In fact
approxloggamma(x) = x * oftype(x, log(x) - 1)
is correct to O(\log x), which is already far more accurate than the precision you’ll ever represent it to.
1 Like
In fact
approxloggamma(x) = x * oftype(x, log(x) - 1)
is correct to O(\log x), which is already far more accurate than the precision you’ll ever represent it to.