[Nek5000-users] Full restart with a different time step

Tue Nov 17 17:29:40 CST 2015

Hi,

when you change the time step during a simulation, then there is no 
problem; the EXT/BDF coefficients are adjusted accordingly, so you will 
keep the accuracy/order even with varying time step.

The reason why you cannot do that during restart (i.e. during the first 
4 steps) is because the stored fields are at a given time (at which they 
were stored, yielding a certain dt), so you cannot assign them a 
different time or dt. The solution in memory is then overwritten with 
the data loaded from disk, until all lag arrays are correctly filled. 
But as soon as the simulation is not reading from the disk any longer, 
then you can go ahead and change the time step.

Best,
PHilipp

On 2015-11-18 00:24, nek5000-users at lists.mcs.anl.gov wrote:
> Hi Adam,
>
> Thank you very much for your comments.
>
> I actually thought about loading the 4 restart files in nek and do
> interpolation between two time steps. But I agree with you that I
> would introduce some error to the flow fields for restarting. In my
> case, I want to keep the data of the old simulation and continue from
> the current checkpoints, so I don't want to ruin the accuracy of the
> flow fields.
>
> Like you said, changing \dt after restarting from the current flow
> fields maight be the simplest solution. However, I have one concern
> about this. When we use 3rd order accuracy in time, the solver would
> use the history information from the last few time steps and that's
> why we need to restart the simulation with 4 continuous flow fields.
> If I change the time step size in the middle of the simulation, how
> much would it affect the solving process? What do you think?
>
> Best regards, Tony
>
> ________________________________________ From:
> nek5000-users-bounces at lists.mcs.anl.gov
> <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of
> nek5000-users-request at lists.mcs.anl.gov
> <nek5000-users-request at lists.mcs.anl.gov> Sent: 17 November 2015
> 18:00 To: nek5000-users at lists.mcs.anl.gov Subject: Nek5000-users
> Digest, Vol 81, Issue 15
>
> Send Nek5000-users mailing list submissions to
> nek5000-users at lists.mcs.anl.gov
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users or, via
> email, send a message with subject or body 'help' to
> nek5000-users-request at lists.mcs.anl.gov
>
> You can reach the person managing the list at
> nek5000-users-owner at lists.mcs.anl.gov
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Nek5000-users digest..."
>
>
> Today's Topics:
>
> 1. Full restart with a different time step
> (nek5000-users at lists.mcs.anl.gov) 2. Re: Full restart with a
> different time step (nek5000-users at lists.mcs.anl.gov)
>
>
> ----------------------------------------------------------------------
>
>  Message: 1 Date: Mon, 16 Nov 2015 19:19:23 +0000 From:
> nek5000-users at lists.mcs.anl.gov To:
> "nek5000-users at lists.mcs.anl.gov" <nek5000-users at lists.mcs.anl.gov>
> Subject: [Nek5000-users] Full restart with a different time step
> Message-ID:
> <mailman.10777.1447701586.24727.nek5000-users at lists.mcs.anl.gov>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Dear Neks,
>
>
> I use full_restart option for my current simulations. One simulation
> crashed after restarting for some time with the Courant number going
> crazy. I suppose I will need to reduce the time step size.
>
>
> Basically, I want to continue my simulation using full_restart option
> but with a smaller time step size. Somehow I remember I saw one post
> saying one shouldn't change the time step size when using the
> full_restart option. I'm just thinking of a way to work around the
> problem.
>
>
> So does anyone have ideas on how I can interpolate the flow fields
> between two time steps or any other ways to fully restart a
> simulation with a different time step size?
>
>
> Thank you very much in advance for any suggestions or comments.
>
>
> Best regards,
>
> Tony -------------- next part -------------- An HTML attachment was
> scrubbed... URL:
> <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151116/ff31eb4f/attachment-0001.html>
>
>  ------------------------------
>
> Message: 2 Date: Tue, 17 Nov 2015 09:27:47 +0100 From:
> nek5000-users at lists.mcs.anl.gov To: nek5000-users at lists.mcs.anl.gov
> Subject: Re: [Nek5000-users] Full restart with a different time step
> Message-ID:
> <mailman.10819.1447748894.24727.nek5000-users at lists.mcs.anl.gov>
> Content-Type: text/plain; charset="windows-1252"; Format="flowed"
>
> Hi
>
> For full_restart you should have four velocity fields and you could
> use nek5000 as a preprocessing tool. You can simply load all 4 fields
> in userchk, perform some kind of interpolation in time and finally
> write down new restart files. The question is how to deal with
> pressure. One solution would be simply interpolate pressure saved in
> the files. The other would be to run modified version of pressure
> solver with uzawa turned on to get pressure corresponding to
> interpolated velocity field. We've played with uzawa for pn-pn-2 and
> we use our version of full_restart, but we haven't done this kind of
> modified restart, so I don't know what is better. On the other hand
> you could restart nek with your current checkpoints and dt
> corresponding to the old simulation, and later (after 4th step)
> gradually decrease time step. I guess this would be the simplest and
> probably best solution. If time resolution in old simulation was
> insufficient, your restart files do not give correct evolution of the
> field and interpolation in time would not remove it. Anyhow you
> introduce some initial error to simulation. In this case I probably
> would simply play with dt keeping it constant during first 4 steps
> and gradually decreasing later. Regards Adam
>
> On 2015-11-16 20:19, nek5000-users at lists.mcs.anl.gov wrote:
>>
>> Dear Neks,
>>
>>
>> I use full_restart option for my current simulations. One
>> simulation crashed after restarting for some time with the Courant
>> number going crazy. I suppose I will need to reduce the time step
>> size.
>>
>>
>> Basically, I want to continue my simulation using full_restart
>> option but with a smaller time step size. Somehow I remember I saw
>> one post saying one shouldn't change the time step size when using
>> the full_restart option. I'm just thinking of a way to work around
>> the problem.
>>
>>
>> So does anyone have ideas on how I can interpolate the flow fields
>> between two time steps or any other ways to fully restart a
>> simulation with a different time step size?
>>
>>
>> Thank you very much in advance for any suggestions or comments.
>>
>>
>> Best regards,
>>
>> Tony
>>
>>
>>
>> _______________________________________________ Nek5000-users
>> mailing list Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>
> -------------- next part -------------- An HTML attachment was
> scrubbed... URL:
> <http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151117/7d8c9274/attachment-0001.html>
>
>  ------------------------------
>
> _______________________________________________ Nek5000-users mailing
> list Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>
>
> End of Nek5000-users Digest, Vol 81, Issue 15
> *********************************************
> _______________________________________________ Nek5000-users mailing
> list Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>