Previous Page

|

Next Page

Definition of Return Codes

The return code, which is represented by the output parameter rc in the optimization subroutines, indicates the reason for optimization termination. A positive value indicates successful termination, while a negative value indicates unsuccessful termination. Table 11.1 gives the reason for termination associated with each return code.

Code	Reason for Optimization Termination
1	ABSTOL criterion satisfied (absolute F convergence)
2	ABSFTOL criterion satisfied (absolute F convergence)
3	ABSGTOL criterion satisfied (absolute G convergence)
4	ABSXTOL criterion satisfied (absolute X convergence)
5	FTOL criterion satisfied (relative F convergence)
6	GTOL criterion satisfied (relative G convergence)
7	XTOL criterion satisfied (relative X convergence)
8	FTOL2 criterion satisfied (relative F convergence)
9	GTOL2 criterion satisfied (relative G convergence)
10	linear independent constraints are active at xr and none of them could be released to improve the function value
-1	objective function cannot be evaluated at starting point
-2	derivatives cannot be evaluated at starting point
-3	objective function cannot be evaluated during iteration
-4	derivatives cannot be evaluated during iteration
-5	optimization subroutine cannot improve the function value (this is a very general formulation and is used for various circumstances)
-6	there are problems in dealing with linearly dependent active constraints (changing the LCSING value in the par vector can be helpful)
-7	optimization process stepped outside the feasible region and the algorithm to return inside the feasible region was not successful (changing the LCEPS value in the par vector can be helpful)
-8	either the number of iterations or the number of function calls is larger than the prespecified values in the tc vector (MAXIT and MAXFU)
-9	this return code is temporarily not used (it is used in PROC NLP indicating that more CPU than a prespecified value was used)
-10	a feasible starting point cannot be computed