Mathematica 9 is now available
Services & Resources / Wolfram Forums / MathGroup Archive
-----

MathGroup Archive 2010

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: problem with RandomInteger

  • To: mathgroup at smc.vnet.net
  • Subject: [mg112203] Re: problem with RandomInteger
  • From: Mark Adler <madler at alumni.caltech.edu>
  • Date: Sat, 4 Sep 2010 04:02:13 -0400 (EDT)
  • References: <i5qhhg$pkm$1@smc.vnet.net>

Interesting and alarming.

The behavior is the same with both Module and Block.  In either case if
any one of the local variables is removed from the local list, the
problem goes away.  All three need to be there.  So it does seem to be
related somehow to the use of Module, Block, and local (dynamic or
lexical) variables.  There is no problem with a non Module/Block
variant of the function, such as
Boole[RandomInteger[{1,3}]==RandomInteger[{1,3}]].

When trying to look at what was going on by putting Print's inside, the
problem went away!  (Heisenberg's Uncertainty Principle.)  Guessing
from that that it was a timing thing, I found that if I put
Pause[$TimeUnit]; anywhere in the function, the problem goes away. 
Some sort of race condition perhaps?

However when I use ParallelTable (which launches another kernel), the
problem goes away.

The behavior changes abruptly at length 250.  Up to length 249, all is
normal.  At length 250, the table is always a few zeros and ones
followed by all ones, or the table is simply all ones.

Beats me what's going on.

Mark


  • Prev by Date: Re: <Null> while building lists
  • Next by Date: Re: Strange colour highlighting in functions
  • Previous by thread: Re: problem with RandomInteger
  • Next by thread: Re: problem with RandomInteger