This repo contains an example showing how one can calculate the two-point correlation function more accurately than with random points at no extra computational cost.
The basic idea is to use a low discrepance sequence instead of random points. We use randomized Halton sequences as provide by SciPy and for the pair-counts we use Corrfunc. A brief explanation why such an approach improves the accuracy with no extra cost can be found in corracc.pdf. See also Kerscher 2022 for more details.
In corracc.py the implementation of the standard Landy & Szalay estimator is given, followed by the estimator using a low discrepancy sequence. Both are using the pair-counts from Corrfunc.
In xi.py the usage of these functions is illustrated with the test data set gals_Mr19ff from the Corrfunc repository.
In corracc.py also the implementation of the exact Landay & Szalay estimator and of an estimator for periodic boundary conditions is given.
For the exact estimator we need the geometric functions from baddeley.py. The implementation closely follows the C-code in sphefrac.c from the R package spatstat (written by A.J. Baddeley).