Hello Øyvind Christiansen,
I am not an expert on BH simulations (though there are some on this list, so it is best to keep the conversation there), but I will try to at least give some pointers.
When changing the location of the black holes you need to also change the parameters to TwoPunctures.
Namely:
TwoPunctures::par_b which is half the distance between the black holes
and (this is not present in bbh.par)
TwoPunctures::center_offset[0] which is the "x" coordinate of the point on the grid from which "par_b" is measured. Ie the black holes are at
TwoPunctures::center_offset[0] + TwoPunctures::par_b
and
TwoPunctures::center_offset[0] - TwoPunctures::par_b
You should ensure that the center of mass between the two black holes is at the origin, otherwise eg extraction of gravitational waves on spheres centered around the origin will give strange results (will mix modes).
Note that TwoPunctures (and in particular its internal coordinate system) was not designed with par_b very large (more than 20 or so might be difficult, more than 100 would find "interesting" though the RIT group in particular has used TwoPunctures to produce initial data for BH that were separated by that much). You may have to increase TwoPunctures and npoints_A, npoints_B parameters (possibly npoints_phi as well though that seems less likely).
You physical accuracy will depend largely on the resolution used I would say. It is almost impossible to predict it without having done at least a couple of trial simulations at lower resolution.
When changing the mass (mass ratio really as you should try and ensure that the total mass is ~1 since many parts of the code and formulation implicitly assume this) you need to keep in mind that you may have to add extra refinement levels around the small black hole to resolve it. You will also have to adjust the
ML_BSSN::BetaDriver
parameter which should be ~ 1/2*mi around a black hole with mass mi. There is a fair amount of literature on this on the arXiv, by the RIT group, and (if my memory serves) a couple years back by Doreen Mueller, and even further back in the past by Erik Schnetter and many others.
With changing mass the the black hole radius changes since the radius is proportional to the mass so you will have to adjust the initial guess.
You usually find at least some information about thorn parameters in the thorn's param.ccl file (which sometimes is the only documentation available unfortunately).
Yours, Roland
Hi,
I'm trying to change at first the bbh.par file from the arxiv-1111.3344 folder in Cactus and later the GW150914 example at the gallery at the einsteintoolkit website, to simulate hyperbolic encounters at large impact parameters where there's overlap of validity with the post-Newtonian approximation.
I'm naively changing the TwoPunctures parameters to aim momentum a bit off of each other and start out at a larger distance. Changing the distance I also change the origin_x and initial_guess__coord_sphere__x_cent, and initial_x parameters in AHFinderDirect and PunctureTracker. I also adjust dtfac because of using larger momenta.
My questions:
Are there any other parameters I should be changing as well to accommodate this scenario?
How accurate and physical results will bbh.par yield? Is GW150914 gallery necessary? If so, should I expect reasonable results using it, when working with it black-box like this?
Are there any subtleties to changing the masses? For example regarding the initial_guess__coord_sphere__radius in AHFinder?
I couldn't find too much documentation on TwoPunctures.
I'm assuming the units are G=c=Ms=1.
Thank you for any help and/or direction!
Øyvind Christiansen
Master student, Oslo