[FLASH-USERS] Domain issues in the core-collapse problem

[Sean Couch]

Hi Cole,

> 1. Excise an inner region with radius less than some value, and simply compute the gravity from the excised mass as a point source. This would seem to involve both modifying the gravity solver to allow xmin > 0 (perhaps not possible with the multipole solver), and creating a hybrid between the poisson gravity solver and the point mass gravity implementation. My main interest will be in the outer regions of the star, so I don’t want my timestep to be limited by the inner regions.

The most direct way of doing this is to, as you say, set xmin > 0 in spherical coordinates.  The New Multipole solver (of Couch, Graziani, & Flock 2013) doesn’t currently support this behavior but the old Multipole solver does.  And since you are doing 1D there really isn’t any benefit in using the new solver.  The old Multipole solver also already includes the addition of a point mass potential on top of the self-gravity of the fluid on the grid.  Just set the runtime parameter point_mass to the mass, in grams, of your mass cut and you should be good to go.

There are other approaches to doing what you want, but this is the most direct.  If you’re happy using spherical coordinates in 3D, then the transition from 1D to 3D should be relatively painless.  

> 2. Have a moving grid, so that the domain can expand over time. I haven’t seen any mention of this capability in FLASH, so I’m guessing it isn’t possible, but I thought I’d ask anyway.

This had been tried long ago in the split PPM solver with little success.  My solution to this (Couch et al. 2009, ApJ, 696, 953) was to periodically interpolate the solution on to an expanded domain.  Matching the solution to the new initial conditions for the regions that are added in this can be tricky, if you’re not careful.  Another approach (Couch et al. 2011, ApJ, 727, 104) is to simply brute-force it and include the entire domain of interest from the very start and allow the AMR to help you.  You can limit the max refinement level as a function of radius and you can also reduce the global max refinement level as a function of time, allowing you to derefine the inner core and increase your time step size.  Check out the runtime parameter descriptions for the Grid unit here:

http://flash.uchicago.edu/site/flashcode/user_support/rpDoc_4p22.py?submit=rp_Grid.txt

And pay particularly attention to gr_lrefineMaxRedDoByTime, gr_lrefineMaxRedDoByLogR, and friends.  Using this, I was able to follow a developing supernova explosion all the way from the core to shock breakout and beyond in 2D.  Anecdotally, I found that each decade in radius included in the sim was roughly the same cost as the previous decade (i.e., the expense scaled logarithmically, not linearly).

Good luck,
Sean



-----------------------------------------------------------
Sean M. Couch
Sr. Postdoctoral Scholar
Theoretical Astrophysics Including Relativity
California Institute of Technology
Mail Code 350-17
Pasadena, CA  91125
(626) 395-4282
www.flash.uchicago.edu/~smc

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20141107/ee83a542/attachment.htm>