forum.alglib.net

I'm using NLEQ to estimate a distortion model for a camera. I already have a Scilab code that estimates the distortion so I know a good solution and alglib's nonlinear solver NLEQ also finds the known solution but returns report code,
-4 ERROR: algorithm has converged to the
stationary point Xf which is local minimum of
f=F[0]^2+...+F[m-1]^2, but is not solution of
nonlinear system.

In this problem, the number of equations m=224 and the dimension of the state is n=61. How does NLEQ know that it has only found a local minimum and that there is a global minimum?

I think that you have too small EpsF (quantity which is set by NLEQSetCond function). If algorithm stops at point where |F|>EpsF, it considers this point at local extrema of F^2 which is NOT solution.

Try increasing EpsF... optimal value is problem-specific.