I'm seeing unexpected behavior in an optimization being performed with fmincon with non-linear constraints.
1) Without non-linear constrains, fmincon finds a minimum that violates my constraints
2) With the default options, and with non-linear constraints, fmincon finds a solution that does not violate the constraints withing the default constraint tolerance
If I set the ConstraintTolerance to a very large number say 1e10, I would expect fmincon to behave as if there were no non-linear constraints at all, and it should return a similar answer to #1. However, it is returning a similar answer to #2. IT's as if the ConstraintTolerance is still the default value
What is going on?
1) Without non-linear constrains, fmincon finds a minimum that violates my constraints
2) With the default options, and with non-linear constraints, fmincon finds a solution that does not violate the constraints withing the default constraint tolerance
If I set the ConstraintTolerance to a very large number say 1e10, I would expect fmincon to behave as if there were no non-linear constraints at all, and it should return a similar answer to #1. However, it is returning a similar answer to #2. IT's as if the ConstraintTolerance is still the default value
What is going on?