You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
using timeit in colab, special.ive takes ~1.5 µs, new cython code takes ~500 ns, pure python takes 18 µs
Error is still large when v and z are small. It probably makes sense to call np.log(special.ive(v, z)) when v and z are small and only use this when v and z are large.
As a follow up to #12244 I had fun adding a function to calculate
log(ive)
.@ev-br mentioned this is already done in _hyp0f1.pxd#L51
This points to DLMF 10.41.10, I extended _hyp0f1 to k=5 using A144617, which reduces max error.
I'd love some advice from @WarrenWeckesser about
stats.ncx2
using
timeit
in colab,special.ive
takes ~1.5 µs, new cython code takes ~500 ns, pure python takes 18 µsError is still large when
v
andz
are small. It probably makes sense to callnp.log(special.ive(v, z))
when v and z are small and only use this when v and z are large.Click to toggle implementation
Click to toggle results
The text was updated successfully, but these errors were encountered: