Escuela Militar de Ingeniera

Mariscal Antonio Jos de Sucre

Materia
Optimizacin
Mercury 2!"
#ocente
$uri %amorano &raun
2!!'
Mercury 2.00 manual
===================
(C) Copyright 1990 Roger Schlafly.
All Rights Resere!.
"ortions of the soft#are (C) Copyright 19$%&$$ 'orlan! (nternational.
All Rights Resere!.
Author)s *ote+ ,his is a !raft of the Mercury 2.00 manual- an! #as the
latest ascii ersion. .nce ( typeset it- it no longer /ecame practical
to up!ate this ersion. Registere! copies of Mercury come #ith a 90
page typeset manual #hich is more complete an! has fe#er errors.
Contents
&&&&&&&&
1. (ntro!uction
0eatures
1istory
Share#are policy
2isclaimer
0iles
1ar!#are Re3uirements
(nstallation
2. "ro/lems
4!iting
*um/ers
.perators
Constants
5. Soling
Solutions
6raphs
Reports
7. 0eature Reference
0unctions
Statements
Comman!s
1. (ntro!uction
(ntro!uction
&&&&&&&&&&&&
Mercury is a program for soling e3uations on the ('M "C. (t is
easy to use- interactie- an! po#erful. 8ith it- you can
9 ealuate mathematical e:pressions.
9 sole for the roots of an e3uation.
9 sole a system of e3uations.
9 ma:imi;e or minimi;e a function- #ith or #ithout constraints.
9 ealuate !eriaties an! !efinite integrals.
9 plot one or more functions.
9 print a report or a graph.
Mercury has a /uilt&in e!itor- pull&!o#n menus- online help-
an! all the coneniences necessary to ma<e it accessi/le to
computer noices.
0or a 3uic< !emonstration of Mercury)s capa/ilities- run M4RC=R>
an! then loa! an! sole some of the sample pro/lems (9.4?A). (f
you get stuc<- @ust press the 01 <ey for help.
Mercury is an outgro#th of 4ure<a- a pro!uct of 'orlan!
(nternational of Scotts Aalley- California. 8hen
'orlan! !eci!e! to concentrate on sprea!sheet an! !ata/ase
pro!ucts- 4ure<a #as spun off. Mercury #as
!eelope! from that co!e /ase- an! is /eing release! in this
form to satisfy customer re3uests for ne# ersions of 4ure<a. (t
is la/elle! ersion 2.0 /ecause it is similar to #hat ersion 2.0 of
4ure<a #oul! hae /een.
4ure<a #as also aaila/le on the Apple Macintosh. *o ersion
of Mercury for the Macintosh is planne! at this time.
Mercury is aaila/le as Share#are. Share#are is a name for
programs #hich are !istri/ute! freely or at nominal charge- /ut
#hich are copyrighte! an! hae restrictions on their usage.
Mercury may /e copie! freely for e:amination purposes. 2istri/utors
may charge a nominal fee- proi!e! that they a!here to the con!itions in
A4*2.R.2.C. Commercial an! institutional use is prohi/ite!
#ithout a registere! license. Such a license may /e o/taine! /y
sen!ing B79 to
Real Soft#are
". 'o: 1C$0
So3uel- CA 9D0%5
=SA
phone or fa:+ 70$&7%C&5DD0
CompuSere+ %CC7C-525
6enie+ R.Schlafly
''S+ 70$&7%C&10DD
"ayment may /e /y chec<- MasterCar!- or Aisa. See R46(S,4R.2.C
for an or!er form. 0oreign or!ers shoul! inclu!e BD for shipping.
,hose #ho register a license #ill get the latest ersion of the
soft#are- an installation program- an! a printe! manual. ,hey #ill
/e eligi/le for technical support- an! #ill /e on the mailing list
for future information an! up!ates.
Site licenses an! other special arrangements are aaila/le.
Comments- /ug reports- an! suggestions are especially #elcome.
2isclaimer
&&&&&&&&&&
,he author has ma!e his /est effort in pro!ucing this soft#are- /ut
here/y !isclaims all #arranties. 1e #ill not /e lia/le for any !amages
e:cee!ing the price pai! for the license to use the soft#are. ,he
person using the soft#are /ears all ris< as to the 3uality an!
performance of the soft#are.
Mercury is not a 'orlan! pro!uct. 'orlan! #ill not proi!e any
technical support. 4ure<a is a registere! tra!emar< of 'orlan!
(nternational. 'orlan! copyrights an! tra!emar<s are use! #ith
permission. Mercury is a tra!emar< of the author.
0iles
&&&&&
Mercury consists of these files.
M4RC=R>.4E4 main program
M4RC=R>.1F" help file
9.'6( graphics !riers
9.C1R fonts
9.4?A sample pro/lems
MA*=AF user manual (this file)
R4A2M4 3uic< instructions
R46(S,4R.2.C registration form
A4*2.R.2.C en!or information
Au:illiary files+
9.4?A more sample pro/lems
9.'6( more graphics !riers
9.'(* e:ternal functions
9.C source to e:ternal functions
MA,1F.1 tools for ma<ing e:ternal functions
9..'G tools for ma<ing e:ternal functions
9.F(' tools for ma<ing e:ternal functions
(f you !o#nloa!e! Mercury from an online serice- you may fin! the
au:illiary files archie! separately. ,he e:tra files are not nee!e!
/y most people- an! are store! separately to sae on connect time.
,he e:tra !riers are for uncommon har!#are only- an! the tools for
ma<ing e:ternal functions re3uire 'orlan!)s ,ur/o CHH.
