[FLASH-USERS] Checkerboard structure in magnetic field using USM solver

[John ZuHone]

Hi Sean,

What are the units of the “resistivity” parameter in this case? We should probably figure out how to translate it to Rukmani’s length / mass / time scales, which are rather large. 

John

> On Oct 9, 2017, at 10:09 AM, Sean M. Couch <couch at pa.msu.edu> wrote:
> 
> Doesn’t take much. In the movie I sent, I set the `resistivity` parameter to 1.e-4. You might also consider trying a smaller CFL. The USM solver _should_ be stable in 3D at 0.8 but I have seen some problems where a lower CFL helped clear this kind of thing up. Consider 0.5, maybe.
> 
> Cheers,
> Sean
> 
> On October 9, 2017 at 10:03:51 AM, Rukmani Vijayaraghavan (rukmani at virginia.edu <mailto:rukmani at virginia.edu>) wrote:
> 
>> Hi Sean,
>> 
>> I'd set both of those to .false., I'll try a version with .true. I've attached my flash.par file, if useful I can also send my much longer .log file (which includes many failed iterations of modifying these parameters). What is the recommended level of magnetic resistivity to minimize this?
>> 
>> Thanks,
>> 
>> Rukmani
>> 
>> 
>> On 10/09/2017 09:43 AM, Sean M. Couch wrote:
>>> I’d agree with Dongwook, it looks like numerical instability. In absence of the log file showing the details of your setup and run, I would suggest you try the following if you haven’t already. Set both E_upwind and E_modification to .TRUE. in your flash.par. What these flags do is improve the order of reconstruction for the electric fields (see Lee & Deane 2009).  See the attached gif of a field loop moving across a jump in refinement with the various flags on or off.  Not exactly the same as your issue, perhaps, but might still be worth a try. You might also try adding in magnetic resistivity to your sim, if all else fails. 
>>> 
>>> Cheers,
>>> Sean
>>> 
>>> On October 9, 2017 at 12:07:50 AM, Dongwook Lee (dongwook at flash.uchicago.edu <mailto:dongwook at flash.uchicago.edu>) wrote:
>>> 
>>>> Dear Rukmani,
>>>> 
>>>> It looks to me that they may come from numerical instability, although it could be an issue related to the use of AMR. What reconstruction scheme and CFL number are you using? Is this 2D or 3D?
>>>> 
>>>> Best,
>>>> Dongwook
>>>> 
>>>> On Sun, Oct 8, 2017 at 12:59 PM, Rukmani Vijayaraghavan <rukmani at virginia.edu <mailto:rukmani at virginia.edu>> wrote:
>>>> Hi everyone,
>>>> 
>>>> I get a strange, unphysical feature in the magnetic field structure with the USM solver, seen in the attached slices of B_z and the magnetic pressure. These features, as you can see, track the block boundaries and look like a checkerboard of varying magnetic field strength (and sign?) in individual grid cells. I've tried modifying every possible USM solver switch, including the Riemann solver, order, slope limiter, prolongation method, steepening, etc. Using Riemann Solver = HLLC instead of HLLD makes things a little better, since HLLC is more diffusive, but the feature still eventually appears. Any advice on solving this or even minimizing its effects would be helpful!
>>>> 
>>>> Thanks,
>>>> 
>>>> Rukmani
>>>> 
>>>> --
>>>> Rukmani Vijayaraghavan
>>>> NSF Astronomy & Astrophysics Postdoctoral Fellow
>>>> University of Virginia
>>>> rukmani at virginia.edu <mailto:rukmani at virginia.edu>
>>>> 
>>>> 
>>>> 
>>>> 
>>>> --
>>>> 
>>>> =========================================
>>>> Dongwook Lee, Ph.D., Assistant Professor
>>>> Applied Mathematics and Statistics
>>>> University of California, Santa Cruz
>>>> Baskin Engineering, Room 353C
>>>> 1156 High Street, Santa Cruz, CA 95064
>>>> https://users.soe.ucsc.edu/~dongwook/ <https://users.soe.ucsc.edu/%7Edongwook/>
>>>> 
>> 
>> --  
>> Rukmani Vijayaraghavan
>> NSF Astronomy & Astrophysics Postdoctoral Fellow
>> University of Virginia
>> rukmani at virginia.edu <mailto:rukmani at virginia.edu><part1.DD5AE1D1.D72BD91E at virginia.edu>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20171009/fcecf2ae/attachment-0001.htm>