stable approximation of (1/cosx)

I'm trying to simulate a function that has a 1/cos(x) term. The value of 1/cos(x) tends to infinity as cos(x) approaches zero. Is there any way to smoothly approximate this function in order to avoid the jumps around pi/2 neighborhood? I'm looking for some sort of smoothing, but in real-time. Please help.

Thanks!

Answer 1

[Jan]

The function 1/cos(x) goes to infinity for cos(x) -> 0. There is absolutely no way to "smooth" this away. The results would be wrong from the view point of mathematical correctness. Even "real-time" is not the problem here.

If you have any good reason why the pole does not appear in the physical system (or what ever this formula describes), please mention it here.