Is it possible to constrain model output values using a NLS solver?

Hey,

I'm using 'lsqcurvefit' to fit a rather complex diffusion model to data in order to forecast the future progress of the diffusion. The system I am trying to model has a limited diffusion rate. I would like to implement this by constraining the model output values. However, I understand that 'lsqcurvefit' only allows lower and upper bounds for the model parameters but not for the output value itself. Is there a way to constrain the model output values using other methods?

Many thanks in advance! Regards Jonas

Answer 1

[Torsten]

Maybe fmincon (which accepts nonlinear constraints) is the suitable tool for you, applied to the objective function f=sum_i(y_sim(i)-y_exp(i))^2.

Best wishes

Torsten.