(j3.2006) (SC22WG5.4124) [ukfortran] Parallel random numbers

[N.M. Maclaren]

On Oct 26 2009, Aleksandar Donev wrote:
>
>>    2) Are you really saying that the Fortran standard should REQUIRE a
>> known, deficient approach, of a nature that every expert in this area
>> regards as unacceptable?  That's not the Fortran I know.
>
>The problem is that your response only says what is wrong with the 
>independent sequence model, it does not say anything about why your 
>proposed approach (processor dependent) is good. If we really want to do 
>something good, we should do it, rather than rely on some compiler 
>intern to get it right. At least the independent sequence model is clear 
>and simple.

No, that's not right.  I did NOT say that the independent sequence model
has anything wrong with it; I said that what was being proposed specified
an ASSOCIATED sequence model.  Yes, it's clear and simple - and VERY bad.

The problem about specifying that this has to be done well is that that
requires some EXTREMELY specialist skills.  I don't know how many people
in the world who are even second-rank experts, but it wouldn't surprise
me if it was still only in the hundreds or perhaps even dozens :-(

I will try to explain in a paper.  I may not succeed.

regards,
Nick.