[FLASH-USERS] Big Problem??

[Seyit Hocuk]

Klaus,

Actually, I think that number might be even too big. It causes the
second order derivative refinement not to work. If the number i reduced
to 1E-99, then the second order derivative refinement works.

Seyit



Klaus Weide wrote:
> On Mon, 18 Aug 2008, Seyit Hocuk wrote:
>
>> Anyway, after these, there were no other errors showing up when 
>> debugging, so the code is a-ok also for our own setup. No 
>> uninitialized parameters. We downloaded the latest compilers of ifort 
>> (10.1.012 / 015 / 017) and tried them each. The difference between 
>> ifort 8.0 and 10.1 seems huge even on the same machine! Refinement is 
>> very different even for normal jeans setup. The differences became 
>> very small when we reduced the boxsize just to check. Normally I use 
>> boxsize of 8.0E19, when we tried 8.0E9 (mind you its 10 orders of 
>> magnitude less), the differences were minimal.
>
> Seyit,
>
> I note that setups/jeans/ref_marking.F90 has a hardwirded "small 
> number" here:
>
>                      do kk = 1, ndim2
>                         num = num + delu2(kk)**2
>                         denom = denom + (delu3(kk) + &
>      &                       (epsil*delu4(kk)+1.e-20))**2
>                      end do
>
> Maybe that has something to do with the dependence of your results on
> box sizes?  You may want to try if changing that number to something 
> many orders of magnitude different changes your findings.
>
> Klaus