Computational times for various routines?

I am solving a "stochatic" optimization problem using fminunc/fmincon. I'm currently using a fairly simple construction, but am slowly adding layers of complexity (additional time periods, variables, dynamics, etc.). Not surprisingly each addition slows the algorithm down considerably.

Is there a way to see (e.g. for each iteration the computational time spent on 'factorKKTmatrix') how much time each subroutine is taking to compute? I have a feeling my bottle neck is the finite-difference approach to computing the gradient/hessian, but I'd like to make sure this is the case.

Thanks

Answer 1

[Azzi Abdelmalek]

You can use tic toc

[Pete]

Yes, I'm using that for the entire routine... but for example how long is the algorithm taking to compute the hessian or the gradient versus the other operations I have within the optimization?

[Azzi Abdelmalek]

In each part of your code use for example

 tic
% part1
t1=toc
tic
%part2
t2=toc

Answer 2

[Kelly Kearney]

Try the Profiler.

doc profile