[Users] Unexpected Hamiltonian constraint violation from refinement when using CCZ4

[CHEN, Kenneth]

Hi, Peter,

Thanks for your suggestions, which are quite helpful.

I was using the independent "Dissipation" thorn
Dissipation::epsdis = 0.1
Dissipation::order = 5
Dissipation::vars                       = "
        ML_CCZ4::ML_log_confac
        ML_CCZ4::ML_metric
        ML_CCZ4::ML_trace_curv
        ML_CCZ4::ML_curv
        ML_CCZ4::ML_Gamma
        ML_CCZ4::ML_lapse
        ML_CCZ4::ML_shift
        ML_CCZ4::ML_dtlapse
        ML_CCZ4::ML_dtshift
"
for RK4, which turned out to be not enough dissipation for damping out the high frequency oscillation of the constrain violation.

Then I turned on the dissipation of ML_CCZ4::epsdiss = 0.1, It worked, no more large violation near refinement boundaries.

However, not very sure about how these two "epsdis" work differently.

BR
KZ





________________________________
From: Peter Diener <diener at cct.lsu.edu>
Sent: Thursday, July 8, 2021 10:47 PM
To: CHEN, Kenneth <kchen at link.cuhk.edu.hk>
Cc: users at einsteintoolkit.org <users at einsteintoolkit.org>
Subject: Re: [Users] Unexpected Hamiltonian constraint violation from refinement when using CCZ4

Hi Kenneth,

It is expected that the mesh refinement boundaries will reflect
Hamiltonian constraint violations. From what you have provided us, it is
not clear whether the level of constraint violations that you see is a
cause of concern. You specifically mention CCZ4, but have you tried with
BSSN? Does the constraint violations look better, the same, or worse with
BSSN?

It is also possible that an issue like this might be caused by some other
problem potentially related to parameter settings, so in order to
investigate this can you provide your thornlist and parameter file?

Cheers,

   Peter

On Sun, 4 Jul 2021, CHEN, Kenneth wrote:

> While using CCZ4 to simulate a single star, the Hamiltonian constraint
> violation initially came from the star surface as expected, but after some
> time, the violation was dominated by areas near corners of refinement levels
> (by Carpet) as shown in following figures, which were strange. Any
> suggestions?
>
> Thanks and regards,
> KZ
>
>
>
>
> [IMAGE]
> [IMAGE]
> [IMAGE]
> [IMAGE]
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20210722/cc3f212f/attachment-0001.html