Registere! customers also get (*S,.C.M- an installation program
for configura/le options.
1ar!#are Re3uirements
&&&&&&&&&&&&&&&&&&&&&
Mercury re3uires an ('M "C or clone #ith at least C70?
of RAM. A/out D00? of free RAM is nee!e!. (n a!!ition-
e:ten!e! memory (4MS memory) is use! if present. (f a
math coprocessor is aaila/le- it #ill /e use!. A math
coprocessor is recommen!e! for serious num/er crunchers.
"lots re3uire a graphics monitor an! a C6A- 46A- A6A- or
1ercules car!.
>ou can print reports an! graphs on any of the popular
printers. Select "rinter an! Mo!el from the .ptions menu
for the list of supporte! printers.
(nstallation
&&&&&&&&&&&&
,here are no special installation re3uirements. ,he simplest
proce!ure is to ma<e a ne# !irectory on your har! !is<- an! to
copy all of the files there. >ou may #ish to put the Mercury
!irectory on your path- or to copy some of the files to your
RAM !is<.
,he program (*S, may /e use! to change the e!it <ey /in!ings-
an! to ma<e a fe# other installation choices. Aarious other
configuration options are aaila/le from #ithin Mercury- an!
these may /e sae! to !is<. See .ptions- /elo#.
Mercury tries to !etect automatically #hat har!#are is present-
/ut there some situations #here you may #ant to oerri!e the
!etection logic.
(f you hae a C6A monitor- it may or may not hae the Isno#I
pro/lem. (f so- you may notice ran!om sno# on the screen as
characters are #ritten there. (f so- you can eliminate it /y
typing
set C6AS*.8C1?=>
at the 2.S prompt /efore loa!ing Mercury. ,here must not /e any spaces
surroun!ing the e3uals sign.
Mercury automatically !etects the math coprocessor- /ut
occasionally Mercury fails /ecause the chip is !efectie. (t can
/e turne! off #ith
set $%=*
at the 2.S prompt.
Mercury also !etects arious graphics car!s- /ut some cannot /e
!etecte!. See the SCR44* statement in Chapter 7.
(f you hae a graphics printer- you #ill #ant to select it at the
"rinter menu. ,he graphics car! an! printer selection can /e
sae! to the M4RC=R>.C06 file- an! then it #ill /e automatic each
time Mercury is loa!e!.
4rrors
&&&&&&
4rror messages are pretty much self&e:planatory. (f an error is
!etecte! #hile parsing the pro/lem file- Mercury puts you in the
4!it #in!o# #ith the cursor at the point of the error. ,he error
message at the top of the #in!o#- an! !isappears as soon as you
start e!iting. .ther errors are !isplaye! in a special /o: #hich
is place! on the screen. After you ac<no#le!ge the error /y
pressing the 4scape <ey- you are put /ac< in the menu system.
2. "ro/lems
A pro/lem is entere! into Mercury /y either loa!ing an e:isting
one from the !is<- or /y creating a ne# one in the 4!it #in!o#. ,his
chapter e:plains ho# to use the e!itor- an! ho# to create a pro/lem
for soling.
4!iting
&&&&&&&
Mercury inclu!es a complete te:t e!itor for creating an!
mo!ifying pro/lem files. =sage of the e!itor is similar
to 8or!Star in non!ocument mo!e- or to
'orlan!)s Si!e?ic<. ,he pro/lem files are
or!inary ascii files- an! can /e e!ite! #ith any other
te:t e!itor.
>ou can moe the cursor aroun! in the e!itor #ith the arro#
<eys- "age =p- "age 2o#n- an! some other <eys liste! /elo#. ,he
e!itor has t#o typing mo!es- (nsert an! .er#rite. (n (nsert
mo!e- anything you type is inserte! at the cursor position. (n
.er#rite mo!e- pressing a <ey replaces the current character
at the cursor position #ith the ne# type! character. (nitially-
the e!itor operates in (nsert mo!e- a !efault you
can change #ith the (*S, program. ,o s#itch mo!es- press the
(nsert <ey or Ctrl&A. (Ctrl&A means to hol! !o#n the Ctrl <ey
on your <ey/oar! #hile you press A. ,he A nee! not /e in upper
case.)
*o 8or!&8rap
=nli<e a more stan!ar! #or! processing program- the e!itor !oes
not hae #or!&#rap. >ou must press 4nter at the en! of each line.
Auto(n!ent
,he e!itor has an autoin!ent feature that allo#s su/se3uent lines
to automatically in!ent. (f you leae seeral spaces at the
start of a line- use Ctrl&.( to in!ent su/se3uent lines the same
amount. ,he comman! se3uence is a toggle- so you nee! only
press Ctrl&.( again to stop in!enting.
Joom
Sometimes you may #ant to see more than the area coere! /y the
current #in!o#. >ou can I;oomI the #in!o#- e:pan!ing it to the
si;e of the entire screen- /y pressing 0D. ,o return the #in!o#
to its usual si;e- press 0D again. 0or other #in!o# comman!s-
such as moing an! resi;ing a #in!o#- see the 8in!o# menu- /elo#.
1ere is a summary of the e!iting comman!s. Many of these
comman!s can /e change! #ith the (*S, installation program.
Cursor Moement Comman!s+
Character left Ctrl&S or Feft arro#
Character right Ctrl&2 or Right arro#
8or! left Ctrl&A
8or! right Ctrl&0
Fine up Ctrl&4 or =p arro#
Fine !o#n Ctrl&E or 2o#n arro#
Scroll up Ctrl&8
Scroll !o#n Ctrl&J
"age up Ctrl&R
"age !o#n Ctrl&C
Feft of line Ctrl&KS or 1ome
Right of line Ctrl&K2 or 4n!
