MATLAB: How to Solve Strictly Concave Quadratic Programming with Lower Bound

Question

Hi All,

My model is a strictly concave quadratic objective function subject to lower bound constraints? And the hessian matrix of objective function is positive semi-definite.

I have calculated the coefficients in objective function, stored in matrix "hessian" and "f_qulwcLR" respectively; then I call the matlab2012a sover "quadprog" as follows:

    lb = zeros(500,1); 
    ub = +inf;
    H=-2*hessian;
    f=-f_qulwcLR; 
    [y,fval,exitflag,output,lambda] = quadprog(H,f,[],[],[],[],lb,ub);

But, it cannot be solved successfully, and the warning message is given below:

 "Optimization terminated: relative function value changing by less than OPTIONS.TolFun."

How Can I Fix this problem?

Many Thanks!

Best Wishes,

Yeeping

Answer 1

You can try passing an options structure to quadprog(), using optimset to create the options structure, specifying a lower tolerance.