Open nilsvu opened 1 year ago
Working on this! :)
Just a note on this, @yoonso0-0 and I have recently found that after some evolution the surface can get a bit "fuzzy" (like the density goes is small but not atmosphere near the surface of the star) so it will be really helpful to have an algorithm which can find along any enthalpy/density contour, +bonus points for bonus points
Here's the plan we came up with today (involved were @PunJustice @ffoucart @nilsdeppe):
Suggested plan for the parallel algorithm / communication pattern:
I don't understand why we need a surface finder component. Can't the elements just send the surface they find directly to the non-const global cache?
We need a place to assemble the radii coming from different elements into one Strahlkorper. I think we need a component with a DataBox for that, to keep track of filled/missing Strahlkorper collocation points.
Ah okay. Could we just have one component rather than an array? Since all elements need to do a reduction over the entire domain, we need to have a vector of N
CoM calculations anyway and each element decidse which one it contributes to. So naively I think a singleton should be the right answer that then just sends to the different control systems.
Find the surface of a neutron star in numeric volume data. This is needed to adapt the computational grid to the stellar surface so the surface discontinuity doesn't spoil exponential convergence of our initial data solver. It's also useful to impose the regularity condition on the elliptic equation for the velocity potential, which we want to solve only within the star. Basic idea:
l_max
andm_max
. Then the goal is to find the radius of the surface at each angular collocation point to construct aStrahlkorper
representing the stellar surface (look atNumericalAlgorithms/SphericalHarmonics/Strahlkorper.hpp
).Literature: Read up on "surface-fitting coordinates" in the context of binary neutron star initial data. Also talk to me.