org.jgrapht.alg.spanning.AhujaOrlinSharmaCapacitatedMinimumSpanningTree Maven / Gradle / Ivy
/*
* (C) Copyright 2018-2021, by Christoph Grüne and Contributors.
*
* JGraphT : a free Java graph-theory library
*
* See the CONTRIBUTORS.md file distributed with this work for additional
* information regarding copyright ownership.
*
* This program and the accompanying materials are made available under the
* terms of the Eclipse Public License 2.0 which is available at
* http://www.eclipse.org/legal/epl-2.0, or the
* GNU Lesser General Public License v2.1 or later
* which is available at
* http://www.gnu.org/licenses/old-licenses/lgpl-2.1-standalone.html.
*
* SPDX-License-Identifier: EPL-2.0 OR LGPL-2.1-or-later
*/
package org.jgrapht.alg.spanning;
import org.jgrapht.*;
import org.jgrapht.alg.cycle.*;
import org.jgrapht.alg.interfaces.*;
import org.jgrapht.alg.util.*;
import org.jgrapht.graph.*;
import org.jgrapht.traverse.*;
import java.util.*;
/**
* Implementation of an algorithm for the capacitated minimum spanning tree problem using a cyclic
* exchange neighborhood, based on Ravindra K. Ahuja, James B. Orlin, Dushyant Sharma, A composite
* very large-scale neighborhood structure for the capacitated minimum spanning tree problem,
* Operations Research Letters, Volume 31, Issue 3, 2003, Pages 185-194, ISSN 0167-6377,
* https://doi.org/10.1016/S0167-6377(02)00236-5.
* (http://www.sciencedirect.com/science/article/pii/S0167637702002365)
*
* A Capacitated Minimum
* Spanning Tree (CMST) is a rooted minimal cost spanning tree that satisfies the capacity
* constrained on all trees that are connected to the designated root. The problem is NP-hard. The
* hard part of the problem is the implicit partition defined by the subtrees. If one can find the
* correct partition, the MSTs can be calculated in polynomial time.
*
* This algorithm is a very large scale neighborhood search algorithm using a cyclic exchange
* neighborhood until a local minimum is found. It makes frequently use of a MST algorithm and the
* algorithm for subset disjoint cycles by Ahuja et al. That is, the algorithm may run in
* exponential time. This algorithm is implemented in two different version: a local search and a
* tabu search. In both cases we have to find the best neighbor of the current capacitated spanning
* tree.
*
* @param the vertex type
* @param the edge type
* @author Christoph Grüne
* @since July 11, 2018
*/
public class AhujaOrlinSharmaCapacitatedMinimumSpanningTree
extends
AbstractCapacitatedMinimumSpanningTree
{
/**
* the maximal length of the cycle in the neighborhood
*/
private final int lengthBound;
/**
* contains whether the best (if true) or the first improvement (if false) is returned in the
* neighborhood search
*/
private final boolean bestImprovement;
/**
* the number of the most profitable operations considered in the GRASP procedure for the
* initial solution.
*/
private final int numberOfOperationsParameter;
/**
* the initial solution
*/
private CapacitatedSpanningTree initialSolution;
/**
* contains whether the local search uses the vertex operation
*/
private final boolean useVertexOperation;
/**
* contains whether the local search uses the subtree operation
*/
private final boolean useSubtreeOperation;
/**
* contains whether a tabu search is used
*/
private final boolean useTabuSearch;
/**
* the tabu time that is the number of iterations an element is in the tabu list
*/
private final int tabuTime;
/**
* the upper limit of non-improving exchanges, this is the stopping criterion in the tabu search
*/
private final int upperLimitTabuExchanges;
/**
* contains whether the algorithm was executed
*/
private boolean isAlgorithmExecuted;
/**
* Constructs a new instance of this algorithm.
*
* @param graph the base graph
* @param root the designated root of the CMST
* @param capacity the edge capacity constraint
* @param demands the demands of the vertices
* @param lengthBound the length bound of the cycle detection algorithm
* @param numberOfOperationsParameter the number of operations that are considered in the
* randomized Esau-Williams algorithm
* {@link EsauWilliamsCapacitatedMinimumSpanningTree} @see
* EsauWilliamsCapacitatedMinimumSpanningTree
*/
public AhujaOrlinSharmaCapacitatedMinimumSpanningTree(
Graph graph, V root, double capacity, Map demands, int lengthBound,
int numberOfOperationsParameter)
{
this(
graph, root, capacity, demands, lengthBound, false, numberOfOperationsParameter, true,
true, true, 10, 50);
}
/**
* Constructs a new instance of this algorithm with the proposed initial solution.
*
* @param initialSolution the initial solution
* @param graph the base graph
* @param root the designated root of the CMST
* @param capacity the edge capacity constraint
* @param demands the demands of the vertices
* @param lengthBound the length bound of the cycle detection algorithm
*/
public AhujaOrlinSharmaCapacitatedMinimumSpanningTree(
CapacitatedSpanningTree initialSolution, Graph graph, V root, double capacity,
Map demands, int lengthBound)
{
this(
initialSolution, graph, root, capacity, demands, lengthBound, false, true, true, true,
10, 50);
}
/**
* Constructs a new instance of this algorithm.
*
* @param graph the base graph
* @param root the designated root of the CMST
* @param capacity the edge capacity constraint
* @param demands the demands of the vertices
* @param lengthBound the length bound of the cycle detection algorithm
* @param bestImprovement contains whether the best (if true) or the first improvement (if
* false) is returned in the neighborhood search
* @param numberOfOperationsParameter the number of operations that are considered in the
* randomized Esau-Williams algorithm
* {@link EsauWilliamsCapacitatedMinimumSpanningTree} @see
* EsauWilliamsCapacitatedMinimumSpanningTree
* @param useVertexOperation contains whether the local search uses the vertex operation
* @param useSubtreeOperation contains whether the local search uses the subtree operation
* @param useTabuSearch contains whether a tabu search is used
* @param tabuTime the tabu time that is the number of iterations an element is in the tabu list
* @param upperLimitTabuExchanges the upper limit of non-improving exchanges, this is the
* stopping criterion in the tabu search
*/
public AhujaOrlinSharmaCapacitatedMinimumSpanningTree(
Graph graph, V root, double capacity, Map demands, int lengthBound,
boolean bestImprovement, int numberOfOperationsParameter, boolean useVertexOperation,
boolean useSubtreeOperation, boolean useTabuSearch, int tabuTime,
int upperLimitTabuExchanges)
{
super(graph, root, capacity, demands);
this.lengthBound = lengthBound;
this.bestImprovement = bestImprovement;
this.numberOfOperationsParameter = numberOfOperationsParameter;
if (!useSubtreeOperation && !useVertexOperation) {
throw new IllegalArgumentException(
"At least one of the options has to be enabled, otherwise it is not possible to excute the local search: useVertexOperation and useSubtreeOperation.");
}
this.useVertexOperation = useVertexOperation;
this.useSubtreeOperation = useSubtreeOperation;
this.useTabuSearch = useTabuSearch;
this.tabuTime = tabuTime;
this.upperLimitTabuExchanges = upperLimitTabuExchanges;
this.isAlgorithmExecuted = false;
}
/**
* Constructs a new instance of this algorithm with the proposed initial solution.