,op of screen Ctrl&K4 or Ctrl&1ome
'ottom of screen Ctrl&KE or Ctrl&4n!
,op of file Ctrl&KR or Ctrl&"g=p
'ottom of file Ctrl&KC or Ctrl&"g2n
Moe to /loc< /egin Ctrl&K'
Moe to /loc< en! Ctrl&K?
Moe to preious point Ctrl&K"
Moe to mar<er 0 Ctrl&K0
Moe to mar<er 1 Ctrl&K1
Moe to mar<er 2 Ctrl&K2
Moe to mar<er 5 Ctrl&K5
Set mar<er 0 Ctrl&?0
Set mar<er 1 Ctrl&?1
Set mar<er 2 Ctrl&?2
Set mar<er 5 Ctrl&?5
(nsert L 2elete Comman!s+
(nsert mo!e onMoff Ctrl&A or (ns
(nsert line Ctrl&*
2elete line Ctrl&>
2elete to en! of line Ctrl&K>
2elete character left Ctrl&1 or 'ac<space
2elete character Ctrl&6 or 2el
2elete #or! right Ctrl&,
'loc< comman!s+
Set /loc< /egin Ctrl&?' or 0%
Set /loc< en! Ctrl&?? or 0$
Mar< #or! Ctrl&?,
1i!e /loc< Ctrl&?1
Copy /loc< Ctrl&?C
Moe /loc< Ctrl&?A
2elete /loc< Ctrl&?>
Rea! /loc< Ctrl&?R
8rite /loc< Ctrl&?8
"rint /loc< Ctrl&?"
Miscellaneous Comman!s+
Kuit e!it- no sae Ctrl&?2- Ctrl&?K- or 4sc
Sae an! e!it Ctrl&?S or 02
*e# file 05
,a/ Ctrl&( or ,a/
,a/ mo!e toggle Ctrl&.,
Auto in!ent onMoff Ctrl&.(
Restore line Ctrl&KF
0in! Ctrl&K0
0in! L replace Ctrl&KA
Repeat last fin! Ctrl&F
Control <ey prefi: Ctrl&"
A/ort operation Ctrl&=
Restore error message Ctrl&K8
Search .ptions+
' 'ac<#ar!s search
6 6lo/al search
n n = any num/er. Search n times
* Replace #ithout as<ing
= (gnore upperMlo#er case
8 Search for #hole #or!s only
S Soun!e:
"ro/lems
&&&&&&&&
A pro/lem is a te:t (ascii) file consisting of comments-
!efinitions- e3uations- an! statements. See the sample
files for e:amples of pro/lems.
Comments are not rea! /y the Mercury soler- an! sere
merely for clarity or !ocumentation. A comment starts
#ith a semicolon (N) an! runs until the en! of the line.
Alternatiely- comments can /e enclose! in curly /races (OP)
an! may occupy seeral lines.
2efinitions may /e either aria/le initiali;ations or function
!efinitions an! are characteri;e! /y the += assignment
operator. (f the left&han!&si!e is a name- it is ta<en to
/e a aria/le an! the right&han!&si!e is an e:pression #hich
ealuates to the initial alue for that aria/le. Such initial
alues can /e critical to the iteratie soler. (f the
left&han!&si!e is a function- as in S3uare(:) += :Q2- then the
right&han!&si!e seres as the !efinition for that function. (n
this case- S3uare is then a function of one argument- an!
S3uare(D) = 2D.
43uations hae a mathematical e:pression on each si!e of
an e3uals sign (=). An e3uation may also /e an ine3uality
constraint- in #hich case the e3uals sign is replace! /y
R- R=- S- or S=.
Statements are special !irecties to Mercury. ,hey are
instructions either to !o something other than sole the e3uations-
or to control the solution process. ,he allo#e! statements
are liste! /elo#- an! e:amples are in the pro/lem files.
*um/ers
&&&&&&&
,he range of num/ers is plus or minus 1e&50% to 1eH50%.
8hen using the e:act soler- a 3uestion mar< in!icates that the
ne:t !igit is uncertain. 0or e:ample- 1.25T in!icates a num/er
that might /e anything /et#een 1.22D an! 1.25D- as these are the
num/ers #hich roun! to 1.25.
(maginary num/ers are in!icate! #ith the suffi: i. 0or e:ample-
: = 2 H 5i is the comple: num/er #ith real part 2 an! imaginary
part 5.
"ercentages may /e e:presse! #ith the suffi: U. 0or e:ample- D.1U
is the same as .0D1.
4:amples
&&&&&&&&
A num/er of e:ample pro/lem files (9.4?A) are on the program
!is<. 4:amination of these is strongly recommen!e!- as these
!emonstrate most of Mercury)s features. See Chapter D.
5. Soling
Soling
&&&&&&&
Mercury features three soler technologies.
1. 2irect soler. ,his metho! consists of applying the rules of
high school alge/ra to manipulate the e3uations until solutions are
o/taine! for all the aria/les. 0or e:ample-
2 : H 5 y = $
7 : H D y = 17
can /e sole! !irectly /y rearranging the first e3uation to
y = ($ & 2 :)M5 an! su/stituting for y in the secon! e3uation.
See the F4A4F statement for user control oer #hat manipulations
are allo#e!.
2. (teratie soler. ,he most !ifficult systems of e3uations cannot
/e sole! analytically. ,he /est #e can !o is fin! some appro:imate
solution an! then apply some iteratie proce!ure to improe the
accuracy. (n such case it is a!isa/le to proi!e starting alues
for the aria/les.
