lsqcurvefit - initial condition

Hi I would like to contact people using the optimization toolbox, specifically lsqcurvefit on matlab version 2022a.

Actually I have some experimental data and I am trying to combine the numerical results with the experimental points by determining new optimized parameters. Unfortunately, I do not get a good fit. Every time I change the initial conditions, the results change, and that's normal. But I would like to know if there is a more practical way to keep lsqcurvefit itself finding the correct result regardless of the initial conditions? Has anyone encountered the same issue?

Thank you in advance.

Answer 1

[Star Strider]

The problem with nonlinear parameter estimation is that the results can be highly dependent on the initial parameter estimates. The best way to approach this is with Global Optimization Toolbox functions, if you have it. I particularly like the ga function, and I have also used Global or Multiple Starting Point Search approaches successfully. They will likely provide an optimal result.

[Sana KHALED]

Thank you for ur answer, will try this.

[Star Strider]

As always, my pleasure!