Author 
Comment/Response 
Bill Simpson

04/19/13 1:19pm
In Response To 'Re: Re: Re: Re: Re: Variance freezes for certai...'  His problem can't be reproduced in several other versions.
But since it works just fine for him with a vector of 9 random numbers and doesn't work with a vector of 10 random numbers and there are no obvious, or even subtle, mistakes in his input...
then it seems really unlikely that "resetting" Mathematica initialization is going to be the cause or the fix, not unless initialization does some pretty silly things.
Similarly it seems really unlikely that libraries could be broken in any way that would work just fine for vectors of length 9 and fail for length 10.
None of the ideas that have been suggested seem like they could explain this. If it worked for short vectors and failed for vectors of 10^6 or 10^9 elements, maybe that I could imagine might be running over some limit, but not 9 versus 10.
Perhaps he could do half a dozen different runs of 9 elements, post all the vectors and the variance result. Then we could stuff his vectors into our variance and verify we get the same result. I can't imagine this would do anything other than verify we all get the same results.
Perhaps he could try other statistical vector functions, like Mean or StandardDeviation and see if they also fail on 9 versus 10.
Perhaps he could try other vector non statistical functions and see if this is a problem with length 10 vectors in general or just Variance. Does {1,2,3,4,5,6,7,8,9,10}+{1,2,3,4,5,6,7,8,9,10} work? Does {1,2,3,4,5,6,7,8,9,10}*{1,2,3,4,5,6,7,8,9,10} work? The idea is to start poking around and see if there is a larger set of things that don't work. What ever the answer to that might be could possibly begin to hint at a pattern and maybe the pattern will provide a clue.
URL: , 