5. 4:act soler. Mercury)s most innoatie feature is its a/ility
to use e:act metho!s in many cases.
Solution 8in!o#
&&&&&&&&&&&&&&&
8hen you apply Sole to a pro/lem file- the result is in the
Solution 8in!o#. ,he #in!o# is e!ita/le li<e the 4!it #in!o#-
an! can /e scrolle! an! ;oome! for easy ie#e!. ,he contents
can /e sae! in a report- if that is !esire!.
,he solution lists all of the aria/les- #ith their final
alues. =sually- the alue is @ust a num/er- !isplaye! to
ho#eer many !igits is specifie! in the 2(6(,S statements or
menu options.
(n some cases- a aria/le #as sole! !irectly an! sym/olically
in terms of other aria/les. (f such an e:pression #as use!
to calculate the alue- then the e:pression is sho#n.
(f no alue #as eer assigne! to the aria/le- then its alue
is gien as I=*240(*42I or I*. S.F=,(.*I. ,his occurs if the aria/le is
use! only as a !ummy aria/le- such as in a function !efinition
or as a aria/le of integration.
.ccasionally- a aria/le alue #ill /e gien as *A*- #hich
means *ot&A&*um/er. ,his means the aria/le coul! not /e compute!
accurately- an! only cru!e /oun!s #ere o/taine!. 0or e:ample-
a result I: = *A*(2.1-%.D)I in!icates that : may /e anything
/et#een 2.1 an! %.D. Sometimes such imprecise results represent
the /est that can /e sai! a/out an ill&pose! pro/lem- /ut sometimes
it represents a failure on the part of Mercury. Such results
shoul! /e use! #ith caution.
8hen a aria/le appears to hae the alue of a simple fraction-
it is note! in the solution. 0or e:ample- soling 5 : = 1 gies
: = 0.555555555555555 O 1 M 5 P. ,he fraction is re!uce!- meaning
that the numerator an! !enominator hae no nontriial factors. ,he
logic #hich recogni;es these fractions is particularly effectie
#ith the 4:act soler.
(n some cases- multiple roots for some aria/le are foun!. ,hese
are liste! after the list of aria/le alues.
(f the pro/lem file has a F(S, statement- then the list of the
a/scissae an! or!inate appear in the solution #in!o#.
,he solution al#ays has a list of resi!uals. ,hese are e:tremely
important for assessing the ali!ity an! accuracy of a solution.
A resi!ual is a 3uantity that shoul! /e ;ero if the e3uation #ere
sole! perfectly. (t is the a/solute alue of the !ifference
/et#een the left an! right han! si!es of the e3uation. 0or
e:ample- soling S(*(:) = .D might gie as a resi!ual
O 14&1C P S(*(:)&.D=0
,he num/er in /races- 14&1C- is the resi!ual. (t is the a/solute
alue of S(*(:)&.D using the sole! alue of :. Resi!uals of this
si;e are 3uite common- as internally Mercury calculates to 1C&!igit
precision- an! there is no machine representa/le : for #hich S(*(:)
is e:actly 0.D.
,he e3uation to the right of the /races is often the same as one of
the input e3uations- /ut sometimes Mercury rearranges some terms.
(t might /e a simple rearrangement as a/oe- /ut it may also the
result of sym/olically soling for one of aria/les an! su/stituting
it into other e3uations. (n any case- the e3uations sho#n #ith the
resi!uals are the actual e3uations use! /y the soler. Some control
of the sym/olic su/stitutions is allo#e! /y the F4A4F statement or
the Feel option un!er the Soler menu.
.ccasionally- you may see aria/les name ,1- ,2- ... ,hese are temporary
aria/les intro!uce! /y Mercury in the course of soling. Most
commonly- they are use! to enforce ine3uality constraints.
,he constraint
4E"(: H y) H y 9 ; R= 10
may /e replace! #ith
4E"(: H y) H y 9 ; H ,1 = 10
,1 S= 0
,he latter has some technical a!antages- een though there is an
e:tra e3uation an! an e:tra aria/le. Aalues for the e:tra aria/le
are not sho#n.
0inally- the solution en!s #ith the largest resi!ual- the metho!- an!
the processor. ,he largest resi!ual is simply the ma:imum of the
preious resi!uals. Assuming your pro/lem is not scale! too /a!ly-
then anything less than a/out 14&10 shoul! /e consi!ere! a success.
,he metho! is 4:act- 2irect- or (teratie. ,he e:act an! !irect
solers are more relia/le- an! Mercury tries them first. ,his logic
can /e altere! #ith the M4,1.2 statement. ,he processor is a $0$%-
$02$%- $05$%- or emulator.
Reports
&&&&&&&
A report is an unformatte! ascii file hol!ing the pro/lem an! the
solution. (t can /e sae! to !is< or printe! to <eep a recor! of
the solution. ,o sae to !is<- press 6o in the Report menu. (,his
menu item !oesn)t /ecome actie until you hae sole! a pro/lem.)
*ormally- the report has the same name as the pro/lem /ut #ith the
.4?A suffi: replace! #ith .R",. >ou can use other file names if you
please.
Another #ay to <eep a recor! of pro/lems an! solutions is to hae
a Fog file. Gust turn Capture .n in the Report menu. ,hen all
pro/lems an! solutions #ill /e appen!e! to a log file.
,he reports an! logs are or!inary ascii files- an! can /e printe!.
