Bayesian Optimization for constrained optimization in fmincon?

Hello, I have been doing constrained optimizations of underdetermined systems of linear equations with fmincon.

The optimization algorithm is too slow for my application, therefore I was wondering if there is a way I can just substitute in the Bayesian Optimization algoirthm instead?

I am interested in using the Bayesian Optimization method to optimize the objective function because I heard the it is computaionally very fast.

Thanks!

Answer 1

[Torsten]

I have been doing constrained optimizations of underdetermined systems of linear equations with fmincon.

Isn't "lsqlin" the appropriate tool for this ?

[SV]

In my case I am not interested in a the least squares solution because there a many non-unique solutions, and the least squares solution is just one of them, and I have constraints on model error that reflect the analytical uncertainty of the chemical analysis instrument.

I am instead using an objective function that solves for the maxiumum and minimum solutions that are valid mass balance solutions to within analytical error.

[SV]

Therefore I am wondering if I can use Bayesian Optimization as the alogithm that finds the minimum of the objective function (which is to find the minumum possible value of a particluar variable, subject to strict linear constraints).

[Torsten]

I think you can use Bayesian Optimization, but my guess is that it's far slower than fmincon.

This should work better for your purpose:

https://de.mathworks.com/matlabcentral/fileexchange/8277-fminsearchbnd-fminsearchcon

For other software, check here

http://plato.asu.edu/sub/nlores.html

under the point

f general, bound constraints only