Algorithms

From Eric's Wiki

Back to Papers

EWR06: Ebeling, White, Rangarajan
ASMOOTH: Kernel smoothing of 2-d data.
Much the same as my numdensity routine. Except the smoothing length is not simply n ^{− m} where n is the number density. Like numdensity, the kernel size is chosen to fix the S/N ratio (i.e. have a constant number of points). Howerver, the S/N computed is relative to the background; so the kernel expands to collect N = N_cts − N_bkg counts. Curiously, the kernel size should, by their definition, scale with density. But it doesn't.