fmincon: Matrix is singular to working precision.

Dear all,

I get the following message but the code seems to be working ok

 In fmincon (line 797)

 Warning: Matrix is singular to working precision.

Is there a way to rectify this problem?

[Matt J]

What version of MATLAB?

[Walter Roberson]

And which options did you pass? For example that message at that line number is plausibly due to a singular Hessian.

[ektor]

version 2017a. I use this option

    options=optimoptions('fmincon', 'Display','off');

and I get this

Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND = 3.584036e-249.

Thanks

[Walter Roberson]

Is it possible that you are working with relatively large values, especially with a lot of variables? If so then the gradient estimation or hessian estimation could produce a singular matrix due to overflow.

[Fabian Friz]

Hi there,

Do you have a solution for this ? Would computing the derivatives yourself solve the problem ?

[Walter Roberson]

Options:

TypicalX Typical x values. The number of elements in TypicalX is equal to the number of elements in x0, the starting point. The default value is ones(numberofvariables,1). fmincon uses TypicalX for scaling finite differences for gradient estimation.