[FLASH-USERS] refinement problems with FLASH 4.2

Thu May 22 12:00:06 EDT 2014

Hi Thomas,

this is very strange. I would like to run your setup myself, but you said it's the SinkRotatingCloudCore setup, which I added to FLASH, in order to test the sink particles. Have you tried running the test in FLASH4.2.1 (latest release) without making any own modifications? I'll also do that asap and see what I get.

Kind regards,

Christoph

Am 22.05.2014 um 04:36 schrieb Thomas Peters:

> Dear Kevin, Christoph and John,
> 
> the attached slice shows that the density field is smooth across refinement boundaries. Since the velocity field jumps across the boundary, the momentum must jump as well.
> 
> I have also attached slices at exactly z = 0, as well as at 10 and 20 times the finest grid size off the midplane. The effect is still clearly visible, there are just fewer jumps in refinement in that region.
> 
> I would not call this "noise" since in these slices, the z-velocity is of the same size as the x- and y-velocities were in the previous slices.
> 
> Not sure if this helps, but I have not seen this effect in any of the collapse simulations I have run with FLASH 2.5.
> 
> Best regards,
> Thomas
> 
> 
> Am 2014-05-21 23:50, schrieb Kevin Olson:
>> Dear Thomas Christoph, John,
>> The variable you might want to look instead of velocity is
>> momentum...momentum is a conserved quantity: FLASH (with Paramesh)
>> guarantees conservation even at refinement jumps.  There is no
>> guarantee that non-conserved quantities will behave properly at
>> refinement jumps.
>> Best,
>> Kevin
>> On May 21, 2014, at 12:54 PM, Thomas Peters wrote:
>>> Dear Kevin, Christoph and John,
>>> this is not a plotting issue, I can see these jumps using yt as well. Also, I don't think this is noise since the maximum velz value is only a factor of 10 smaller than the maxmimum velx value (in magnitude). The slices are not centered around zero but on the cell with the highest density. I am using the default paramesh library and hydro solver of the setup in FLASH 4.2.0.
>>> Best regards,
>>> Thomas
>>> Am 2014-05-21 18:43, schrieb John ZuHone:
>>>> Thomas,
>>>> I would also suggest looking at the dataset in yt:
>>>> http://yt-project.org [2]
>>>> A python script that would get you going with looking at slices is:
>>>> from yt.mods import *
>>>> ds = load(your_filename)
>>>> slc = SlicePlot(ds, "z", ["velx", "vely", "velz"])
>>>> slc.annotate_grids()
>>>> slc.save()
>>>> you can find more examples here:
>>>> http://yt-project.org/docs/2.6/cookbook/simple_plots.html [3]
>>>> On May 21, 2014, at 11:58 AM, Christoph Federrath
>>>> <christoph.federrath at monash.edu> wrote:
>>>>> Hi Thomas, Hi Kevin,
>>>>> I don't think it's the refinement criterion. There shouldn't be big
>>>>> jumps and I don't think there are. Isn't it that you simply plot
>>>>> values very close to zero in velz? This is basically a slice in the
>>>>> mid plane of the disk, which should actually have zero velocity in z
>>>>> direction, because everything is symmetric in z direction? Thus, the
>>>>> velocity in the z=0 plane should be zero and what you are plotting
>>>>> is basically numerical noise around zero values?
>>>>> I'm also wondering whether this particular plotting issue might have
>>>>> something to do with the way xflash interpolates values? Have you
>>>>> tried using visit and making the same cut in z? Definitely, the velz
>>>>> axis you are showing is funny and seems skewed to negative values,
>>>>> but they are all quite small compared to sound speed. If this is
>>>>> really a cut at exactly z=0, then the z velocity should be zero.
>>>>> What paramesh lib (implementation) are you using? And what hydro/MHD
>>>>> solver are you using? Is this in the latest FLASH4.2.1?
>>>>> Kind regards,
>>>>> Christoph
>>>>> ________________________________
>>>>> Dr. Christoph Federrath
>>>>> Monash Centre for Astrophysics,
>>>>> School of Mathematical Sciences,
>>>>> Monash University,
>>>>> Clayton, VIC 3800, Australia
>>>>> +61 3 9905 9760
>>>>> http://www.ita.uni-heidelberg.de/~chfeder/index.shtml?lang=en [1]
>>>>> Am 21.05.2014 um 08:47 schrieb Kevin Olson:
>>>>> Hi Thomas,
>>>>> This is due to the fact that the Sink particle test uses only a
>>>>> Jeans criterion based refinement test. This does nothing to detect
>>>>> shocks and other discontinuities in the flow. I think you will find
>>>>> that if you turn on the standard FLASH refinement criterion plus the
>>>>> Jeans length criterion, you problem will go away...but at the cost
>>>>> of having many more refinements.
>>>>> Also, the problem you are running is basically that of a collapsing
>>>>> cloud and the regions where you are seeing large errors probably
>>>>> don't have a large impact on the overall solution you are looking
>>>>> for since those regions probably don't have a very high density.
>>>>> Hope this helps.
>>>>> Kevin Olson
>>>>> On May 21, 2014, at 11:18 AM, Thomas Peters wrote:
>>>>> Dear FLASH users,
>>>>> I am having some problems with the grid refinement in FLASH 4.2. I
>>>>> have attached some slices of a simulation run with the standard
>>>>> setup SinkRotatingCloudCore. The only thing I modified was to set
>>>>> lrefine_max to a larger number in order to get more refinement
>>>>> levels.
>>>>> The attached z-slices show very strong jumps in velz at refinement
>>>>> boundaries, while velx and vely remain smooth. However, velx has
>>>>> jumps in x-slices and vely has jumps in y-slices as well.
>>>>> Has anyone seen such a problem before and knows how to get rid of
>>>>> it?
>>>>> Thanks for your help,
>>>>> Thomas
>>>> <BB_hdf5_plt_cnt_blk_velz0142.png><BB_hdf5_plt_cnt_velx0142.png><BB_hdf5_plt_cnt_vely0142.png><BB_hdf5_plt_cnt_velz0142.png>
>>>> Links:
>>>> ------
>>>> [1] http://www.ita.uni-heidelberg.de/~chfeder/index.shtml?lang=en
>>>> [2] http://yt-project.org
>>>> [3] http://yt-project.org/docs/2.6/cookbook/simple_plots.html
> <BB_hdf5_plt_cnt_dens0142.png><BB_hdf5_plt_cnt_velz0142_z0.png><BB_hdf5_plt_cnt_velz0142_z10.png><BB_hdf5_plt_cnt_velz0142_z20.png>