optimization with many variables - lsqnonlin
41 views (last 30 days)
Hi, I have an optimization problem with a large amount of variables. could be 100-10,000. Assuming that I can produce a similar amount of function, and that I have good starting guess, How many variables can lsqnonlin handle?
Is there a better mathod to solve these type of problems?
Alan Weiss on 13 Feb 2014
There is no set limit on the number of variables you can use. People routinely use Optimization Toolbox to solve nonlinear problems with thousands of variables. With a good starting guess, you can expect lsqnonlin to work.
That said, there are techniques that can speed lsqnonlin, such as using a Jacobian pattern or analytic Jacobian, or even a Jacobian multiply function. See lsqnonlin options.
MATLAB mathematical toolbox documentation