(f you hae a "ostscript printer- you #ill hae to select a mo!e
in #hich it #ill accept an ascii file- or loa! a utility for
printing ascii files. (*o such utility is proi!e! #ith Mercury.)
6raphics
&&&&&&&&
Mercury may /e use! to generate graphs. ,hese isually sho#
the /ehaior of functions. A graphics monitor is re3uire!.
,o get a graph- you must first !efine a function of one aria/le
in your pro/lem file. An e:ample from the file '=M">.4?A is
/umpy(:) += :Q2 S(*(:)
#hich !efines a function calle! I/umpyI. ,he I:I is a !ummy aria/le.
,he has alues func(0) = 0Q2 S(*(0)- func(1) = 1Q2 S(*(1)- etc. ,o
graph the function from &1D to H1D- use the statements+
"F., /umpy
6'.=*2S &1D- H1D
,o assimilate this information into the Mercury soler- you must !o
a Sole- een though you are not really soling for anything in this
case. .nce the pro/lem has /een sole!- the Aie#- "rint- an! 8rite
comman!s /ecome aaila/le un!er the 6raph menu. Select Aie# to see
the graph on the screen.
(f the graph loo<s a little ragge!- perhaps not enough points #ere
use! to plot it. ,he !efault is 50. (ncreasing it to 100 #ill gie
a much smoother graph. ,his can /e !one /y either a!!ing the
6".(*,S 100 statement to the pro/lem file an! soling it again- or
/y selecting 6"oints from the 6raph menu.
(t is also possi/le to hae IparametricI plots- #here /oth E an!
> are functions of some other in!epen!ent aria/le. Merely use
the EAE(S statement to specify the E function. ,he 6'.=*2S statement
then refers to the in!epen!ent aria/le.
,hese graphs can /e printe! on graphics printers. ,his inclu!es
most printers- e:cept for !aisy&#heel printers. Selecting "rint
#ill print the graph to #hateer printer has /een selecte! un!er
the .ptionsM"rinter menu.
>ou can also select 8rite- #hich prints the graph to a file. Such
a file can /e sent to a printer at a later time. ,he primary
usefulness of this feature is to e:port the graph in Fotus or
"ostscript format- #here it can /e importe! /y another program.
,he arious options in the 6raph menu are really the !efaults- an!
are oerri!!en /y the correspon!ing statements in the pro/lem file.
(f there are statements in the pro/lem file setting graph options-
then the menu may not accurately reflect the options use! to !ra#
the graph. *eertheless- once a graph has /een create!- these
options can /e mo!ifie! from the graph menu- an! then Aie# #ill
sho# the reise! graph.
7. 0eature Reference
,his chapter gies an alpha/etical loo<up for the Mercury /uilt&in
functions an! statements.
.perators
&&&&&&&&&
H a!!.
& su/tract.
9 multiply. a space may also /e use!.
M !ii!e.
Q po#er. 4g- 2Q5 = $.
V comple: con@ugate.
R less than.
R= less than.
S greater than.
S= greater than.
= e3uals.
+= aria/le initiali;ation or function !efinition.
+ e3uation separator.
U percent- eg- 1%U = .1%.
,hese hae the usual mathematical prece!ence rules- #ith Q haing
higher prece!ence than 9 an! M- #hich in turn are higher than H an! &.
Constants
&&&&&&&&&
,hese t#o constants are /uilt&in.
(*0 plus infinity
"( 5.171D92CD5D... >ou can also use the gree< letter /y typing Alt&".
0unctions
&&&&&&&&&
0unctions ta<e one or more arguments. ,he arguments are separate!
/y commas an! enclose! in parentheses. 1ere is a list of the
/uilt&in functions. A!!itional functions can /e !efine! #ithin
a pro/lem file.
A'S(:)
,he a/solute alue of :. >ou can also use W:W.
AC.S(:)
,he arc cosine of :- inerse to C.S(:).
,he result is in ra!ians.
AS(*(:)
,he arc sine of :- inerse to S(*(:).
,he result is in ra!ians.
A,A*(:)
,he arc tangent of :- inerse to ,A*(:).
,he result is in ra!ians.
'4SSG(n-:)
,he 'essel function of the first <in!.
,he first argument must /e an integer.
'4SS>(n-:)
,he 'essel function of the secon! <in!- sometimes
also calle! the 8e/er function.
,he first argument must /e an integer.
0or each n- it is a solution to 'essel)s
!ifferential e3uation #ith a singularity at ;ero.
'(*.M(n-<)
,he num/er of com/inations- choosing < of n items.
6ien /y nX M (<X (n&<)X).
C4(F(:)
,he smallest integer greater than or e3ual to :.
4g- C4(F(7.%) = D.
C.S(:)
,he cosine of :. ,he angle : is measure! in ra!ians.
C.S1(:)
,he hyper/olic cosine of :- (4E"(:) H 4E"(&:))M2.
C,4RM(rate-f-p)
,he num/er of perio!s re3uire! for a gien present
alue (p) to increase to a gien future alue (f)
if it gro#s at the gien interest rate. ,he function
is compati/le #ith KuattroMFotus.
24R(A(f(:)-:)
,he !eriatie of the function f(:) #ith respect to :.
(n most cases- the !eriatie is compute! sym/olically.
2F6AMMA(n-:)
,he n&th !eriatie of F.66AMMA- n = 0-1-2-...
4R0(:)
,he error function of :. (t is relate! to *C=M(:).
4R0(:) = 2 *C=M(: SKR,(2)) & 1
4g- 4R0(&(*0) = &1- 4R0(0) = 0- 4R0(H(*0) = H1.
4R0C(:)
,he complementary error function of :. See 4R0 an! *C=M.
