ApPENDIX F
Q,erf & erfcFunctions
F.l TheGFunction
Computation of probabilities that involve a Gaussianprocessrequire finding the areaunder the
tail of the Gaussian(normal) probability densityfunction asshownin Figure F.l.
trL J$ x
Figure F.l Gaussianprobabilitydensityfunction.Shadedareais Pr(x> xs)
Gaussianrandomvariable.
for a
Appendix F . Q, ert & erfc Functions
Figure F.l illustrates the probability that a Gaussian random variable .r exceeds x0,
Pr(x>.rs), whichis evaluatedas
' | 1 - r x - m t : /  2 o 2 t *
Prx2xo)=
l-et 6 "J/.TE
(F.1)
(F.2)
(F.3)
(F.4)
(F.5)
(F.6)
(F.7)
r0
The Gaussianprobability densityfunction in Equation(F.1)cannotbe integratedin closedform'
Any Gaussianprobability density function may be rewritten through useof the substitution
to yield
x-my = --o
r,(r>ry)= j
f-^'-r"o'( x o - m 
 o l
where the kernel of the integral on the right-handside of Equation (F.3) is the normalized
Gaussianprobability densityfunction with meanof 0 and standarddeviationof 1. Evaluation
of the integralin Equation(F.3)is designatedasthe Q-function,which is definedas
o(:)= 11,'"'a,JrJ2n
HenceEquations(F.1)or (F.3)canbe evaluatedas
/ x^-m (ro-.!
= o(z.lrt-J =g(.-o
) -
The o-function is boundedby two analyticalexpressionsasfollows:
/ r  | - l t 1 - . 2 , )
I r _1l*"-"' <QQ)3-i-e'
-
 ,'z)7J2n zJTn
For valuesof z greater3.0,both of theseboundsclosely approximateQ(z) .
Two important propertiesof Qk) are
Qer)= r-QQ)
O0 =,
A graphof Qk) versus{ is givenin FigureF2'
A tabulationof theQ-functionfor variousvaluesof z is givenin TableF.l.
t-,
i
l:
*.
i:
::
{:
t;
i,:t
The O+unction
TableF.1 TabulationoftheGfunction
aQl aQ)
0.50000 2.0 0.02275
0.1 0.46017 2.1 0.01786
0.42074 2.2 0.0r390
0.38209 2.3 0.01072
0.4 0.34458 0.00820
0.30854 2.5 0.00621
0.27425 2.6 0.00466
0.24t96 2.7 0.00347
0.21I86 2.8 0.00256
0.18406 2.9 0.00187
0.15866 3.0 0.00135
0.13567 J . l 0.00097
0.11507 0.00069
l . J 0.09680 J . J 0.00048
0.08076 3 . +
0.06681 3.5 0.00023
0.05480 3.6 0.00016
0.04457 3.7
0.03593 3.8
0.02872 3.9
0.0
0.2
0.3
0.5
0.6
0.7
0.8
0.9
1.0
l . l
) . 2r.2
0.000341.4
1.5
t . o
t t 0.00011
1 , 8 0.00007
1.9
i:
i
f'
v.
F
L
* t -
ta*,*.- ..
'0.00005
0 0.5 1.0 1.5 2.0 2.5
FigureF.2 PlotoftheGlunction.
erfc(z)= t-erf(z)
F.2 TheerfandertcFunctions
The enor function (erf; is defined as
n '
er.f(a)= 4lr-" a,
J"t,
andthe complementaryerror function (erfc) is defined as
" i
- 2
erfcz) = !-le-^ dr
Jnr,
The erfc function is relatedto the et'function by
Appendix F . Q, ert & ertc Functions
(F.10)
f
,.'
L
f.
t'
t
-l
t 0
10-'
(F.8)
(F.9)
The erfand elc Function
The Q-function is relatedto the erf and erfc functions by
l T / z  ) I / - 
Qc)= ;lt
-er;,ll = ;,,J,li). 4 2 . , - . 4 2 ,
erfc(z)= 2Q(J1z)
erf(z)= 1-2Q0J2z)
The relationshipsin Equations(F.11)-(F.13)arewidely usedin eror probabilitycomputa-
tions.TableF.2displaysvaluesfor the ef function.
TableF.2 TabulationoftheErrorFunctionerf(z)
z ertQl z ertQl
0.1 0.1t246 1.6 0.97635
0.2 0.22270 l ' 1 0.98379
0.32863 1 . 8 0.98909
0.4 0.42839 1 . 9 0.992'79
0.-5 0.52049 2.0 0.99532
0.6 0.60385 2 . 1 0.99702
0.67780 2.2 0.99814
0.8 0.74210 L . 3 0.99885
0.79691 . A
0.99931
0.842'70 2.5 0.99959
1 . 1 0.88021 2.6 0.99976
0.91031 2.1 0.99987
l . J 0.93401 2.8 0.99993
1 . 4 0.95228 2.9 0.99996
(F.11)
(F.12)
/ E I ? . 
0.7
0.9
i.0
t.2
,.
:.,
t
Tl:
a.
I
?,1:.
