Error with lsqlin with inequality constraint using interior-point-convex

Hi there, I was using the following code

options = optimoptions('lsqlin','Algorithm','interior-point',...
'MaxIter',1000,'Display','iter','TolFun',1e-1,'TolCon',1e-1);
[x,resnorm,residual,exitflag,output,lambda]  =...
lsqlin(A_matrix,b_matrix,A_inequal,b_inequal,A_equal,b_equal,lb,ub,x0,options);

It works fine with one dataset, when I append the dataset with some more data, I have the error

Error using ipqpcommon>ipConvexQP (line 122)
The interior-point-convex algorithm requires all objective and constraint values to be finite.
Error in ipqpcommon (line 59)
Error in lsqlin (line 384)
    [X,~,exitflag,output,lambda] = ipqpcommon(H,f,A,b,Aeq,beq,lb,ub,X0, ....

I searched around but couldn't find detailed documentation on this error. Could anyone give me a guidance? Thanks, Dave

Answer 1

[Matt J]

I searched around but couldn't find detailed documentation on this error. Could anyone give me a guidance?

Not definitively without having the data to play with (you could attach it to your post for us in a .mat file), but I suspect there are some inadvertent Inf entries in A_matrix,b_matrix,A_inequal,b_inequal,A_equal,b_equal. This is not allowed.

[Dave Lee]

@Matt J: Thanks a lot! It was the issue when some data input was nan. I was led to the wrong track on debugging this. Took me almost one day.

Dave