The Sequential Quadratic Programming Solver |
Solver Termination Messages
The SQP solver terminates with one of the following messages:
-
Converged
- The SQP solver has found a local minimum within the convergence tolerance specified.
-
Maximum number of iterations reached
- The SQP solver has reached the limit on the maximum number of iterations, but has not satisfied the convergence criteria.
-
Maximum specified time reached
- The SQP solver has spent more time than the prespecified maximum real time for the optimization process, but has not satisfied the convergence criteria.
-
Line search cannot improve further
- The SQP solver cannot make any progress in the line search. One possibility is that a solution has been found, but the convergence tolerance is set too small.
-
Out of memory
- The NLP problem is too large for the SQP solver to solve. This could be due to a limitation in the memory allocated to the procedure.
-
Objective function cannot be evaluated at starting point
- The objective function cannot be evaluated at the starting point.
-
At least one constraint cannot be evaluated at starting point
- At least one of the constraints cannot be evaluated at the starting point.
-
Gradient of objective function cannot be evaluated at starting point
- The gradient of the objective function cannot be evaluated at the starting point.
-
Gradient of at least one constraint cannot be evaluated at starting point
- The gradient of at least one of the constraints cannot be evaluated at the starting point.
-
Did not converge
- The NLP problem has not been solved. There could be several reasons why this happens. For instance, maybe a QP subproblem could not be solved within a preset number of iterations.
-
Second order optimality is not satisfied
- SQP has located a point satisfying the first-order condition but not the second-order condition.
-
Problem may be unbounded
- SQP has located a feasible point, but its objective function value is extremely low. SQP can improve the objective function even further as well.
-
Problem may be infeasible
- SQP cannot locate a feasible point.
-
Problem is infeasible
- SQP has identified an infeasibility in the problem.
-
Problem solved by the OPTMODEL presolver
- The problem was solved by the OPTMODEL presolver.
Copyright © 2008 by SAS Institute Inc., Cary, NC, USA. All rights reserved.