:i:-
r+:
it.
w
1 . 5 0.9661r 3.0 0.99998

Q and erf functions

  • 1.
    ApPENDIX F Q,erf &erfcFunctions F.l TheGFunction Computation of probabilities that involve a Gaussianprocessrequire finding the areaunder the tail of the Gaussian(normal) probability densityfunction asshownin Figure F.l. trL J$ x Figure F.l Gaussianprobabilitydensityfunction.Shadedareais Pr(x> xs) Gaussianrandomvariable. for a
  • 2.
    Appendix F .Q, ert & erfc Functions Figure F.l illustrates the probability that a Gaussian random variable .r exceeds x0, Pr(x>.rs), whichis evaluatedas ' | 1 - r x - m t : / 2 o 2 t * Prx2xo)= l-et 6 "J/.TE (F.1) (F.2) (F.3) (F.4) (F.5) (F.6) (F.7) r0 The Gaussianprobability densityfunction in Equation(F.1)cannotbe integratedin closedform' Any Gaussianprobability density function may be rewritten through useof the substitution to yield x-my = --o r,(r>ry)= j f-^'-r"o'( x o - m o l where the kernel of the integral on the right-handside of Equation (F.3) is the normalized Gaussianprobability densityfunction with meanof 0 and standarddeviationof 1. Evaluation of the integralin Equation(F.3)is designatedasthe Q-function,which is definedas o(:)= 11,'"'a,JrJ2n HenceEquations(F.1)or (F.3)canbe evaluatedas / x^-m (ro-.! = o(z.lrt-J =g(.-o ) - The o-function is boundedby two analyticalexpressionsasfollows: / r | - l t 1 - . 2 , ) I r _1l*"-"' <QQ)3-i-e' - ,'z)7J2n zJTn For valuesof z greater3.0,both of theseboundsclosely approximateQ(z) . Two important propertiesof Qk) are Qer)= r-QQ) O0 =, A graphof Qk) versus{ is givenin FigureF2' A tabulationof theQ-functionfor variousvaluesof z is givenin TableF.l. t-, i l: *. i: :: {: t; i,:t
  • 3.
    The O+unction TableF.1 TabulationoftheGfunction aQlaQ) 0.50000 2.0 0.02275 0.1 0.46017 2.1 0.01786 0.42074 2.2 0.0r390 0.38209 2.3 0.01072 0.4 0.34458 0.00820 0.30854 2.5 0.00621 0.27425 2.6 0.00466 0.24t96 2.7 0.00347 0.21I86 2.8 0.00256 0.18406 2.9 0.00187 0.15866 3.0 0.00135 0.13567 J . l 0.00097 0.11507 0.00069 l . J 0.09680 J . J 0.00048 0.08076 3 . + 0.06681 3.5 0.00023 0.05480 3.6 0.00016 0.04457 3.7 0.03593 3.8 0.02872 3.9 0.0 0.2 0.3 0.5 0.6 0.7 0.8 0.9 1.0 l . l ) . 2r.2 0.000341.4 1.5 t . o t t 0.00011 1 , 8 0.00007 1.9 i: i f' v. F L * t - ta*,*.- .. '0.00005
  • 4.
    0 0.5 1.01.5 2.0 2.5 FigureF.2 PlotoftheGlunction. erfc(z)= t-erf(z) F.2 TheerfandertcFunctions The enor function (erf; is defined as n ' er.f(a)= 4lr-" a, J"t, andthe complementaryerror function (erfc) is defined as " i - 2 erfcz) = !-le-^ dr Jnr, The erfc function is relatedto the et'function by Appendix F . Q, ert & ertc Functions (F.10) f ,.' L f. t' t -l t 0 10-' (F.8) (F.9)
  • 5.
    The erfand elcFunction The Q-function is relatedto the erf and erfc functions by l T / z ) I / - Qc)= ;lt -er;,ll = ;,,J,li). 4 2 . , - . 4 2 , erfc(z)= 2Q(J1z) erf(z)= 1-2Q0J2z) The relationshipsin Equations(F.11)-(F.13)arewidely usedin eror probabilitycomputa- tions.TableF.2displaysvaluesfor the ef function. TableF.2 TabulationoftheErrorFunctionerf(z) z ertQl z ertQl 0.1 0.1t246 1.6 0.97635 0.2 0.22270 l ' 1 0.98379 0.32863 1 . 8 0.98909 0.4 0.42839 1 . 9 0.992'79 0.-5 0.52049 2.0 0.99532 0.6 0.60385 2 . 1 0.99702 0.67780 2.2 0.99814 0.8 0.74210 L . 3 0.99885 0.79691 . A 0.99931 0.842'70 2.5 0.99959 1 . 1 0.88021 2.6 0.99976 0.91031 2.1 0.99987 l . J 0.93401 2.8 0.99993 1 . 4 0.95228 2.9 0.99996 (F.11) (F.12) / E I ? . 0.7 0.9 i.0 t.2 ,. :., t Tl: a. I ?,1:. :i:- r+: it. w 1 . 5 0.9661r 3.0 0.99998