simplify trigonometric and hyperbolic functions with argument I*something #162
Conversation
This is pretty optimal, hard to see how to make it faster. |
You are right; I've overestimated the cost of the additional The following behavior seems inconsistent to me:
We should probably stop raising errors at poles and simply return |
|
I'm a bit irritated where the Nevertheless, related to #164? |
Nevermind, just found out that this is in fact pynac's |
Yes, |
I'm very sorry that I haven't looked at this for so long. I have now fixed the conflict, I get a funny fail:
|
I'll implement |
The code here currently implements the simplifications
sin(I*something) --> I*sinh(something)
andcos(I*something) --> cosh(something)
.Performance probably is a bit worse (a call to
real_part
is needed, which is python/cython).This should only be considered mergeable after the following TODOs are done: