I added a wrapper lib for Gil Segura Temme maths...

I added a wrapper lib for Gil-Segura-Temme maths with the following signatures:

Scorer functions

• double scorer_gi(double x);
• double scorer_hi(double x);

Parabolic cylinder functions.

• double parab_cylinder_u(double a, double x);
• double parab_cylinder_v(double a, double x);

Incomplete gamma functions.

• double pgamma(double a, double x);
• double qgamma(double a, double x);

Toroidal harmonic functions

• double tor_harmonic_p(unsigned int l, unsigned int m, double x);
• double tor_harmonic_q(unsigned int l, unsigned int m, double x);

Spheroidal harmonic functions

• double pro_sph_harmonic_p(unsigned int l, unsigned int m, double x);
• double pro_sph_harmonic_q(unsigned int l, unsigned int m, double x);
• double obl_sph_harmonic_r(unsigned int l, unsigned int m, double x);
• double obl_sph_harmonic_t(unsigned int l, unsigned int m, double x);

I believe distribution functions like pgamma and qgamma might be better off not to return single numbers but should return both p and q. Invariably, you have to compute both and if one is tiny - near a machine epsilon or less - (often q) you'll just lose information with the P function. So I'm thinking of adding:

Incomplete gamma functions.

• gamma_cdf_t gamma_cdf(double a, double x);

Inverse Incomplete gamma functions.

• double inv_gamma_cdf(double a, gamma_cdf_t pq);

Where gamma_cdf_t is a pair basically. I just need to figure out how I want the API to look. Maybe a generic cdf_t, maybe just std::pair and be done...

Of course, I'm not done with any of these functions in the library yet.