*
* @param initialSolution the initial solution
* @param graph the base graph
* @param root the designated root of the CMST
* @param capacity the edge capacity constraint
* @param demands the demands of the vertices
* @param lengthBound the length bound of the cycle detection algorithm
* @param bestImprovement contains whether the best (if true) or the first improvement (if
* false) is returned in the neighborhood search
* @param useVertexOperation contains whether the local search uses the vertex operation
* @param useSubtreeOperation contains whether the local search uses the subtree operation
* @param useTabuSearch contains whether a tabu search is used
* @param tabuTime the tabu time that is the number of iterations an element is in the tabu list
* @param upperLimitTabuExchanges the upper limit of non-improving exchanges, this is the
* stopping criterion in the tabu search
*/
public AhujaOrlinSharmaCapacitatedMinimumSpanningTree(
CapacitatedSpanningTree initialSolution, Graph graph, V root, double capacity,
Map demands, int lengthBound, boolean bestImprovement,
boolean useVertexOperation, boolean useSubtreeOperation, boolean useTabuSearch,
int tabuTime, int upperLimitTabuExchanges)
{
this(
graph, root, capacity, demands, lengthBound, bestImprovement, 0, useVertexOperation,
useSubtreeOperation, useTabuSearch, tabuTime, upperLimitTabuExchanges);
if (!initialSolution.isCapacitatedSpanningTree(graph, root, capacity, demands)) {
throw new IllegalArgumentException(
"The initial solution is not a valid capacitated spanning tree.");
}
this.initialSolution = initialSolution;
}
@Override
public CapacitatedSpanningTree getCapacitatedSpanningTree()
{
if (isAlgorithmExecuted) {
return bestSolution.calculateResultingSpanningTree();
}
// calculates initial solution on which we base the local search
bestSolution = getInitialSolution();
// map that contains all spanning trees of the current partition
Map> partitionSpanningTrees =
new HashMap<>();
// map that contains the subtrees of all vertices
Map, Double>> subtrees = new HashMap<>();
// set that contains all part of the partition that were affected by an exchange operation
Pair, Set> affected = Pair.of(bestSolution.getLabels(), new HashSet<>());
// the improvement graph
ImprovementGraph improvementGraph = new ImprovementGraph(bestSolution);
// tabu list
Set tabuList = new HashSet<>();
// tabu time list
Map> tabuTimeList = new HashMap<>();
// tabu timer
int tabuTimer = 0;
// number of tabu echanges
int numberOfTabuExchanges = 0;
// the solution int he current iteration
CapacitatedSpanningTreeSolutionRepresentation currentSolution = bestSolution;
// the difference from the current solution and the best solution
double costDifference = 0;
double currentCost;
// do local improvement steps
while (true) {
partitionSpanningTrees = calculateSpanningTrees(
currentSolution, partitionSpanningTrees, affected.getFirst());
if (useSubtreeOperation) {
subtrees = calculateSubtreesOfVertices(
currentSolution, subtrees, partitionSpanningTrees, affected.getFirst());
}
improvementGraph
.updateImprovementGraph(
currentSolution, subtrees, partitionSpanningTrees, affected.getFirst(),
tabuList);
AhujaOrlinSharmaCyclicExchangeLocalAugmentation<
Pair,
DefaultWeightedEdge> ahujaOrlinSharmaCyclicExchangeLocalAugmentation =
new AhujaOrlinSharmaCyclicExchangeLocalAugmentation<>(
improvementGraph.improvementGraph, lengthBound,
improvementGraph.cycleAugmentationLabels, bestImprovement);
GraphWalk, DefaultWeightedEdge> cycle =
ahujaOrlinSharmaCyclicExchangeLocalAugmentation.getLocalAugmentationCycle();
currentCost = cycle.getWeight();
costDifference += currentCost;
if (useTabuSearch) { // do tabu search step
if (currentCost < 0) {
affected = executeNeighborhoodOperation(
currentSolution, improvementGraph.improvementGraphVertexMapping,
improvementGraph.pathExchangeVertexMapping, subtrees, cycle);
if (costDifference < 0) {
bestSolution = currentSolution;
costDifference = 0;
}
} else {
if (upperLimitTabuExchanges <= numberOfTabuExchanges) {
break;
}
// clone solution such that a non-improving exchange does not override a good
// solution
if (currentSolution == bestSolution) {
currentSolution = currentSolution.clone();
}
affected = executeNeighborhoodOperation(
currentSolution, improvementGraph.improvementGraphVertexMapping,
improvementGraph.pathExchangeVertexMapping, subtrees, cycle);
// update tabu list
tabuList.addAll(affected.getSecond());
tabuTimeList.put(tabuTimer, affected.getSecond());
numberOfTabuExchanges++;
}
// update tabu list
Set set = tabuTimeList.remove(tabuTimer - tabuTime - 1);
if (set != null) {
tabuList.removeAll(set);
}
tabuTimer++;
} else { // do normal local search step
if (currentCost < 0) {
affected = executeNeighborhoodOperation(
currentSolution, improvementGraph.improvementGraphVertexMapping,
improvementGraph.pathExchangeVertexMapping, subtrees, cycle);
} else {
break;
}
}
}
this.isAlgorithmExecuted = true;
return bestSolution.calculateResultingSpanningTree();
}
/**
* Calculates an initial solution depending on whether an initial solution was transferred while
* construction of the algorithm. If no initial solution was proposed, the algorithm of
* Esau-Williams is used.
*
* @return an initial solution
*/
private CapacitatedSpanningTreeSolutionRepresentation getInitialSolution()
{
if (initialSolution != null) {
return new CapacitatedSpanningTreeSolutionRepresentation(
initialSolution.getLabels(), initialSolution.getPartition());
}
return new EsauWilliamsCapacitatedMinimumSpanningTree<>(
graph, root, capacity, demands, numberOfOperationsParameter).getSolution();
}
/**
* Executes the move operations induced by the calculated cycle in the improvement graph. It
* returns the set of labels of the subsets that were affected by the move operations.
