What is the best (fastest + physically proper) way to combine four isochrones (the corners of the rectangle in the (z, a) grid) into a new average isochrone? #553
Currently each dimension (magnitude, colors, initial mass) is interpolated individually in tracksPrep.interpIsochs(). This makes every isochrone have the same number of points, which allows the application of the (weighted) average later on in zaWAverage(). The problem with this is that the isochrones are not mass-aligned, which causes issues if the (z, a) grid is not dense enough (since we are combining stars of different masses as if they were of equal masses). It also causes artifacts like the one described in this issue, due to the small number of points per-isochrone (a few hundreds).
Increasing the number of interpolating points helps with this, but it also affects the performance. The topic to investigate would appear to be Multidimensional irregular grid interpolation. Related:
Currently each dimension (magnitude, colors, initial mass) is interpolated individually in
tracksPrep.interpIsochs(). This makes every isochrone have the same number of points, which allows the application of the (weighted) average later on inzaWAverage(). The problem with this is that the isochrones are not mass-aligned, which causes issues if the(z, a)grid is not dense enough (since we are combining stars of different masses as if they were of equal masses). It also causes artifacts like the one described in this issue, due to the small number of points per-isochrone (a few hundreds).Increasing the number of interpolating points helps with this, but it also affects the performance. The topic to investigate would appear to be Multidimensional irregular grid interpolation. Related:
Related #508