The incomplete gamma and beta functions arise in a LOT of cumulative density functions. Very accurate implementations of them are quite common and also in Boost. What I have not seen yet are implementations of their logarithms. These would be very helpful to compute accurate logarithms of CDF and SFs. One example is an issue in SciPy where a user wants to compute the logcdf of the Poisson distribution for not even very extreme values: scipy/scipy#8424. Here the logarithm of the incomplete regularized gamma function would come very much in handy.

SciPy includes the logarithm of the normal distribution CDF already but incomplete beta and gamma functions are of course a larger beast to kill.

Is this something you would find worthwile for Boost?

Edit: CC ing the special function wizards @steppi @WarrenWeckesser