# SCIP

Solving Constraint Integer Programs

Public Variable Graph Methods

## Detailed Description

methods to create a variable graph and perform breadth-first search

## Functions

SCIP_EXPORT SCIP_RETCODE SCIPvariablegraphBreadthFirst (SCIP *scip, SCIP_VGRAPH *vargraph, SCIP_VAR **startvars, int nstartvars, int *distances, int maxdistance, int maxvars, int maxbinintvars)

SCIP_EXPORT SCIP_RETCODE SCIPvariableGraphCreate (SCIP *scip, SCIP_VGRAPH **vargraph, SCIP_Bool relaxdenseconss, SCIP_Real relaxdensity, int *nrelaxedconstraints)

SCIP_EXPORT void SCIPvariableGraphFree (SCIP *scip, SCIP_VGRAPH **vargraph)

## Function Documentation

 SCIP_EXPORT SCIP_RETCODE SCIPvariablegraphBreadthFirst ( SCIP * scip, SCIP_VGRAPH * vargraph, SCIP_VAR ** startvars, int nstartvars, int * distances, int maxdistance, int maxvars, int maxbinintvars )

Perform breadth-first (BFS) search on the variable constraint graph.

The result of the algorithm is that the distances array contains the correct distances for every variable from the start variables. The distance of a variable can then be accessed through its problem index (calling SCIPvarGetProbindex()). Hence, The method assumes that the length of distances is at least SCIPgetNVars(). Variables that are not connected through constraints to the start variables have a distance of -1.

Limits can be provided to further restrict the breadth-first search. If a distance limit is given, the search will be performed until the first variable at this distance is popped from the queue, i.e., all variables with a distance < maxdistance have been labeled by the search. If a variable limit is given, the search stops after it completes the distance level at which the limit was reached. Hence, more variables may be actually labeled. The start variables are accounted for those variable limits.

If no variable variable constraint graph is provided, the method will create one and free it at the end This is useful for a single use of the variable constraint graph. For several consecutive uses, it is advised to create a variable constraint graph via SCIPvariableGraphCreate().

Parameters
 scip SCIP data structure vargraph pointer to the variable graph, or NULL to let the function create a local graph startvars array of start variables to calculate distance from nstartvars number of starting variables, at least 1 distances array to keep distance in vargraph from start variables for every variable maxdistance maximum distance >= 0 from start variable (INT_MAX for complete BFS) maxvars maximum number of variables to compute distance for maxbinintvars maximum number of binary or integer variables to compute distance for

Definition at line 1491 of file heur.c.

## ◆ SCIPvariableGraphCreate()

 SCIP_EXPORT SCIP_RETCODE SCIPvariableGraphCreate ( SCIP * scip, SCIP_VGRAPH ** vargraph, SCIP_Bool relaxdenseconss, SCIP_Real relaxdensity, int * nrelaxedconstraints )

initialization method of variable graph data structure

Parameters
 scip SCIP data structure vargraph pointer to the variable graph relaxdenseconss should dense constraints (at least as dense as density) be ignored by connectivity graph? relaxdensity density (with respect to number of variables) to relax constraint from graph nrelaxedconstraints pointer to store the number of constraints that were relaxed, or NULL if not needed

Definition at line 1792 of file heur.c.