Auglag using Cobyla for constraints not working

[tmalewic]

I saw that in a previous issue someone posted that the lower bounds should be 1e-6. I have done this but the constraints are not satisfied when using Auglag as the global model an Cobyla as the local model.
I tried setting the equality and inequality tolerances both in Auglag and Cobyla.

Now, if I just use Cobyla as the only NLOPT model then the constraints are satisfied and a solution is found.
I am just trying to get a global model to work with equality and inequality constraints.

Has anyone gotten the this combination to work?

[stevengj]

Maybe it's a problem of your convergence tolerances?

I don't understand why you would use both auglag and cobyla? The point of auglag is to take an algorithm that doesn't support constraints and add constraints to it, but cobyla already supports constraints.

[tmalewic]

I have tried all combinations and I can't get Auglag to work. If I use only Cobyla, the equality constraint works great but it is a local model. I am looking for a global model with equality and inequality constraints.

If you do use Auglag and a local model like Nelder Mead do you provide the equality tolerance for Auglag or Nelder Mead?
I have also tried Auglag with the NLOPT_GN_DIRECT_L model, and really doesn't work.