APPENDIX D
Q, erf& erfe Functions
D.1 The GFunction
Computation of probabilities that involve a Gaussian process require find-
ing the area under the tail of the Gaussian (normal) probability density function
as shown in Figure Dl.
— (x — rn)
2&
p(x)= I
e
a
m x
Figure D.1
Gaussian probability densityfunction. Shaded area is Pr (x x0) for a Gaussian random
variable./
593
594 App.D'Q,erf&erfcFunctions
Figure D. 1 illustrates the probability that a Gaussian random variable x
exceeds x0, Pr (x x0) , which is evaluateclas
( —(x—rnh/ (202)
Pr(x x0) = I
e dx (D.1)
xo
The Gaussian probability density function in equation (D.1) cannot be integrated
in closed form.
Any Gaussian probability density function may be rewritten through use of
the substitution
= x—nt
(P.2)
to yield
(
) .f pe i —y2/2
dy (113)
(xo — m
where the kernel of the integral on the right hand side of equation (D.3) is the
normalized Gaussian probability density function with mean of 0 and standard
deviation of 1. Evaluation of the integral in equation (D.3) is designated as the
Q-function, which is defined as
Q(z) =
tl —y/2
dy (D.4)
z
Hence equations (P.1) or (D.3) can be evaluated as
= Q(XO 7?2)
= Q(z) (D.5)
The Q-function is bounded by two analytical expressions as follows:
( ii
1—91
I
z
e
—z/2 I
i
1
z.r2,t
For values of z greater 3.0, both of these bounds closely approximate Q (z).
Two important properties of Q (z) are
Q(—z) = 1 —Q(z) (D.6)
Q(0) = (D.7)
A graph of Q (z) versus z is given in Figure D.2.
A tabulation of the [Link] for various values of z is given in Table D.1.
App. D .0,8(16 erIc Functions 595
100
to_I
'os
to-'
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5
Figure D.2
Plot of the Q-function.
0.2 The ml and erfc functions
The error function (en) is defined as
erf(z) = (D.8)
and the complementary error function (erfe) is defined as
er/'c(z) = 27je *
cix (D.9)
t
The erfc function is related to the enj' function by
erfc(z) = l—erf(z) (D.10)
The Q-function is related to the erf and erfe functions by
596 App.D.Q,ert&erfcFunctions
Table D.1 Tabulation of the Gfunctlon
z Q(z) z Q(z)
0.0 0.50000 2.0 0.02275
0.1 0.46017 2.1 0.01786
0.2 0.42074 2.2 0.01390
0.3 0.38209 2.3 0.01072
0.4 0.34458 2.4 0.00820
0.5 0.30854 2.5 0.00621
0.6 0.27425 2.6 0.00466
0.7 0.24196 2.7 0.00347
0.8 0.21186 2.8 0.00256
0.9 0.18406 2.9 0.00187
1.0 0.15866 3.0 0.00135
1.1 0.13567 3.1 0.00097
1.2 0.11507 3.2 0.00069
1.3 0.09680 3.3 0.00048
1.4 0.08076 3.4 0.00034
1.5 0.06681 3.5 0.00023
1.6 0.05480 3.6 0.00016
1.7 0.04457 3.7 0.00011
1.8 0.03593 3.8 0.00007
1.9 0.02872 3.9 0.00005
Q(z) = = (D.11)
erfc (z) = 2Q (D.12)
erf(z) = I (D.13)
The relationships in equations (D.11) - (D.13) are widely used in error prob-
ability computations. Table D.2 displays values for the eTf function.
App. D .Q en & eric Functions 597
C Table D.2 Tabulation of the Error Function erf(z)
z erf(z) z erf(z)
0.1 0.11246 1.6 0.97635
0.2 0.22270 1.7 0.98379
0.3 0.32863 1.8 0.98909
OA 0.42839 1.9 0.99279
0.5 0.52049 2.0 0.99532
0.6 0.60385 2.1 0.99702
0.7 0.67780 2.2 0.99814
0.8 - 0.74210 2.3 0.99885
0.9 0.79691 2.4 0.99931
1.0 0.84270 2.5 0.99959
1.1 0.88021 2.6 0.99976
1.2 0.91031 2.7 0.99987 •
1.3 0.93401 2.8 0.99993
1.4 0.95228 2.9 0.99996
1.5 0.96611 3.0 0.99998