*
* @param improvementGraphVertexMapping the mapping from the index of the improvement graph
* vertex to the correspondent vertex in the base graph
* @param pathExchangeVertexMapping the mapping from the improvement graph pseudo vertices to
* their subset that they represent
* @param subtrees the map containing the subtree for every vertex
* @param cycle the calculated cycle in the improvement graph
* @return the set of affected labels of subsets that were affected by the move operations
*/
private Pair, Set> executeNeighborhoodOperation(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map improvementGraphVertexMapping,
Map, Integer> pathExchangeVertexMapping,
Map, Double>> subtrees,
GraphWalk, DefaultWeightedEdge> cycle)
{
Set affectedVertices = new HashSet<>();
Set affectedLabels = new HashSet<>();
Iterator> it = cycle.getVertexList().iterator();
if (it.hasNext()) {
Pair cur = it.next();
Integer firstLabel;
switch (cur.getSecond()) {
case SINGLE:
firstLabel =
currentSolution.getLabel(improvementGraphVertexMapping.get(cur.getFirst()));
break;
case SUBTREE:
firstLabel =
currentSolution.getLabel(improvementGraphVertexMapping.get(cur.getFirst()));
break;
default:
firstLabel = -1;
}
while (it.hasNext()) {
Pair next = it.next();
switch (cur.getSecond()) {
/*
* A vertex is moved form the part of cur to the part of next. Therefore, both parts
* are affected. We only consider the label of cur to be affected for now, the label
* of next will be add to the affected set in the next iteration.
*/
case SINGLE: {
V curVertex = improvementGraphVertexMapping.get(cur.getFirst());
Integer curLabel = currentSolution.getLabel(curVertex);
Integer nextLabel;
if (it.hasNext()) {
switch (next.getSecond()) {
case SINGLE:
nextLabel = currentSolution
.getLabel(improvementGraphVertexMapping.get(next.getFirst()));
break;
case SUBTREE:
nextLabel = currentSolution
.getLabel(improvementGraphVertexMapping.get(next.getFirst()));
break;
case PSEUDO:
nextLabel = pathExchangeVertexMapping.get(next);
break;
default:
throw new IllegalStateException(
"This is a bug. There are invalid types of vertices in the cycle.");
}
} else {
nextLabel = firstLabel;
}
affectedVertices.add(curVertex);
affectedLabels.add(curLabel);
currentSolution.moveVertex(curVertex, curLabel, nextLabel);
break;
}
/*
* A subtree is moved from the part of cur to the part of next. Therefore, the part
* of cur is affected.
*/
case SUBTREE: {
V curVertex = improvementGraphVertexMapping.get(cur.getFirst());
Integer curLabel = currentSolution.getLabel(curVertex);
Integer nextLabel;
if (it.hasNext()) {
switch (next.getSecond()) {
case SINGLE:
nextLabel = currentSolution
.getLabel(improvementGraphVertexMapping.get(next.getFirst()));
break;
case SUBTREE:
nextLabel = currentSolution
.getLabel(improvementGraphVertexMapping.get(next.getFirst()));
break;
case PSEUDO:
nextLabel = pathExchangeVertexMapping.get(next);
break;
default:
throw new IllegalStateException(
"This is a bug. There are invalid types of vertices in the cycle.");
}
} else {
nextLabel = firstLabel;
}
affectedVertices.add(curVertex);
affectedLabels.add(curLabel);
// get the whole subtree that has to be moved
Set subtreeToMove = subtrees.get(curVertex).getFirst();
currentSolution.moveVertices(subtreeToMove, curLabel, nextLabel);
break;
}
/*
* cur is the end of a path exchange. Thus, the part of cur is affected because
* vertices were inserted.
*/
case PSEUDO: {
Integer curLabel = pathExchangeVertexMapping.get(cur);
affectedLabels.add(curLabel);
break;
}
/*
* This is the beginning of a path exchange. We have nothing to do.
*/
case ORIGIN: {
break;
}
default:
throw new IllegalStateException(
"This is a bug. There are invalid types of vertices in the cycle.");
}
cur = next;
}
}
/*
* The subsets in the partition may include more than one subtree rooted at root. We create
* a subset for all subtrees rooted at root.
*/
Set moreAffectedLabels = new HashSet<>();
Iterator affectedLabelIterator = affectedLabels.iterator();
while (affectedLabelIterator.hasNext()) {
int label = affectedLabelIterator.next();
Set vertexSubset = currentSolution.getPartitionSet(label);
if (vertexSubset.isEmpty()) {
affectedLabelIterator.remove();
} else {
moreAffectedLabels
.addAll(currentSolution.partitionSubtreesOfSubset(vertexSubset, label));
}
}
affectedLabels.addAll(moreAffectedLabels);
// clean up the partition such that only current subsets are represented
currentSolution.cleanUp();
return Pair.of(affectedLabels, affectedVertices);
}
/**
* Updates the map containing the MSTs for every subset of the partition.
*
* @param partitionSpanningTrees the map containing the MST for every subset of the partition
* @param affectedLabels the labels of the subsets of the partition that were changed due to the
* multi-exchange
* @return the updated map containing the MST for every subset of the partition
*/
private Map> calculateSpanningTrees(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map> partitionSpanningTrees,
Set affectedLabels)
{
for (Integer label : affectedLabels) {
Set set = currentSolution.getPartitionSet(label);
currentSolution.getPartitionSet(label).add(root);
partitionSpanningTrees
.put(
label,
new PrimMinimumSpanningTree<>(new AsSubgraph<>(graph, set)).getSpanningTree());
currentSolution.getPartitionSet(label).remove(root);
}
return partitionSpanningTrees;
}
/**
* Updates the map containing the subtrees of all vertices in the graph with respect to the MST
* in the partition and returns them in map.
*
* @param subtrees the subtree map to update
* @param partitionSpanningTree the map containing the MST for every subset of the partition
* @param affectedLabels the labels of the subsets of the partition that were changed due to the
* multi-exchange
* @return the updated map of vertices to their subtrees
*/
private Map, Double>> calculateSubtreesOfVertices(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map, Double>> subtrees,
Map> partitionSpanningTree,
Set affectedLabels)
{
for (Integer label : affectedLabels) {
Set modifiableSet = new HashSet<>(currentSolution.getPartitionSet(label));
modifiableSet.add(root);
for (V v : currentSolution.getPartitionSet(label)) {
Pair, Double> currentSubtree =
subtree(currentSolution, modifiableSet, v, partitionSpanningTree);
subtrees.put(v, currentSubtree);
}
}
return subtrees;
}
/**
* Calculates the subtree of v with respect to the MST given in
* partitionSpanningTree.
*
* @param v the vertex to calculate the subtree for
* @param partitionSpanningTree the map from labels to spanning trees of the partition.
* @return the subtree of v with respect to the MST given in
* partitionSpanningTree.
