Hacker News new | ask | show | jobs
by jason_s 4010 days ago
No mention of Padé approximants? I'd just use a Chebyshev approximation transformed into a Padé approximation, then perturb the coefficients to optimize maximum error.