This is a Python wrapper for the FFTLog code in CCL. Specifically, it wraps the fftlog_ComputeXi2D function and tries to reproduce the behaviour of ccl_tracer_corr_fftlog by doing the same power law extrapolation of the Cl's at high ell. The agreement is very good for xi_minus but xi_plus seems to be dependent on the details of the interpolation, such that the agreement is only ~1e-3 at scales of 10 degrees.
The usual
python setup.py install
should do the trick.
xi_plus = pycl2xi.fftlog.Cl2xi(Cl, ell, theta, bessel_order=0)
xi_minus = pycl2xi.fftlog.Cl2xi(Cl, ell, theta, bessel_order=4)
The ranges of ell used for FFTLog can be set with ell_min_fftlog, ell_max_fftlog, and n_ell_fftlog.
There is currently one test in test/ that compares the correlation functions of this module with that of CCL. Passing --plot to test_against_ccl.py creates a plot of the fractional difference.