[Nek5000-users] Optimal response for nek5000

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

Dear Adam,

I first need to understand if you are solving

  .steady Stokes

or

  .unsteady Stokes


They differ by what is on the diagonal of the Stokes operator.

For steady Stokes -

    A -D'
   -D  0

for unsteady Stokes we have

    H -D'
   -D  0

where H = 1/dt*B + A


The conditioning of the systems is quite different and hence
the solution strategy is also different.

For steady Stokes, I can confirm that PnPn-2 works well ---
I'm less certain of PnPn, which is generally designed as a
high Reynolds number solver (but it also works fine at low
Re, I'm just not certain of Re=0 for PnPn).

Paul





----- Original Message -----
From: nek5000-users at lists.mcs.anl.gov
To: nek5000-users at lists.mcs.anl.gov
Sent: Thursday, September 5, 2013 9:11:19 AM
Subject: Re: [Nek5000-users] Optimal response for nek5000

Dear Paul

I switch between time-dependent NS and time-dependent Stokes (both for 
Euler time-stepping). I missed the fact that prsolvv gives steady state 
solution. I guess it will be PnPn-2, but not sure. Is there a crucial 
difference between PnPn and PnPn-2 when it comes to switching between NS 
and Stokes?
regards
Adam

On 05/09/13 15:46, nek5000-users at lists.mcs.anl.gov wrote:
> Adam,
>
> Just to clarify,
>
> are you switching between time-dependent NS and time-dependent Stokes.
>
> Or, are you switching between time-dependent NS and steady Stokes ?
>
> Also, are you using PnPn or PnPn-2 ?
>
> Thanks,
>
> Paul
>
> ----- Original Message -----
> From: nek5000-users at lists.mcs.anl.gov
> To: nek5000-users at lists.mcs.anl.gov
> Sent: Thursday, September 5, 2013 7:49:01 AM
> Subject: [Nek5000-users] Optimal response for nek5000
>
> Dear neks
>
> I'm going to implement optimal response to forcing in nekton and I have
> question related to pressure solver. During calculations I have to
> continuously switch between time dependent N-S and Stokes problem. I was
> going to do something similar to prsolvv. However, in nek_init the
> choice of pressure solver depends on IFTRAN suggesting different solver
> for Stokes problem, but prslovv seem to be called without
> reinitialisation of pressure solver. So can I use the same pressure
> solver (say itr) for both full N-S and Stokes?
> Regards
> Adam
> _______________________________________________
> 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

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users