1 * discrete inverse distributions - easy apart from hypergeometric(?)
3 * look for integer overflow in the discrete functions - this could be
4 hard to find - perform computations in double to avoid.
6 gsl_cdf_binomial_P (unsigned int k, double p, unsigned int n);
7 gsl_cdf_binomial_Q (unsigned int k, double p, unsigned int n);
8 gsl_cdf_poisson_P (unsigned int k, double mu);
9 gsl_cdf_poisson_Q (unsigned int k, double mu);
10 gsl_cdf_geometric_P (unsigned int k, double p);
11 gsl_cdf_geometric_Q (unsigned int k, double p);
12 gsl_cdf_negative_binomial_P (unsigned int k, double p, double n);
13 gsl_cdf_negative_binomial_Q (unsigned int k, double p, double n);
14 gsl_cdf_pascal_P (unsigned int k, double p, unsigned int n);
15 gsl_cdf_pascal_Q (unsigned int k, double p, unsigned int n);
16 gsl_cdf_hypergeometric_P (unsigned int k, unsigned int n1,
17 unsigned int n2, unsigned int t);
18 gsl_cdf_hypergeometric_Q (unsigned int k, unsigned int n1,
19 unsigned int n2, unsigned int t);
21 * Unify the beta_inc function with the special functions, put all the
22 functionaity in gsl_sf_beta_inc, providing a new function for the
23 AXPY part if necessary (can we do everything with gsl_sf_beta_inc
24 and gsl_sf_beta_inc_1mx keeping in mind that we cannot do 1-x
25 because of cancellation for small x)