MATLAB: Matlab 2019b optimoptions not work on sparse jacobPattern anymore

jacobianoptimoptions

I have just recently update to Matlab2019b, but the optmoptions did not work as I expect.

For example I have sparse jacobian pattern matrix create by A=speye(100000) and feed to optimoption by

'options = optimoptions(@fsolve,'Display','iter','JacobPattern',J,'Algorithm','trust-region','SubproblemAlgorithm','cg');'

It just complain as follows, seems Matlab2019b tries to construct a dense matrix from the sparse. While this code will work on Matlab2018b.

Anyone can help for that, please? Thanks

Best Answer

This is a bug in optimoptions in R2019b. Thanks for reporting this!

As a workaround until it is fixed, you can use optimset

options = optimset('Display','iter','JacobPattern',J,'Algorithm','trust-region','SubproblemAlgorithm','cg');

======== Update ==========

Note, this bug is fixed in R2020a and later releases.

Related Solutions

MATLAB: Using lsqnonlin on large-scale problems – Question about “JacobMult” option

If you use JacobMult, I imagine it will try to solve for the Newton steps iteratively, instead of with QR decomposition, which could be faster, depending on how sparse your Jacobian really is.

Basically, the Newton step amounts to solving a linear equation

J*x=F

where J is your Jacobian. There are iterative algorithms which can do this using a series of multiplications with J and J.'. In your case, the JacobMult function could be implemented simply as

function W=jmfun(J,Y,flag)
   if flag==0
    W = J'*(J*Y);
   elseif flag>0
     W= J*Y;
   else%flag<0
     W=J.'*Y;
   end
end

MATLAB: Warning about optimset and using optimoptions

Your best option is in the error message:

set option Algorithm to 'trust-region' using optimset.

Best Answer

Related Solutions

MATLAB: Using lsqnonlin on large-scale problems – Question about “JacobMult” option

MATLAB: Warning about optimset and using optimoptions

Related Question