[Users] McLachlan crash

Wed Feb 27 11:53:18 CST 2013

Vassili

Yes, this plot looks good.

Do you expect the black hole to be stationary? If so, you could look at the
spacetime quantities near the puncture (the McLachlan variables, not the
ADMBase variables) and see which begins to vary or to drift, causing the
trouble in the end.

-erik

On Wed, Feb 27, 2013 at 11:30 AM, Vassilios Mewes <vassilios.mewes at uv.es>wrote:

> hello Erik,
>
> i set ML_BSSN::MinimumLapse  = 1e-8 in the par file...
>
> i think the output of the minimum lapse is not correct (in the plot i sent
> you i plotted admbase::lapse.minimum.asc), because the minimum lapse is
> actually much smaller...
>
> i have attached a plot of the lapse along x (admbase::lapse.x.asc) at the
> final time step that was output..
>
> best wishes,
>
> Vassili
>
>
> On Wed, Feb 27, 2013 at 5:11 PM, Erik Schnetter <schnetter at cct.lsu.edu>wrote:
>
>> Vassilli
>>
>> This lapse minimum looks strange. During evolution, if there is a
>> puncture, the lapse minimum should be much smaller than 0.4.
>>
>> Of course, in the end (when the lapse minimum becomes negative), things
>> go wrong. You could set McLachlan's parameter to enforce a positive lapse,
>> e.g. enforcing that the lapse does not go below 10^-10.
>>
>> -erik
>>
>>
>> On Wed, Feb 27, 2013 at 11:01 AM, Vassilios Mewes <vassilios.mewes at uv.es>wrote:
>>
>>> hello Erik,
>>>
>>> i have attached a plot of the minimum value of the lapse vs time....
>>>
>>> here is the segment of the par file where i set the dissipation:
>>>
>>> Dissipation::order = 5
>>> Dissipation::vars = "
>>>         ML_BSSN::ML_log_confac
>>>         ML_BSSN::ML_metric
>>> ML_BSSN::ML_trace_curv
>>>         ML_BSSN::ML_curv
>>>         ML_BSSN::ML_Gamma
>>>         ML_BSSN::ML_lapse
>>>         ML_BSSN::ML_shift
>>>   ML_BSSN::ML_dtlapse
>>>         ML_BSSN::ML_dtshift
>>> "
>>>
>>> best wishes,
>>>
>>> Vassili
>>>
>>>
>>> On Wed, Feb 27, 2013 at 4:45 PM, Erik Schnetter <schnetter at cct.lsu.edu>wrote:
>>>
>>>> Vassili
>>>>
>>>> In my experience, one needs to "control" the puncture by ensuring that
>>>> (a) things evolve only slowly there (keeping the lapse small) and (b)
>>>> adding sufficient dissipation.
>>>>
>>>> What is the value of the lapse near the puncture?
>>>> How much dissipation are you adding?
>>>> To which variables are you adding dissipation?
>>>>
>>>> -erik
>>>>
>>>>
>>>> On Wed, Feb 27, 2013 at 8:53 AM, Vassilios Mewes <vassilios.mewes at uv.es
>>>> > wrote:
>>>>
>>>>> Hello all,
>>>>>
>>>>> still no success with the full evolution of BH+torus...
>>>>>
>>>>> i have attached the par file i have been using lately, in which i
>>>>> added two more levels of mesh refinement, as Roland suggested, increased
>>>>> the Dissipation order to 5th and removed the parameter
>>>>> ML_BSSN::dt_lapse_shift_method = "noLapseShiftAdvection"..
>>>>>
>>>>> this has improved things a bit, but the L_inf norm of the hamiltonian
>>>>> constraint still starts growing at around t=200 (after having decreased and
>>>>> then having been rather constant up to that point) and then the simulation
>>>>> will crash later (either due to the shift being nan at the puncture or by
>>>>> detecting nans)
>>>>>
>>>>>
>>>>> in the meantime, i have evolved a single puncture, the bondi accretion
>>>>> onto a puncture and an "inverse cowling" (setting up the spacetime and
>>>>> hydro variables of the BH+torus system but only evolving the spacetime),
>>>>> all using the same parameters, and all of those work...its only the full
>>>>> system of BH+torus when i run into problems...this is using the ET 2012.11.
>>>>> release..
>>>>>
>>>>> do you have any ideas what might go wrong with the spacetime evolution
>>>>> when evolving the full system?
>>>>>
>>>>> best wishes,
>>>>>
>>>>> Vassili
>>>>>
>>>>>
>>>>> On Tue, Feb 12, 2013 at 2:07 AM, Roland Haas <
>>>>> roland.haas at physics.gatech.edu> wrote:
>>>>>
>>>>>> Hello Vassilios,
>>>>>>
>>>>>> > as mentioned earlier in the ET phone call, i have attached the par
>>>>>> file
>>>>>> > used in the run...
>>>>>> >
>>>>>> > i tried checking for the nans in the output, but the last hdf5
>>>>>> output was
>>>>>> > too many iterations earlier and they aren't visible in 1D ascii
>>>>>> output
>>>>>> > either
>>>>>> >
>>>>>> > as said, the puncture tracker actually aborted the run, not the nan
>>>>>> > checker...
>>>>>> Thank you for the parameter file. As a unsubstantiated guess, I would
>>>>>> try what happens if you move the outer boundary further out. Right now
>>>>>> it sits at 50M which I would find uncomfortably close if there is any
>>>>>> say "junk" radiation (eg from setting up a BH with spin in flat
>>>>>> background) present. I'd move it to at least 200M (ie two more levels
>>>>>> of
>>>>>> mesh refinement). Not sure how likely this explanation is since your
>>>>>> NaNs appear after many (9) light crossing times.
>>>>>>
>>>>>> Yours,
>>>>>> Roland
>>>>>>
>>>>>> --
>>>>>> My email is as private as my paper mail. I therefore support
>>>>>> encrypting
>>>>>> and signing email messages. Get my PGP key from http://keys.gnupg.net
>>>>>> .
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> Users mailing list
>>>>>> Users at einsteintoolkit.org
>>>>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Users mailing list
>>>>> Users at einsteintoolkit.org
>>>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>> Erik Schnetter <schnetter at cct.lsu.edu>
>>>> http://www.perimeterinstitute.ca/personal/eschnetter/
>>>>
>>>
>>>
>>
>>
>> --
>> Erik Schnetter <schnetter at cct.lsu.edu>
>> http://www.perimeterinstitute.ca/personal/eschnetter/
>>
>
>

-- 
Erik Schnetter <schnetter at cct.lsu.edu>
http://www.perimeterinstitute.ca/personal/eschnetter/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.einsteintoolkit.org/pipermail/users/attachments/20130227/410a9565/attachment.html 