4R0C(:) = 1 & 4R0(:)
4g- 4R0C(&(*0) = H1- 4R0C(0) = 0- 4R0C(H(*0) = &1.
4E"(:)
,he e:ponential function of :- eQ:- e = 2.%1$2$...
(t is the inerse to F*(:).
0AC,(:)
: factorial- sometimes #ritten :X.
4g- 0AC,(D) = D97959291 = 120.
0F..R(:)
,he largest integer less than or e3ual to :.
4g- 0F..R(7.%) = 7.
0RAC(:)
,he fractional part of :- : & 0F..R(:).
4g- 0RAC(7.%) = 0.%.
0A(pmt-rate-nper)
,he future alue of payments (pmt) at the en! of nper
perio!s- assuming the gien interest rate.
,he function is compati/le #ith KuattroMFotus.
0AAF(rate-nper-pmt-p)
6ies the future alue of a financial #ith a payment
pmt in each of nper perio!s- #ith the gien present
alue p- an! the gien interest rate.
,he function is compati/le #ith Kuattro "ro.
(t uses the financial mo!e specifie! /y the 0(*A*C4
statement to !etermine #hether the payments are at the
en! of the perio! (the !efault) or the /eginning.
1>".,(:-y)
,he polar coor!inate ra!ius correspon!ing to cartesian
coor!inates (:-y). 4g- 1>".,(5-7) = D.
(0".S(:-y-;)
6ies y if : S= 0- else it gies ;. 4g- (0".S(&5-C-%) = %.
(M(:)
,he imaginary part of :. ,he C.M"F4E option must /e in use.
(*,46(f(:)-:-a-/)
(*,46(f(:)-:-a-/) gies the numerical integral of f(:) #ith
respect to : from a to /. (mproper integrals are allo#e!- so a
can /e &(*0- / can /e H(*0- an! f can /e singular at a or /. ,he
appro:imation metho! is a!aptie an! nonlinear- an! often is
accurate to 1D !igits.
(*,461(f(:)-:-a-/)
(*,461 is similar to (*,46- /ut uses a simpler
strategy. (t is nona!aptie an! linear- an! may /e preferre! in
situations #here the result is suppose! to /e a smooth function
of a an! /. (*,461 is not nearly as accurate as (*,46 on
singular functions.
F(M(,(f(:)-:-a-(*0)
4stimates the limit of f(:)- as : goes from a to infinity.
(t uses a numerical metho! to compute the limit
of f(:)- as : goes from a to infinity. (t is an nonlinear
e:trapolation /ase! on a fairly small num/er of terms. ,he
limit is only relia/le if the function is #ell&/ehae!- an!
this feature shoul! /e use! #ith caution.
F*(:)
,he natural logarithm (/ase e) of :.
(t is the inerse to 4E"(:).
F.610(:)
,he logarithm of : to the /ase 10. ,he inerse is 10Q:.
F.66AMMA(:)
,he logarithm of the gamma function of :.
MAE(:-...)
,he ma:imum alue of the arguments.
4g- MAE(&10-D-%) = %.
M(*(:-...)
,he minimum alue of the arguments.
4g- M(*(&10-D-%) = &10.
*C=M(:)
,he normal cumulatie pro/a/ility function. *C=M(E) is the
pro/a/ility that a normal ran!om aria/le is /elo# its mean-
plus E stan!ar! !eiations.
4g- *C=M(&(*0) = 0- *C=M(0) = .D- *C=M(H(*0) = 1.
"A>M,(rate-nper-p-f)
6ies the present alue of a financial #ith a payment
pmt in each of nper perio!s- #ith the gien future
alue f- an! the gien interest rate.
,he function is compati/le #ith Kuattro "ro.
(t uses the financial mo!e specifie! /y the 0(*A*C4
statement to !etermine #hether the payments are at the
en! of the perio! (the !efault) or the /eginning.
"M,(p-rate-nper)
6ies the amount of the payments re3uire! in each of nper
perio!s to /alance the gien present alue p an! interest rate.
,he function is compati/le #ith KuattroMFotus.
".FAR(:-y)
,he polar coor!inate angle correspon!ing to cartesian (:-y).
Same as the KuattroMFotus YA,A*2(:-y) or the 0ortranMC atan2(y-:).
".F>(:-a0-a1-...-an)
,he polynomial a0 :Qn H ... H an.
4g- the polynomial 2 :Q5 & 5 : H % can also /e #ritten
".F>(:-2-0-&5-%). ,he chief a!antage of the latter is
notational conenience- an! so it can /e recogni;e! /y the
special polynomial soler.
".S(:)
,he positie part of :- ie- : if positie an! 0 if negatie.
"A(pmt-rate-nper)
6ies the present alue of nper payments of amount pmt-
using the gien interest rate.
,he function is compati/le #ith KuattroMFotus.
"AAF(rate-nper-pmt-f)
6ies the present alue of a financial #ith a payment
pmt in each of nper perio!s- #ith the gien future
alue f- an! the gien interest rate.
,he function is compati/le #ith Kuattro "ro.
(t uses the financial mo!e specifie! /y the 0(*A*C4
statement to !etermine #hether the payments are at the
en! of the perio! (the !efault) or the /eginning.
RA,4(f-p-nper)
6ies the interest rate re3uire! for the present
alue (p) to increase to the future alue (f) in
nper perio!s. ,he function is compati/le #ith KuattroMFotus.
R4(:)
,he real part of :. ,he C.M"F4E option must /e in use.
R4M(:-y)
,he remain!er of : !ii!e! /y y. 4g- R4M(20-5) = 2.
