Running the lsqlin interior-point algorithm for specific optimization problems is much slower in R2017a than it was in R2016b.
Example:
n=2000;A(:,1)=(1:n)';A(:,2)=ones(n,1);b=rand(2000,1);tic; x1 = lsqlin(A,b,A,b,[],[],[],[],[],optimoptions(@lsqlin,'Algorithm', 'active-set')); toc tic; x2 = lsqlin(A,b,A,b,[],[],[],[],[],optimoptions(@lsqlin,'Algorithm', 'interior-point')); toc
returns
Elapsed time is 0.021589 seconds.
Elapsed time is 3.926372 seconds. (!)
Best Answer