*/
private Pair, Double> subtree(
CapacitatedSpanningTreeSolutionRepresentation currentSolution, Set modifiableSet, V v,
Map> partitionSpanningTree)
{
/*
* initializes graph that is the MST of the current subset rooted
*/
SpanningTreeAlgorithm.SpanningTree partSpanningTree =
partitionSpanningTree.get(currentSolution.getLabel(v));
Graph spanningTree =
new AsSubgraph<>(graph, modifiableSet, partSpanningTree.getEdges());
/*
* calculate subtree rooted at v
*/
Set subtree = new HashSet<>();
double subtreeWeight = 0;
Iterator depthFirstIterator = new DepthFirstIterator<>(spanningTree, v);
Set currentPath = new HashSet<>();
double currentWeight = 0;
boolean storeCurrentPath = true;
while (depthFirstIterator.hasNext()) {
V next = depthFirstIterator.next();
if (spanningTree.containsEdge(next, v)) {
storeCurrentPath = true;
subtree.addAll(currentPath);
subtreeWeight += currentWeight;
currentPath = new HashSet<>();
currentWeight = 0;
}
/*
* This part of the subtree is connected to the root, thus, this particular tree is not
* part of the subtree of the current vertex v.
*/
if (next.equals(root)) {
storeCurrentPath = false;
currentPath = new HashSet<>();
currentWeight = 0;
}
if (storeCurrentPath) {
currentPath.add(next);
currentWeight += demands.get(next);
}
}
return Pair.of(subtree, subtreeWeight);
}
/**
* This enums contains the vertex types of the improvement graph.
*/
private enum ImprovementGraphVertexType
{
SINGLE,
SUBTREE,
PSEUDO,
ORIGIN
}
/**
* This class realises the improvement graph for the composite multi-exchange large neighborhood
* search. The improvement graph encodes two exchange classes: - cyclic exchange (on vertices
* and subtrees) - path exchange (on vertices and subtrees)
*
* DEFINITION EXCHANGES Let T[i] be the subtree rooted at i of the MST implicitly defined by the
* vertex partition. Cyclic Exchange: A cyclic exchange is defined on vertices i_1, ..., i_r,
* i_1, where the vertices represent either itself in the base graph or the subtrees rooted at
* i_k for k = 1, ..., r, where T[i_a] != T[i_b] for a != b. The cyclic exchange on i_1, ...,
* i_r, i_1 moves the i_a (or T[i_a]) to the subset of i_b, where b = a+1 mod r+1. Such a cyclic
* exchange is feasible if the capacity constraint is not violated. We can represent the cost of
* the cyclic exchange by the following formulas: Let S[i_k] be the subset of i_k in the
* implicitly defined partition. - exchange of vertices: $$ c(T_new) - c(T) = \sum_{a = 1}^{r}
* c(\{i_{a - 1}\} \cup S[i_{i_a}] \setminus \{i_a\}] $$ - exchange of rooted subtrees: $$
* c(T_new) - c(T) = \sum_{a = 1}^{r} c(T[i_{a - 1}] \cup S[i_{i_a}] \setminus T[i_a]] $$ where
* c is the given edge cost function and T_new is the CMST resulting by executing the cyclic
* exchange. Thus, an exchange is profitable if c(T_new) - c(T) < 0.
*
* Path Exchange: A path exchange follows the same idea as the cyclic exchange but it does not
* end at the same vertex. That is, the path exchange is defined on i_1, ..., i_r. The cost
* function has to be adapted at the start and end point of the path.
*
* DEFINITION NEIGHBORHOOD Furthermore, we have to define the neighborhood. These are all
* capacitated spanning trees that are reachable by using such an exchange as given above.
*
* DEFINITION IMPROVEMENT GRAPH The improvement graph is based on a feasible capacitated
* spanning tree and uses a one-to-one correspondence between the vertices in the base graph and
* the vertices in the improvement graph. We want to define the arc set of the improvement graph
* such that each subset disjoint directed cycle (see construction) correspond to a cyclic
* exchange (or a path exchange, we come to that later). Furthermore, the cost of the cycle in
* the improvement graph and the cost of the corresponding cyclic exchange has to be equal.
*
* CONSTRUCTION OF THE IMPROVEMENT GRAPH The improvement graph IG = (V, A) has the vertex set V,
* which is equal to the vertex set of the base graph. The arc set A is defined in the
* following: A directed arc (i, j) in IG represents that we move the node i (or the subtree
* T[i]) to the subset in which vertex j is. That is, vertex i and j are removed from their
* subset and i (or the subtree T[i]) is moved to the subset of j. This arc only exists if the
* exchange is feasible. Then, the cost can be defined as $$ c(\{T[i]\} \cup S[j] \setminus
* \{T[j]\}) - c(S[j]). $$ A directed cycle i_1, ..., i_r, i_1 in this graph subset disjoint if
* the subsets of the nodes are pairwise disjoint. By this definition, there is a one-to-one
* cost-preserving correspondence between the cyclic exchanges and the subset disjoint directed
* cycles in the improvement graph IG.
*
* Identifying path exchanges: For the conversion of path exchanges into subset disjoint cycles,
* we have to introduce two more node types in the improvement graph: pseudo nodes and a origin
* node. On the one hand, pseudo nodes represent a subset of the implicitly defined partition
* and mark the end of the end of a path exchange. On the other hand, the origin node marks a
* beginning of a path exchange. Therefore, the pseudo node are connected to the origin node to
* induce subset disjoint cycles. The costs of the arcs from and to the pseudo nodes and the
* origin nodes are defined as follows: We denoted the original nodes in the improvement graph
* as regular nodes - c(p, o) = 0 for all pseudo nodes p and origin node o - c(o, r) = c(S[j]
* \setminus \{T[j]\}) - c(S[j]) for origin node o and for all regular nodes r - c(r, p) =
* c(\{T[i]\} \cup S[j]) - c(S[j]) for all regular nodes r and for all pseudo nodes p Again,
* those arc exists only if the exchange is feasible.
*
* IDENTIFYING SUBSET DISJOINT CYCLES This is done via a heuristic which can be found here
* {@link AhujaOrlinSharmaCyclicExchangeLocalAugmentation} @see
* AhujaOrlinSharmaCyclicExchangeLocalAugmentation.
*/
private class ImprovementGraph
{
/**
* the improvement graph itself
*/
Graph, DefaultWeightedEdge> improvementGraph;
/**
* the current solution corresponding to the improvement graph
*/
CapacitatedSpanningTreeSolutionRepresentation capacitatedSpanningTreeSolutionRepresentation;
/**
* mapping form all improvement graph vertices to their labels corresponding to the base
* graph for the CMST problem
*/
Map, Integer> cycleAugmentationLabels;
/**
* mapping from the vertex index in the improvement graph to the vertex in the base graph
*/
Map improvementGraphVertexMapping;
/**
* mapping from the base graph vertex to the vertex index in the improvement graph
*/
Map initialVertexMapping;
/**
* mapping from the label of the subsets to the corresponding vertex mapping
*/
Map> pseudoVertexMapping;
/**
* mapping from the pseudo vertices to the label of the subset they are representing
*/
Map, Integer> pathExchangeVertexMapping;
/**
* the origin vertex
*/
Pair origin;
/**
* dummy label of the origin vertex
*/
final Integer originVertexLabel = -1;
/**
* Constructs an new improvement graph object for this CMST algorithm instance.
