Hi,

My optimization problem is NLP (Linear+Conic constraints & quadratic functions in objective function).

The linear model (relaxed) of the original problem gives an objective function value = 951.3748095 (best solution found during preprocessing) using the BARON solver.

However, when I solve the original NLP problem with the BARON solver, it never converges. It gives lower bound=797.715 & upper bound= 956.704180994 (which is also the best solution found during preprocessing). The lower & upper bounds don’t update in subsequent iterations.

We know the relaxed model (Linear in this case) always gives a lower bound to the original problem but then why BARON would report lower bound=797.715 & won’t converge.

IPOPT/IPOPTH solves the original NLP problem with optimal solution=956.704.

MOSEK says ‘The problem contains both conic and nonlinear constraints. (*1501*) The problem & solution status=Unknown’.

MINOS exited with problem infeasibility.

CONOPT = 956.704 ** Feasible solution. The tolerances are minimal and there is no change in objective although the reduced gradient is greater than the tolerance.

CONVERT = No solution returned