XSLPsetcbprenode
XSLPsetcbprenode |
Purpose
Set a user callback to be called during MISLP after the set-up of the SLP problem to be solved at a node, but before SLP optimization
Synopsis
int XPRS_CC XSLPsetcbprenode(XSLPprob Prob, int (XPRS_CC *UserFunc) (XSLPprob myProb, void *myObject, int *feas), void *Object);
Arguments
Prob
|
The current SLP problem.
|
UserFunc
|
The function to be called after the set-up of the SLP problem to be solved at a node.
UserFunc returns an integer value. If the return value is nonzero, or if the feasibility flag is set nonzero, then further processing of the node will be terminated (it is declared infeasible).
|
myProb
|
The problem passed to the callback function.
|
myObject
|
The user-defined object passed as
Object to
XSLPsetcbprenode.
|
feas
|
Address of an integer containing the feasibility flag. If
UserFunc sets the flag nonzero, the node is declared infeasible.
|
Object
|
Address of a user-defined object, which can be used for any purpose by the function.
Object is passed to
UserFunc as
myObject.
|
Example
The following example sets up a callback function to be executed at each node before the SLP optimization starts. The array
IntList contains a list of integer variables, and the function prints the bounds on these variables.
int *IntList; XSLPsetcbprenode(Prob, CBPreNode, IntList);
A suitable callback function might resemble the following:
int XPRS_CC CBPreNode(XSLPprob myProb, void *Obj, int *feas) { XPRSprob xprob; int i, *IntList; double LO, UP; IntList = (int *) Obj; XSLPgetptrattrib(myProb, XSLP_XPRSPROBLEM, &xprob); for (i=0; IntList[i]>=0; i++) { XPRSgetlb(xprob,&LO,IntList[i],IntList[i]); XPRSgetub(xprob,&UP,IntList[i],IntList[i]); if (LO > 0 || UP < XPRS_PLUSINFINITY) printf("\nCol %d: %lg <= %lg",LO,UP); } return 0; }
Further information
If a node can be identified as infeasible by the callback function, then the initial optimization at the current node is avoided, as well as further exploration of the node.
This callback must be used in place of XPRSsetcbprenode when optimizing with MISLP.
Related topics