[FLASH-USERS] 2D +spherical point mass AND self-gravity question

[Sean Couch]

Hi Ivo,

Just catching up on this.  The easiest solution to your problem is to switch to the “old” multipole solver.  That has both the additional point mass and the ability for rmin > 0.

Norbert is correct that I have added an additional point mass potential to the “new” multipole solver.  This is really trivial as it is just a additional -GM/r term in the final potential calculation.  The new multipole solver still lacks the ability to have rmin > 0, although this is also not difficult, in principal.  If the old one can do it, the new one can too.  I just haven’t looked into doing it.

All that said, for 2D spherical the only thing the new solver will by you is speed (it’s much faster), but for a 2D problem, who cares?  The old solver will do the job admirably.  I still use it from time to time when I need rmin > 0, or a 3D octant geometry…

Sean

> On May 5, 2015, at 2:44 PM, Christoph Federrath <christoph.federrath at gmail.com> wrote:
> 
> 
> Hi Ivo, Hi Norbert,
> 
> as an alternative, the (new) sink particle unit can be used to simulate multiple (active) point masses. However, it is currently only guaranteed to work in 3D cartesian geometry. But I’d be happy to try a generalization of the module to work in any geometry. BTW: The gravitational interactions of the sinks with the gas are computed by direct summation over all grid cells and sink particles, so it is decoupled from the gas gravity solver anyway; the sink particles basically have their own gravity solver. Please let me know if you are interested in looking into this option.
> 
> Kind regards,
> 
> Christoph
> 
> 
> On 5 May 2015, at 23:32, Norbert Flocke <flocke at flash.uchicago.edu> wrote:
> 
>> Hello Ivo,
>> 
>> The new multipole solver does NOT support additional point mass. This is a feature that remains to be implemented. The old multipole solver does support an additional point mass, however the old code is outdated and has some numerical flaws, so you have to use it at your own risk. I think that Sean Couch (at Caltech right now) implemented his local version of the new multipole solver with additional point mass as an option, so probably you should contact him. Right now I simply don't have the time to add this additional feature to FLASH's official new multipole solver. Maybe in the future sometime, when my workload is a bit less and there is sufficient demand for the point mass option.
>> 
>> Best regards,
>> Norbert
>> 
>> 
>> On Tue, 5 May 2015, Ivo Seitenzahl wrote:
>> 
>>> Dear FLASH-users,
>>> I have run into a problem with the gravity for my problem. The geometry is +spherical and dimensionality is 2D.
>>> 
>>> What I've been trying to set up is self-gravity (as far as I can tell only multi-pole is supported in 2D+spherical) IN ADDITION to a single central "point_mass".  Essentially what I have is a central compact object (think of it as a black hole) surrounded by a self-gravitating envelope. So a further complication is that xmin > 0.0 (I have excised a central region from the computational domain, still, so the point_mass would sit outside the computational domain).
>>> 
>>> Looking here it seems to me that Poisson gravity and point_mass is a valid combination.
>>> http://flash.uchicago.edu/site/flashcode/user_support/rpDoc_4p22.py?submit=rp_Gravity.txt
>>> 
>>> However, it seems to be a strict requirement (by that the grid should begin at r=0 (i.e. xmin=0), which is communicated clearly as an ERROR by gr_mpoleInit.F90 if I build the problem with +newMpole.
>>> 
>>> *****
>>> if (radialXGrid) then
>>>    if (gr_mpoleDomainXmin /= ZERO) then
>>>        call Driver_abortFlash ('[gr_mpoleInit] ERROR: radial X-coord out of range')
>>>    end if
>>> end if
>>> *****
>>> 
>>> Am I just SOL or am I missing something? If anyone has any insight how to setup self-gravity in 2D+spherical (with xmin > 0.0) and a central (fixed at the origin) point mass then I would greatly appreciate any suggestions/help.
>>> 
>>> Best wishes from Down-Under,
>>> 
>>> Ivo
>>>