*/
public ImprovementGraph(
CapacitatedSpanningTreeSolutionRepresentation capacitatedSpanningTreeSolutionRepresentation)
{
this.capacitatedSpanningTreeSolutionRepresentation =
capacitatedSpanningTreeSolutionRepresentation;
this.improvementGraphVertexMapping = new HashMap<>();
this.initialVertexMapping = new HashMap<>();
this.pseudoVertexMapping = new HashMap<>();
this.pathExchangeVertexMapping = new HashMap<>();
/*
* We initialize this map such that it can be used in the subset-disjoint cycle
* detection algorithm. This map redirects the getters to the corresponding maps in this
* improvement graph such that it realises the correct functionality.
*/
this.cycleAugmentationLabels = getImprovementGraphLabelMap();
this.improvementGraph = createImprovementGraph();
}
/**
* Initializes the improvement graph, i.e. adds single, subtree and pseudo vertices as well
* as the origin vertex. Furthermore, it initializes all mappings.
*
* @return the improvement graph itself.
*/
public Graph,
DefaultWeightedEdge> createImprovementGraph()
{
Graph, DefaultWeightedEdge> improvementGraph =
new DefaultDirectedWeightedGraph<>(DefaultWeightedEdge.class);
int counter = 0;
for (V v : graph.vertexSet()) {
if (v.equals(root)) {
continue;
}
if (useVertexOperation) {
Pair singleVertex =
new Pair<>(counter, ImprovementGraphVertexType.SINGLE);
improvementGraph.addVertex(singleVertex);
}
if (useSubtreeOperation) {
Pair subtreeVertex =
new Pair<>(counter, ImprovementGraphVertexType.SUBTREE);
improvementGraph.addVertex(subtreeVertex);
}
// we have to add these only once
improvementGraphVertexMapping.put(counter, v);
initialVertexMapping.put(v, counter);
counter++;
}
Pair origin =
new Pair<>(counter, ImprovementGraphVertexType.ORIGIN);
improvementGraph.addVertex(origin);
this.origin = origin;
pathExchangeVertexMapping.put(origin, originVertexLabel);
for (Integer label : capacitatedSpanningTreeSolutionRepresentation.getLabels()) {
Pair pseudoVertex =
new Pair<>(origin.getFirst() + label + 1, ImprovementGraphVertexType.PSEUDO);
pseudoVertexMapping.put(label, pseudoVertex);
pathExchangeVertexMapping.put(pseudoVertex, label);
improvementGraph.addVertex(pseudoVertex);
}
/*
* connection of pseudo nodes and origin node
*/
for (Pair v : pseudoVertexMapping.values()) {
improvementGraph.setEdgeWeight(improvementGraph.addEdge(v, origin), 0);
}
return improvementGraph;
}
/**
* Updates the improvement graph. It updates the vertices and edges in the parts specified
* in labelsToUpdate.
*
* @param currentSolution the current solution
* @param subtrees the mapping from vertices to their subtree
* @param partitionSpanningTrees the mapping from labels of subsets to their spanning tree
* @param labelsToUpdate the labels of all subsets that has to be updated (because of the
* multi-exchange operation)
*/
public void updateImprovementGraph(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map, Double>> subtrees,
Map> partitionSpanningTrees,
Set labelsToUpdate, Set tabuList)
{
this.capacitatedSpanningTreeSolutionRepresentation = currentSolution;
this.cycleAugmentationLabels = getImprovementGraphLabelMap();
updatePseudoNodesOfNewLabels(currentSolution);
for (V v1 : graph.vertexSet()) {
if (v1.equals(root)) {
continue;
}
Pair vertexOfV1Single =
Pair.of(initialVertexMapping.get(v1), ImprovementGraphVertexType.SINGLE);
Pair vertexOfV1Subtree =
Pair.of(initialVertexMapping.get(v1), ImprovementGraphVertexType.SUBTREE);
if (updateTabuVertices(tabuList, v1, vertexOfV1Single, vertexOfV1Subtree)) {
continue;
}
updateOriginNodeConnections(
currentSolution, subtrees, partitionSpanningTrees, labelsToUpdate, v1,
vertexOfV1Single, vertexOfV1Subtree);
/*
* update the connections to regular nodes and pseudo nodes
*/
for (Integer label : currentSolution.getLabels()) {
/*
* only update if there is a change induced by a changed part. This potentially
* saves a lot of time.
*/
if (label.equals(currentSolution.getLabel(v1))
|| (!labelsToUpdate.contains(currentSolution.getLabel(v1))
&& !labelsToUpdate.contains(label)))
{
continue;
}
Pair pseudoVertex =
pseudoVertexMapping.get(label);
Set modifiableSet = new HashSet<>(currentSolution.getPartitionSet(label));
// add root to the set for MST calculations
modifiableSet.add(root);
double oldWeight = partitionSpanningTrees.get(label).getWeight();
updateSingleNode(
currentSolution, subtrees, tabuList, label, oldWeight, modifiableSet,
pseudoVertex, v1, vertexOfV1Single);
updateSubtreeNode(
currentSolution, subtrees, tabuList, label, oldWeight, modifiableSet,
pseudoVertex, v1, vertexOfV1Subtree);
}
}
}
/**
* Updates the pseudo nodes corresponding to new subsets in the partition. That is, new
* pseudo nodes for new labels in the label set are added and pseudo nodes of labels that
* are no more in the label set are removed.
*
* @param currentSolution the current solution in the iteration
*/
private void updatePseudoNodesOfNewLabels(
CapacitatedSpanningTreeSolutionRepresentation currentSolution)
{
if (!currentSolution.getLabels().equals(pseudoVertexMapping.keySet())) {
for (Integer label : currentSolution.getLabels()) {
if (!pseudoVertexMapping.keySet().contains(label)) {
Pair pseudoVertex = new Pair<>(
origin.getFirst() + label + 1, ImprovementGraphVertexType.PSEUDO);
pseudoVertexMapping.put(label, pseudoVertex);
pathExchangeVertexMapping.put(pseudoVertex, label);
improvementGraph.addVertex(pseudoVertex);
DefaultWeightedEdge newEdge =
improvementGraph.addEdge(pseudoVertex, origin);
improvementGraph.setEdgeWeight(newEdge, 0);
}
}
if (currentSolution.getLabels().size() != pseudoVertexMapping.keySet().size()) {
Iterator labelIterator = pseudoVertexMapping.keySet().iterator();
while (labelIterator.hasNext()) {
int label = labelIterator.next();
if (!currentSolution.getLabels().contains(label)) {
Pair pseudoVertex = new Pair<>(
origin.getFirst() + label + 1, ImprovementGraphVertexType.PSEUDO);
labelIterator.remove();
pathExchangeVertexMapping.remove(pseudoVertex);
improvementGraph.removeVertex(pseudoVertex);
}
}
}
}
}
/**
* Updates all nodes that correspond to v1 and returns if the vertex
* v1. That is, all incident edges of v1 are removed if
* v1 is in the tabu list.