S6*(:)
6ies H1 if : S 0- &1 if : R 0.
S(*(:)
,he sine of :. ,he angle : is measure! in ra!ians.
S(*1(:)
,he hyper/olic sine of :- (4E"(:) & 4E"(&:))M2.
SKR,(:)
,he s3uare root of :.
S=M(f(:)-:-a-/)
,he sum of f from a to /-
f(a) H f(a H 1) H ... H f(/)
(f / = H(*0- the limit is estimate! as #ith the F(M(, function.
,A*(:)
,he tangent of :- S(*(:)MC.S(:).
,he angle : is measure! in ra!ians.
,A*1(:)
,he hyper/olic tangent of :- S(*1(:)MC.S1(:).
,4RM(pmt-rate-f)
,he num/er of perio!s re3uire! for a gien payment
stream (pmt) to increase to a gien future alue (f)
if it gro#s at the gien interest rate.
,he function is compati/le #ith KuattroMFotus.
Statements
&&&&&&&&&&
'.=*2S lo#er-upper
Specify /oun!s for the !omains of the aria/les. ,he !efault is
'.=*2S &(*0-(*0. Some soling time can /e sae! #ith the
statement '.=*2S &1eC-1eC if you are sure that all the releant
alues of your aria/les are less than a million.
CAS40.F2
(gnore case in names. .ther#ise- the aria/le : is !ifferent from E.
CAS4S,R(C,
=sing CAS4S,R(C,- #ith no args- causes /uilt&in functions-
constants- an! statements to /e upper case only. *ormally such
pre!efine! names are not case sensitie- so S(*(:) is the same as
sin(:). =sing this option re!uces the chance of a name conflict.
C.M"F4E
2eclares that all the aria/les are to /e consi!ere! comple:
aria/les. ,heir solutions hae a real an! imaginary part.
,hese /uilt&in functions may /e use! in a comple: pro/lem+
A'S- C.S- C.S1- 4E"- (M- F*- R4- S(*- S(*1- SKR,-
S=M- ,A*- ,A*1.
2(6(,S n
2isplay real num/ers to n !igits of precision. ,he !efault is 1$.
4"S(F.* :
0orces num/ers ery close to ;ero to /e !isplaye! as ;ero. =seful
for sho#ing comple: num/ers #hich are actually real. 2efault is 4"S(F.* 0.
4E,4R*AF function
2eclares the name to /e an e:ternally !efine! function- to /e
foun! on the !is< in the pro/lem !irectory #ith the file name
function.'(*. ,hese functions are #ritten in ,ur/o CHH or ,ASM-
an! compile! into .'(* files. See the samples.
0(*A*C4 type
6ies the type parameter to the Kuattro compati/le financial
functions- "AAF- "A>M,- an! 0AAF. ,he !efault is
type = 0- meaning that the payments are at the en! of the
perio!. =se type = 1 if the payments are at the /eginning.
6'.=*2S :left- :right
,he lo#er (:left) an! upper (:right) /oun!s for the E&a:is in a graph.
6".(*,S n
,he num/er of points (function alues) use! to !ra# a graph. ,he
!efault is 50. (f your graph is too ragge!- try increasing the
num/er of points.
6S,>F4 n
Allo#s graph cures to /e thic< (6S,>F4 0) or thin (6S,>F4 1).
(*CF=24 filename
,he file #ith the gien name is inserte! in the pro/lem at
that point. ,he filename must /e in 3uotes- as in (*CF=24 IC.*S,.(*CI. ,he
!efault file name e:tension is .(*C. (nclu!e files cannot /e neste!.
(,4RA,(.*S n
,he ma:imum num/er of iterations use! /y the iteratie soler. ,he
!efault is 100.
F464*2
"uts a legen! on the graph #hich i!entifies the function or functions.
Recommen!e! #hen plotting more than one function.
F4A4F n
Allo#s user control oer #hat sym/olic su/stitutions are ma!e.
0. no sym/olic soling
1. sole E = constant
2. sole E = constant e:pression
5. sole only #hen E occurs once- no other aria/les
7. sole only #hen E occurs once- other aria/les possi/le
D. sole linear e3uations
C. sole s3uares- using s3uare roots
%. sole 3ua!ratic e3uations
$. sole po#ers
,he !efault is F4A4F D- soling linear e3uations /ut not using
s3uare roots. Strictly spea<ing- the su/stitutions inoling
roots are not mathematically e3uialent- as the su/stitution must
ar/itrarily choose the positie or negatie s3uare root.
(t is sometimes prefera/le to use F4A4F 0 to turn off all the
sym/olic manipulations. ,hen Mercury effectiely #eights all
e3uations e3ually- an! the solution can /e consi!ere! the solution
to a least s3uares pro/lem.
F(S, function
Fist one or more functions in the solution #in!o#. Synta: is the
same as for "F.,.
MAE(M(J4 :
,he aria/le : is ma:imi;e!.
M4,1.2 n
Specify the soler metho!- #here n = 1- 2- or 5.
1. 4:act
2. 2irect
5. (teratie
,he !efault is M4,1.2 1- the e:act soler. (f it fails- the !irect
soler is trie!- an! it that also fails- the iteratie soler is
trie!.
M(*(M(J4 :
,he aria/le : is minimi;e!.
"4*AF,> n
,he alue of the penalty constant- an ar/itrary constant use!
to enforce constraints #hen the iteratie soler is use! to
ma:imi;e or minimi;e a aria/le. ,he !efault is 50. (ncreasing
it forces the appro:imate solutions closer to the constraint
manifol! at the cost of ma<ing the e:tremum har!er to fin!. Rea!ers
of ,he 1itchhi<er)s 6ui!e to the 6ala:y suggest using the alue 72.
