From octave-sources-request at bevo dot che dot wisc dot edu  Thu Apr  9 07:33:05 1998
Subject: sanity test for gammai
From: Jim Van Zandt <jrv at mbunix dot mitre dot org>
To: octave-sources at bevo dot che dot wisc dot edu
Cc: jrv at vanzandt dot mv dot com
Date: Thu, 09 Apr 1998 08:33:03 -0400

John -

Octave 2.0.5 finds:

    octave:46> sqrt(pi)*erf(1)
    ans =  1.49364826562485e+00
    octave:47> gammai(.5,1)*gamma(.5)
    ans =  1.49364826562485e+00
    octave:48> gammainc(1,.5)*gamma(.5)
    ans =  1.49364826562485e+00

which are correct.  (Unfortunately, as has been pointed out, 2.0.11
results are not correct.)  

I would like to propose the following for
test/octave.test/arith/gammai-1.m:

    # reference: Abramowitz & Stegun section
    # 6.5, where P(a,x) is the Octave
    # function gammai(a,x)
    x = [sqrt(pi)*erf(1), gammai(.5,1)*gamma(.5), gammainc(1,.5)*gamma(.5)];
    v = [1.49364826562485, 1.49364826562485, 1.49364826562485];
    all (abs (x - v) < sqrt(eps))

Unfortunately, octave 2.0.5 chokes on this (infinite loop or
something).  I have not been able to determine why.

                                 - Jim Van Zandt