problem with rand ?

[Dmitri A. Sergatskov]

I found that random series returned by rand (from octave-forge) returns
a higher number of _identical_ values than I would expect:

octave:1> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); any(dss1==0)
ans = 1
octave:2> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans = [](1x0)
octave:3> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans = 60641
octave:4> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans = [](1x0)
octave:5> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans =

  26998  27604

octave:6> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans = 16034
octave:7> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans = 23032
octave:8> s1=rand(1,128*1024); ss1=sort(s1); dss1=diff(ss1); find(dss1==0)
ans = 121716

etc...

I would think that for random 2^17 numbers probability of two identical
numbers would be something like 2^17 / 2^54 (assuming ieee 64 bit double
representation)

Matlab does not show this behavior even for larger series:

>> s1=rand(1,256*1024); ss1=sort(s1); dss1=diff(ss1); any(dss1==0)

ans =

     0

>> s1=rand(1,1024*1024); ss1=sort(s1); dss1=diff(ss1); any(dss1==0)

ans =

     0

>> s1=rand(1,1024*1024); ss1=sort(s1); dss1=diff(ss1); any(dss1==0)

ans =

     0

Am I missing something here?

Sincerely,

Dmitri.



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------