[Nek5000-users] Residuals

[nek5000-users at lists.mcs.anl.gov]

Hi JC and Paul,

there is a typo -- it should be

n = nx1*ny1*nz1*nelv

Best,
Aleks


On Tue, 15 Jun 2010, nek5000-users at lists.mcs.anl.gov wrote:

>
> PS - I forgot to divide by dt in the routine below.
>
>     To do this,
>
>
>     dti = 1./dt
>     call opcmult(uo,vo,wo,dti)
>
>     after opcopy
>
>
> On Tue, 15 Jun 2010, nek5000-users at lists.mcs.anl.gov wrote:
>
>> 
>> JC,
>> 
>> Given that (v,u_t) = (v,r), where r is the residual, I would
>> assume that you can get the desired residual estimate by looking
>> at u_t := du/dt ~= || u^n - u^n-1 || / dt ?
>> 
>> Typically, I would do this in userchk, creating a set of variables
>> in a common block, e.g.
>>
>>      parameter (lt=lx1*ly1*lz1*lelt)
>>      common /vold/ uo(lt),vo(lt),wo(lt)
>>      real linf,l2,h1,semi
>>
>>      n = nx1*ny1*nz1*nz1*nelv
>>
>>      if (istep.gt.0) then
>>         call opsub2 (uo,vo,wo,vx,vy,vz)
>>         call normvc (h1,semi,l2,linf,uo,vo,wo)
>>         if (nid.eq.0) write(6,1) istep,time,h1,semi,l2,linf
>>    1    format(i8,1p5e13.6,' resnorm')
>>      endif
>>      call opcopy (uo,vo,wo,vx,vy,vz) ! Save current velocity
>> 
>> 
>> Then "grep resnorm logfile > t.t"  would put the time history
>> of this quantity into the file t.t, assuming your stdout is
>> directed to logfile.
>> 
>> Paul
>> 
>> 
>> On Tue, 15 Jun 2010, nek5000-users at lists.mcs.anl.gov wrote:
>> 
>>> Hi Nek's,
>>> 
>>> I have a question concerning residuals.
>>> I wanna run a calculation that yields to a stationnary flow. In order to 
>>> be sure that my calculation has converged, I'd like to see the residuals.
>>> How/where could I output them ?
>>> 
>>> Regards,
>>> JC
>>> _______________________________________________
>>> Nek5000-users mailing list
>>> Nek5000-users at lists.mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>> 
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>