*
* @param tabuList the tabu list of the current iteration
* @param v1 the vertex to update the nodes in the improvement graph for
* @param vertexOfV1Single the node in the improvement graph representing the exchange of
* the vertex v1
* @param vertexOfV1Subtree the node in the improvement graph representing the exchange of
* the subtree rooted at v1
* @return true iff v1 is in the tabu list
*/
private boolean updateTabuVertices(
Set tabuList, V v1, Pair vertexOfV1Single,
Pair vertexOfV1Subtree)
{
if (tabuList.contains(v1)) {
// remove all edges from the vertex
if (useVertexOperation) {
improvementGraph.removeVertex(vertexOfV1Single);
improvementGraph.addVertex(vertexOfV1Single);
}
if (useSubtreeOperation) {
improvementGraph.removeVertex(vertexOfV1Subtree);
improvementGraph.addVertex(vertexOfV1Subtree);
}
return true;
}
return false;
}
/**
* Updates the edges to the origin vertex.
*
* @param currentSolution the current solution in the iteration
* @param subtrees the mapping from vertices to their subtree
* @param partitionSpanningTrees the mapping from labels of subsets to their spanning tree
* @param labelsToUpdate the labels of all subsets that has to be updated (because of the
* multi-exchange operation)
* @param v1 the vertex to update the nodes in the improvement graph for
* @param vertexOfV1Single the node in the improvement graph representing the exchange of
* the vertex v1
* @param vertexOfV1Subtree the node in the improvement graph representing the exchange of
* the subtree rooted at v1
*/
private void updateOriginNodeConnections(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map, Double>> subtrees,
Map> partitionSpanningTrees,
Set labelsToUpdate, V v1,
Pair vertexOfV1Single,
Pair vertexOfV1Subtree)
{
double newWeight, oldWeight;
SpanningTreeAlgorithm.SpanningTree spanningTree;
/*
* update connections to origin node
*/
if (labelsToUpdate.contains(currentSolution.getLabel(v1))) {
oldWeight = partitionSpanningTrees.get(currentSolution.getLabel(v1)).getWeight();
/*
* edge for v1 vertex remove operation
*/
Set partitionSetOfV1 =
currentSolution.getPartitionSet(currentSolution.getLabel(v1));
partitionSetOfV1.add(root);
if (useVertexOperation) {
partitionSetOfV1.remove(v1);
spanningTree =
new PrimMinimumSpanningTree<>(new AsSubgraph<>(graph, partitionSetOfV1))
.getSpanningTree();
if (spanningTree.getEdges().size() == partitionSetOfV1.size() - 1) {
newWeight = spanningTree.getWeight();
} else {
newWeight = Double.NaN;
}
updateImprovementGraphEdge(origin, vertexOfV1Single, 0, newWeight - oldWeight);
partitionSetOfV1.add(v1);
}
/*
* edge for v1 subtree remove operation If the subtree of v1 contains only the
* vertex itself, it is the same operation as removing v1 as vertex. Thus, do not
* add edges.
*/
if (useSubtreeOperation) {
if (subtrees.get(v1).getFirst().size() > 1 || !useVertexOperation) {
partitionSetOfV1.removeAll(subtrees.get(v1).getFirst());
spanningTree =
new PrimMinimumSpanningTree<>(new AsSubgraph<>(graph, partitionSetOfV1))
.getSpanningTree();
if (spanningTree.getEdges().size() == partitionSetOfV1.size() - 1) {
newWeight = spanningTree.getWeight();
} else {
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
origin, vertexOfV1Subtree, 0, newWeight - oldWeight);
partitionSetOfV1.addAll(subtrees.get(v1).getFirst());
} else {
improvementGraph.removeVertex(vertexOfV1Subtree);
improvementGraph.addVertex(vertexOfV1Subtree);
}
}
partitionSetOfV1.remove(root);
}
}
/**
* Updates all edges from vertexOfV1Single to nodes in the subset represented
* by label.
*
* @param currentSolution the current solution in the iteration
* @param subtrees the mapping from vertices to their subtree
* @param tabuList the tabu list of the current iteration
* @param label the current label to update the edges for
* @param oldWeight the old weight of the subset
* @param modifiableSet a modifiable version of the subset of nodes represented by label
* inclusive the root node
* @param pseudoVertex the pseudo vertex representing the subset represented by label
* @param v1 the vertex to update the nodes in the improvement graph for
* @param vertexOfV1Single the node in the improvement graph representing the exchange of
* the vertex v1
*/
private void updateSingleNode(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map, Double>> subtrees, Set tabuList, int label, double oldWeight,
Set modifiableSet, Pair pseudoVertex, V v1,
Pair vertexOfV1Single)
{
double newCapacity, newWeight;
SpanningTreeAlgorithm.SpanningTree spanningTree;
// add v1 to the set for MST calculations
modifiableSet.add(v1);
/*
* Adding of edges for v1 vertex replacing an object in v2. We need to considers this
* only if vertex operations should be used.
*/
if (useVertexOperation) {
for (V v2 : currentSolution.getPartitionSet(label)) {
if (v2.equals(root)) {
throw new IllegalStateException(
"The root is in the partition. This is a bug.");
}
if (tabuList.contains(v2)) {
continue;
}
/*
* edge for v1 vertex replacing v2 vertex
*/
modifiableSet.remove(v2);
spanningTree = new PrimMinimumSpanningTree<>(
new AsSubgraph<>(graph, modifiableSet, graph.edgeSet())).getSpanningTree();
if (spanningTree.getEdges().size() == modifiableSet.size() - 1) {
newCapacity = calculateMaximumDemandOfSubtrees(
modifiableSet, spanningTree, currentSolution.getPartitionWeight(label)
+ demands.get(v1) - demands.get(v2));
newWeight = spanningTree.getWeight();
} else {
newCapacity = Double.NaN;
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
vertexOfV1Single,
Pair.of(initialVertexMapping.get(v2), ImprovementGraphVertexType.SINGLE),
newCapacity, newWeight - oldWeight);
modifiableSet.add(v2);
// end edge for v1 vertex replacing v2 vertex
/*
* edge for v1 vertex replacing v2 subtree If the subtree of v2 contains only
* the vertex itself and both operations are used, it is the same operation as
* moving v2 as vertex. Thus, do not add edges.
