ƒ{iN e
!i transformation of basis vectors and §3-1.4
one-forms
V {N D ƒ{iN V i ; p{N D ƒi{N pi transformation of components (2.12)
ei D ƒ|iN e|N D ƒ|iN ƒj|N e j the transformation matrix goes in
both directions
) ƒ|iN ƒj|N D ıij matrix inverses (2.18)
Notes: (1) these look complicated to remember, but as long as each ƒ has one barred
and one unbarred index, you’ll find there’s only one place to put each index, consistent with
the summation convention. (2) This shows why it’s useful to have the bars on the indexes.
(3) Some books use hats for the alternate bases: e {O .
GRG I, notation Einstein’s summation convention

Einstein’s summation convention

See §2-2.5.
X
pi V i  pi V i :
i

The convention works only for two repeated indexes, one up, one down. This is one of
the reasons why one-form components are written with lowered indexes and vectors with
raised ones; the other is to distinguish the components pi of the one-form e
p from the compo-
nents p i D g ij pj of the related vector p D g.e
p ; e
/. Points to watch:
 A term should have at most two duplicate indexes, one up, one down; if you find
something like U i V i or U i Ti i , you’ve made a mistake.
 All the terms in an expression should have the same unmatched index(es): Ai D B j T i j C
C i is all right, Ai D B j T k j is a mistake (typo or thinko).
 You can change or swap repeated indexes: Aij Tij , Ai k Ti k and Aj i Tj i all mean exactly
the same thing, but all are different from Aij Tj i (unless T happens to be symmetric).

However, sometimes we will refer to particular components of a tensor or matrix, such

as referring to the diagonal elements of the metric as g i i – there’s no summation convention
here, so the proscriptions above aren’t relevant. Context again – sorry.
GRG I, notation Miscellaneous

Miscellaneous
(
i ij 1 if i D j
ı j Dı D ıij D Kronecker delta symbol §2-1.1
0 otherwise
 D diag. 1; 1; 1; 1/ the metric (tensor) of Minkowski
space
ŒA; B D AB BA the ‘commutator’ before (3.51)
Note: ƒi|N
and €ji k
are matrices, not the components of tensors, so the indexes don’t
correspond to arguments, and so don’t have to be staggered (though I sometimes do, in the
notes, for no particularly good reason).
Different authors adopt different conventions for the ‘signature’ of the metric (the sum
of the signs on the diagonal), and adopt different conventions for the Riemann tensor (3.43).
This ‘+2’ convention matches Schutz.
In general, component indexes are roman letters, i , j , and so on. When discussing
specifically spacetime, it is traditional but not universal to use greek letters such as , , ˛,
ˇ, and so on, for component indexes ranging over f0; 1; 2; 3g, and roman letters for indexes
ranging over the spacelike components f1; 2; 3g.
Components are usually standing in for numbers, however we’ll sometimes replace them
with letters when a particular coordinate system suggests them. For example e x rather than
 1
e 1 in the context of cartesian coordinates, or write € rather than, say, €22 when writing
the Christoffel symbols for coordinates .; /. There shouldn’t be confusion (context, again),
because x, y,  and  are never used as (variable) component indexes; see eg. (3.13).

Copyright 2002–13, Norman Gray. These notes are made available under the Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International Licence. To view a copy of this licence,
visit http://creativecommons.org/licenses/by-nc-nd/4.0/. The home URL of these notes is
http://purl.org/nxg/text/general-relativity