"F., function
Creates a plot of the name! function. ,o ie# the graph- select
6raph from the main menu. ,o plot seeral functions on the same
graph- list the names separate! /y commas. ,he functions must /e
!efine! in the pro/lem file- an! /e functions of one aria/le. 0or
e:ample- to plot S(*(:) an! :Q5&: from &D to HD- use
f(:) += S(*(:)
g(:) += :Q5 & :
6'.=*2S &D- HD
"F., f- g
"R(*,
Same as "R(*,0(F4 IF",1I. (t prints the graph to the 2.S !eice F",1-
#hich is usually a printer on parallel port 1. ,he printing can /e
configure! to use another port from the .ptions menu- an! the result
sae! to a setup file.
"R(*,0(F4 filename
"rint the graph to the gien file- #hose name is in 3uotes.
,hese names hae special meaning.
I"AR1I parallel port 1
IS4R1I serial port 1
I"AR2I parallel port 2
IS4R2I serial port 2
IF",1I 2.S lpt1
IF",2I 2.S lpt2
IF",5I 2.S lpt5
IF",7I 2.S lpt7
I4",I ('M postscript port
"R.6R4SS :
Sho# the alue of the aria/le : in the "rogress /o:. Mercury
!oes not sole for all aria/les at once- so sometimes there
might /e no alue for the aria/le until the en! of the solution
process. (n such cases- the "rogress #in!o# sho#s I: = *A*I-
meaning I: is not a num/erI.
(f no "R.6R4SS statement is gien- the "rogress #in!o# sho#s the
alue of the IResi!ualI. ,his is the sum of s3uares of 3uantities
#hich shoul! /e ;ero- an! #hich Mercury is trying to minimi;e. ,he
smaller the alue- the more li<ely that Mercury is getting close to
the solution.
R4".R, filename
Allo#s naming a report file from #ithin a pro/lem.
R..,S(6* n
8hen soling an e3uation li<e :Q2 = a- Mercury may ta<e the s3uare
root of /oth si!es. ,his intro!uces t#o roots- HSKR,(a) an!
&SKR,(a)- an! Mercury !oes not <no# #hich to ta<e. >ou can specify
that you al#ays #ant the positie root #ith R..,S(6* H1- or
al#ays the negatie root #ith R..,S(6* &1. ,he !efault is
R..,S(6* 0- #hich leaes Mercury una/le to ealuate the s3uare
root.
SCR44* n
Choose a graphics screen. ,he !efault is SCR44* 0- #hich means
Mercury automatically tries to !etect your graphics har!#are.
Some graphics car!s an! mo!es cannot /e !etecte!- an! must /e
selecte! #ith a SCR44* statement.
0. Auto2etect
1. C6A
2. MC6A
5. 46A C7<
7. 46A 2DC<
D. 46A Mono
C. ('M $D17
%. 1ercules
$. A,, 700 line
9. A6A
10. 52%0 "C
11. ,an!y 1000
12. A,( Super A6A
15. .rchi! Super A6A
17. "ara!ise Super A6A
1D. Aega Super A6A
1C. Compro A'F4 (( Super A6A
1%. S.,A Super A6A
0rom the main menu- the graphics screen can /e selecte! #ith
6raphMMo!e. Such a selection is ineffectie after the first
graph is !ra#n- as the graphics has /een initiali;e! at that
point. 8e recommen! saing the !esire! graphics mo!e to the
M4RC=R>.C06 configuration file.
(*ote+ ,he !riers for some of these graphics mo!es is not
inclu!e! #ith some ersions of Mercury. Registere! customers
receie the full set of !riers.)
S1.8 :
Sho# specifie! aria/les in the specifie! or!er. .ther aria/les
are not !isplaye! in the solution #in!o#.
S.FA4 :
(nstructs Mercury to sole only for the aria/le :. 'y !efault- it
soles for all aria/les name! in the pro/lem.
(f S.FA4 is gien #ith a polynomial function name- as in
S.FA4 f
f(:) += ".F>(:-1-0-0-1)
then Mercury #ill apply a special polynomial root fin!er to get all
real an! comple: roots.
S,AR, :
,he starting alue for uninitiali;e! aria/les #hile using the
iteratie soler. ,he !efault is 1.00.
,4RMS n
,he most num/er of terms use! /y the (*,46 an! F(M(, functions. ,he
!efault is D0.
,(M4 n
,he ma:imum time allo#e! for solution. ,he !efault is 200 secon!s.
Same as MAE,(M4 in 4ure<a 1.0.
,(,F4 la/el
Argument is a 3uote! string- to /e use! as a graph title.
2efault is no title.
S=',(,F4 la/el
Argument is a 3uote! string- to /e use! as a graph su/title.
2efault is no title.
EFA'4F la/el
A la/el for the E&a:is of a graph. ,he la/el must /e in 3uotes.
>FA'4F la/el
A la/el for the >&a:is of a graph. ,he la/el must /e in 3uotes.
EAE(S function
Specifies that the E&a:is /e a function instea! of an in!epen!ent
aria/le. ,he function must /e !efine! else#here as a function
of one aria/le.
,he E a:is of the graph is a function #ith the gien name. =seful
for parametric plots. 4g- to graph (:-y) = (cos(t)-sin(t)) as
t goes from 0 to 29"(- use
:(t) += cos(t)
y(t) += sin(t)
6'.=*2S 0- 2 "(
EAE(S :
"F., y