Is Matlab generally faster at computing polynomials or trigonometric functions?

Hi, I’ve got a pretty complicated mathematical function involving a lot of sin, cos, square roots, etc and it is called a large number of times in the program which can lead to decently long runtimes. The exact value isn’t necessarily important, so I can approximate my function with a sufficiently accurate polynomial. Is this worth it in general? I.e. is matlab ‘faster’ at computing powers than it is at computing trigonometric functions?

[KSSV]

The mat libraries behind calculate Trigonometirc functions by approsimating it to a polynomial. Rememebr Taylor series. You need not to worry about time. Read about the functions tic and toc to find the time taken. Read about profile to check your code.

Answer 1

[Rik]

You mean you want to approximate your complicated function with many calls by a single polynomial? If that is possible with your data, that would indeed be faster, since a single polynomial calculation will be faster than many calculations put together.

To find a middle ground in terms of accuracy you could even considering determining a few domains of input ranges and compute the polynomial for each domain. That will be a trade-off between determining the polynomial and accuracy (since you obviously shouldn't compute a polynomial for every unique input).

If you need help with implementation, you should post your current function, along with reasonable example inputs.