*/
if (useSubtreeOperation) {
if (subtrees.get(v2).getFirst().size() > 1) {
modifiableSet.removeAll(subtrees.get(v2).getFirst());
spanningTree = new PrimMinimumSpanningTree<>(
new AsSubgraph<>(graph, modifiableSet, graph.edgeSet()))
.getSpanningTree();
if (spanningTree.getEdges().size() == modifiableSet.size() - 1) {
newCapacity = calculateMaximumDemandOfSubtrees(
modifiableSet, spanningTree,
currentSolution.getPartitionWeight(label) + demands.get(v1)
- subtrees.get(v2).getSecond());
newWeight = spanningTree.getWeight();
} else {
newCapacity = Double.NaN;
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
vertexOfV1Single,
Pair
.of(
initialVertexMapping.get(v2),
ImprovementGraphVertexType.SUBTREE),
newCapacity, newWeight - oldWeight);
modifiableSet.addAll(subtrees.get(v2).getFirst());
}
}
// end edge for v1 vertex replacing v2 subtree
}
/*
* edge for v1 vertex replacing no object
*/
spanningTree = new PrimMinimumSpanningTree<>(
new AsSubgraph<>(graph, modifiableSet, graph.edgeSet())).getSpanningTree();
if (spanningTree.getEdges().size() == modifiableSet.size() - 1) {
newCapacity = calculateMaximumDemandOfSubtrees(
modifiableSet, spanningTree,
currentSolution.getPartitionWeight(label) + demands.get(v1));
newWeight = spanningTree.getWeight();
} else {
newCapacity = Double.NaN;
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
vertexOfV1Single, pseudoVertex, newCapacity, newWeight - oldWeight);
// end edge for v1 vertex replacing no object
// remove v1 from the set
modifiableSet.remove(v1);
}
}
/**
* Updates all edges from vertexOfV1Single to nodes in the subset represented
* by label. This method does adds the subtree of v1 to
* modifiableSet.
*
* @param currentSolution the current solution in the iteration
* @param subtrees the mapping from vertices to their subtree
* @param tabuList the tabu list of the current iteration
* @param label the current label to update the edges for
* @param oldWeight the old weight of the subset
* @param modifiableSet a modifiable version of the subset of nodes represented by label
* @param pseudoVertex the pseudo vertex representing the subset represented by label
* @param v1 the vertex to update the nodes in the improvement graph for
* @param vertexOfV1Subtree the node in the improvement graph representing the exchange of
* the subtree rooted at v1
*/
private void updateSubtreeNode(
CapacitatedSpanningTreeSolutionRepresentation currentSolution,
Map, Double>> subtrees, Set tabuList, int label, double oldWeight,
Set modifiableSet, Pair pseudoVertex, V v1,
Pair vertexOfV1Subtree)
{
double newCapacity, newWeight;
SpanningTreeAlgorithm.SpanningTree spanningTree;
/*
* Adding of edges for v1 subtree replacing an object in v2. We need to considers this
* only if subtree operations should be used.
*
* If the subtree of v1 contains only the vertex itself and both operations are used, it
* is the same operation as moving v1 as vertex. Thus, do not add edges.
*/
if (useSubtreeOperation
&& (subtrees.get(v1).getFirst().size() > 1 || !useVertexOperation))
{
// add the subtree of v1 to the set for MST calculations
modifiableSet.addAll(subtrees.get(v1).getFirst());
for (V v2 : currentSolution.getPartitionSet(label)) {
if (v2.equals(root)) {
throw new IllegalStateException(
"The root is in the partition. This is a bug.");
}
if (tabuList.contains(v2)) {
continue;
}
/*
* edge for v1 subtree replacing v2 vertex
*/
if (useVertexOperation) {
modifiableSet.remove(v2);
spanningTree = new PrimMinimumSpanningTree<>(
new AsSubgraph<>(graph, modifiableSet, graph.edgeSet()))
.getSpanningTree();
if (spanningTree.getEdges().size() == modifiableSet.size() - 1) {
newCapacity = calculateMaximumDemandOfSubtrees(
modifiableSet, spanningTree,
currentSolution.getPartitionWeight(label)
+ subtrees.get(v1).getSecond() - demands.get(v2));
newWeight = spanningTree.getWeight();
} else {
newCapacity = Double.NaN;
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
vertexOfV1Subtree,
Pair
.of(
initialVertexMapping.get(v2),
ImprovementGraphVertexType.SINGLE),
newCapacity, newWeight - oldWeight);
modifiableSet.add(v2);
}
// end edge for v1 subtree replacing v2 vertex
/*
* edge for v1 subtree replacing v2 subtree
*/
modifiableSet.removeAll(subtrees.get(v2).getFirst());
spanningTree = new PrimMinimumSpanningTree<>(
new AsSubgraph<>(graph, modifiableSet, graph.edgeSet())).getSpanningTree();
if (spanningTree.getEdges().size() == modifiableSet.size() - 1) {
newCapacity = calculateMaximumDemandOfSubtrees(
modifiableSet, spanningTree,
currentSolution.getPartitionWeight(currentSolution.getLabel(v2))
+ subtrees.get(v1).getSecond() - subtrees.get(v2).getSecond());
newWeight = spanningTree.getWeight();
} else {
newCapacity = Double.NaN;
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
vertexOfV1Subtree,
Pair.of(initialVertexMapping.get(v2), ImprovementGraphVertexType.SUBTREE),
newCapacity, newWeight - oldWeight);
modifiableSet.addAll(subtrees.get(v2).getFirst());
// end edge for v1 subtree replacing v2 subtree
}
/*
* edge for v1 subtree replacing no object
*/
spanningTree = new PrimMinimumSpanningTree<>(
new AsSubgraph<>(graph, modifiableSet, graph.edgeSet())).getSpanningTree();
if (spanningTree.getEdges().size() == modifiableSet.size() - 1) {
newCapacity = calculateMaximumDemandOfSubtrees(
modifiableSet, spanningTree,
currentSolution.getPartitionWeight(label) + subtrees.get(v1).getSecond());
newWeight = spanningTree.getWeight();
} else {
newCapacity = Double.NaN;
newWeight = Double.NaN;
}
updateImprovementGraphEdge(
vertexOfV1Subtree, pseudoVertex, newCapacity, newWeight - oldWeight);
// end edge for v1 subtree replacing no object
}
}
/**
* Adds an edge between v1 and v2 to the improvement graph if
* newCapacity does not exceed the capacity constraint. The weight of the edge
* is newCost.
