Improved algorithm for calculating high accuracy values of the Chandrasekhar function
Description
The recently published set of programs H_FUN for calculation of the Chandrasekhar function [Comput. Phys. Commun. 235 (2019) 489-501] is replaced with a new set H_FUN_v2 which has the following improvements: (i) increased accuracy from 14 decimals to 15–16 decimals, (ii) the upper limit of albedo values increased from to w = 0.85 to w = 1, (iii) a reasonably short execution time, and (iv) brevity and simplicity of the code. The previously edited program HFUNELE dedicated to applications in electron transport theory is replaced with a short universal program HFUNIV that allows to reach the desired accuracy for small values of arguments, for albedo values close to unity, and also for conservative scattering. This program turns out to be very compact with listing close to one page (96 Fortran lines). Additionally, a new test program is added that illustrates the procedure for determination of accuracy by comparison with the enclosed extensive database containing the Chandrasekhar function reference values determined with accuracy of 21 decimals.