Change owens_t_T1_accelerated to use larger n and stop on convergence. Also returns an error bound.
Add arbitrary precision versions of T2 and T4.
Refactor owens_t_dispatch into different version for different precisions.
Arbitrary precision code now pretty much works, but gets rather inaccurate for a very close to 1.
(No files)
Note:
See TracChangeset
for help on using the changeset viewer.