*
* @param v1 start vertex (the vertex or subtree induced by v1 that will be
* moved to the subset of v2)
* @param v2 end vertex (the vertex or subtree induced by v2 that will be
* removed from the subset of v2)
* @param newCapacity the used capacity by adding the vertex or subtree induced by
* v1 to the subset of v2 and deleting the vertex or
* subtree induced by v2
* @param newCost the cost of the edge (the cost induced by the operation induced by
* v1 and v2)
*/
public void updateImprovementGraphEdge(
Pair v1,
Pair v2, double newCapacity, double newCost)
{
if (!Double.isNaN(newCapacity) && newCapacity <= capacity && !Double.isNaN(newCost)) {
DefaultWeightedEdge edge;
edge = improvementGraph.getEdge(v1, v2);
if (edge == null) {
edge = improvementGraph.addEdge(v1, v2);
}
improvementGraph.setEdgeWeight(edge, newCost);
} else {
improvementGraph.removeEdge(v1, v2);
}
}
/**
* Calculates the maximum demand over all new subtrees induced by the minimum spanning tree
* spanningTree. A spanning tree induces more than one subset in the partition
* if the root vertex of the base graph connects more than one subtree of the spanning tree.
*
* @param vertexSubset the vertex subset spanning Tree is defined on
* @param spanningTree the spanning tree
* @param totalDemand the total demand of the whole spanning tree
* @return the maximum demand over all new subtrees induced by the minimum spanning tree
* spanningTree
*/
public double calculateMaximumDemandOfSubtrees(
Set vertexSubset, SpanningTreeAlgorithm.SpanningTree spanningTree,
double totalDemand)
{
Graph spanningTreeGraph =
new AsSubgraph<>(graph, vertexSubset, spanningTree.getEdges());
/*
* The subtree does not evolve to more than 1 partition subsets, thus, we can return the
* total demand.
*/
int degreeOfRoot = spanningTreeGraph.degreeOf(root);
if (degreeOfRoot == 1) {
return totalDemand;
}
double maximumDemand = 0;
DepthFirstIterator depthFirstIterator =
new DepthFirstIterator<>(spanningTreeGraph, root);
if (depthFirstIterator.hasNext()) {
depthFirstIterator.next();
}
int numberOfRootEdgesExplored = 0;
double exploredVerticesDemand = 0;
double currentDemand = 0;
while (depthFirstIterator.hasNext()) {
V next = depthFirstIterator.next();
// exploring new subtree
if (spanningTreeGraph.containsEdge(root, next)) {
exploredVerticesDemand += currentDemand;
if (maximumDemand < currentDemand) {
maximumDemand = currentDemand;
}
// we can stop the exploration
if (maximumDemand >= 0.5 * totalDemand
|| exploredVerticesDemand + maximumDemand >= totalDemand)
{
return maximumDemand;
}
// we can stop the exploration, all subtrees but one are explored
if (numberOfRootEdgesExplored + 1 == degreeOfRoot) {
return Math.max(maximumDemand, totalDemand - exploredVerticesDemand);
}
numberOfRootEdgesExplored++;
currentDemand = 0;
}
currentDemand += demands.get(next);
}
return maximumDemand;
}
/**
* Returns the mapping that is used in the valid cycle detection algorithm, i.e. the vertex
* label map.
*
* @return the vertex label map used in the valid cycle detection algorithm
*/
private Map,
Integer> getImprovementGraphLabelMap()
{
return new AbstractMap, Integer>()
{
@Override
public int size()
{
return improvementGraphVertexMapping.size() + pathExchangeVertexMapping.size()
+ (origin == null ? 0 : 1);
}
@Override
public boolean isEmpty()
{
return improvementGraphVertexMapping.isEmpty()
&& pathExchangeVertexMapping.isEmpty() && origin == null;
}
@Override
public boolean containsKey(Object key)
{
if (key instanceof Pair) {
return improvementGraphVertexMapping.containsKey(((Pair) key).getFirst())
|| pathExchangeVertexMapping.containsKey(key) || key.equals(origin);
}
return false;
}
@Override
public boolean containsValue(Object value)
{
return improvementGraphVertexMapping.containsValue(value)
|| pathExchangeVertexMapping.containsValue(value)
|| value.equals(originVertexLabel);
}
@Override
public Integer get(Object key)
{
if (key instanceof Pair) {
if (improvementGraphVertexMapping.containsKey(((Pair) key).getFirst())) {
return capacitatedSpanningTreeSolutionRepresentation
.getLabel(
improvementGraphVertexMapping.get(((Pair) key).getFirst()));
}
if (key.equals(origin)) {
return originVertexLabel;
}
}
return pathExchangeVertexMapping.get(key);
}
@Override
public Integer put(Pair key, Integer value)
{
throw new IllegalStateException();
}
@Override
public Integer remove(Object key)
{
throw new IllegalStateException();
}
@Override
public void putAll(
Map, ? extends Integer> m)
{
throw new IllegalStateException();
}
@Override
public void clear()
{
throw new IllegalStateException();
}
@Override
public Set> keySet()
{
Set> keySet = new HashSet<>();
for (Integer i : improvementGraphVertexMapping.keySet()) {
if (useVertexOperation) {
keySet.add(Pair.of(i, ImprovementGraphVertexType.SINGLE));
}
if (useSubtreeOperation) {
keySet.add(Pair.of(i, ImprovementGraphVertexType.SUBTREE));
}
}
keySet.addAll(pathExchangeVertexMapping.keySet());
keySet.add(origin);
return keySet;
}
@Override
public Collection values()
{
return capacitatedSpanningTreeSolutionRepresentation.getLabels();
}
@Override
public Set, Integer>> entrySet()
{
Set, Integer>> entrySet =
new HashSet<>();
for (Integer i : improvementGraphVertexMapping.keySet()) {
Integer label = capacitatedSpanningTreeSolutionRepresentation
.getLabel(improvementGraphVertexMapping.get(i));
if (useVertexOperation) {
entrySet
.add(
new AbstractMap.SimpleEntry<>(
Pair.of(i, ImprovementGraphVertexType.SINGLE), label));
}
if (useSubtreeOperation) {
entrySet
.add(
new AbstractMap.SimpleEntry<>(
Pair.of(i, ImprovementGraphVertexType.SUBTREE), label));
}
}
for (Pair pseudoVertex : pathExchangeVertexMapping
.keySet())
{
entrySet
.add(
new AbstractMap.SimpleEntry<>(
pseudoVertex, pathExchangeVertexMapping.get(pseudoVertex)));
}
entrySet.add(new AbstractMap.SimpleEntry<>(origin, originVertexLabel));
return entrySet;
}
};
}
}
}