Registration of problem sizes

Functions
int COI_CALLCONV	COIDEF_NumVar (coiHandle_t cntvect, int numvar)
	defines the number of variables in the model.
int COI_CALLCONV	COIDEF_NumCon (coiHandle_t cntvect, int numcon)
	defines the number of constraints in the model.
int COI_CALLCONV	COIDEF_NumNz (coiHandle_t cntvect, int numnz)
	defines the number of nonzero elements in the Jacobian.
int COI_CALLCONV	COIDEF_NumNlNz (coiHandle_t cntvect, int numnlnz)
	defines the Number of Nonlinear Nonzeros.
int COI_CALLCONV	COIDEF_NumHess (coiHandle_t cntvect, int numhess)
	defines the Number of Hessian Nonzeros.
int COI_CALLCONV	COIDEF_OptDir (coiHandle_t cntvect, int optdir)
	defines the Optimization Direction.
int COI_CALLCONV	COIDEF_ObjVar (coiHandle_t cntvect, int objvar)
	defines the Objective Variable.
int COI_CALLCONV	COIDEF_ObjCon (coiHandle_t cntvect, int objcon)
	defines the Objective Constraint.

Detailed Description

Before any model data can be loaded into CONOPT, the dimensions of the problem must be first specified.

This includes the number of variables, the number of constraints, the number of non-zeros, etc. The routines for registering the problem sizes are given in this section. These sizes are used in callbacks when loading the problem matrix and evaluating the first derivatives.

Function Documentation

COIDEF_NumVar()

int COI_CALLCONV COIDEF_NumVar	(	coiHandle_t	cntvect,
		int	numvar )

defines the number of variables in the model.

Attention

Mandatory routine. The number must be positive.

defines the number of variables in the model. The number does not include any slack or artificial variables.

Parameters

cntvect	the control vector
numvar	the number of variables

COIDEF_NumCon()

int COI_CALLCONV COIDEF_NumCon	(	coiHandle_t	cntvect,
		int	numcon )

defines the number of constraints in the model.

Attention

Mandatory routine. The number must be positive.

defines the number of constraints in the model. The number includes the objective function if the objective is defined as an expression (see COIDEF_ObjCon() and COIDEF_ObjVar()).

Parameters

cntvect	the control vector
numcon	the number of constraints

COIDEF_NumNz()

int COI_CALLCONV COIDEF_NumNz	(	coiHandle_t	cntvect,
		int	numnz )

defines the number of nonzero elements in the Jacobian.

Attention

Mandatory routine. The number must be positive.

defines the number of nonzero elements in the Jacobian of the model (the matrix of first derivatives of all constraints with respect to all variables).

Parameters

cntvect	the control vector
numnz	the number of nonzero elements

COIDEF_NumNlNz()

int COI_CALLCONV COIDEF_NumNlNz	(	coiHandle_t	cntvect,
		int	numnlnz )

defines the Number of Nonlinear Nonzeros.

Attention

Mandatory routine.

defines the number of nonlinear nonzeros in the Jacobian. The number is zero if the model is linear and positive if the model is nonlinear.

Parameters

cntvect	the control vector
numnlnz	the number of nonlinear nonzeros

COIDEF_NumHess()

int COI_CALLCONV COIDEF_NumHess	(	coiHandle_t	cntvect,
		int	numhess )

defines the Number of Hessian Nonzeros.

defines the number of nonzeros in the Hessian. The number is zero if the model is linear and positive if the model is nonlinear.

Parameters

cntvect	the control vector
numhess	the number of nonzeros in Hessian

COIDEF_OptDir()

int COI_CALLCONV COIDEF_OptDir	(	coiHandle_t	cntvect,
		int	optdir )

defines the Optimization Direction.

defines the optimization direction. OptDir = +1 defines maximization and OptDir = -1 defines minimization. Setting an optimization direction is optional. If no optimization direction is set, the CONOPT will search for a feasible solution and then stop.

Parameters

cntvect	the control vector
optdir	the optimization direction

COIDEF_ObjVar()

int COI_CALLCONV COIDEF_ObjVar	(	coiHandle_t	cntvect,
		int	objvar )

defines the Objective Variable.

Parameters

cntvect	the control vector
objvar	the index of the objective variable

COIDEF_ObjCon()

int COI_CALLCONV COIDEF_ObjCon	(	coiHandle_t	cntvect,
		int	objcon )

defines the Objective Constraint.

Note

The constraint must be a Free Row, see argument TYPE in ReadMatrix in section ReadMatrix.

If both an objective variable and constraint are set, the last one set will be used in the optimization. You can turn a previously defined objective off by defining variable or constraint 0 as the objective (Fortran notation) or variable or constraint -1 (C notation).

Parameters

cntvect	the control vector
objcon	the index of the objective constraint