JaCoP.constraints.knapsack
Class Tree

java.lang.Object
  JaCoP.constraints.knapsack.Tree

public class Tree
extends java.lang.Object

This tree is a main data structure used and required by Knapsack constraint.

Author:

Radoslaw Szymanek and Wadeck Follonier

Field Summary

int	alreadyObtainedProfit It specifies the already obtained profit due to the items which are already included in the solution.
int	alreadyUsedCapacity It specifies the already used capacity due to the items which are already included in the solution.
int	availableWeightOfCriticalItem It specifies the fraction of the critical item which has been not included in the optimal non-integral solution.
TreeLeaf	criticalLeaf It specifies the leaf containing the critical item.
int	criticalLeftLeaf It specifies the leaf containing the left most item which is being used during computeForbidden().
int	criticalRightLeaf It specifies the leaf containing the last right item which is being used during computeMandatory().
boolean	exhaustedLeftItems It specifies that computeForbidden part of the consistency function has run out of left mandatory items.
boolean	exhaustedRightItems It specifies if the mandatory check has run out of right items to complement mandatory items.
TreeLeaf	first It specifies the first (counting from left to right), the most efficient item in the tree.
TreeLeaf	last It specifies the last (counting from left to right), the least efficient item in the tree.
double	optimalProfit It specifies the optimalProfit of possibly non-integral solution generated by LP relaxation.
TreeNode	root It specifies the root of the tree.
int	takenWeightOfCriticalItem It specifies how much weight is used by an optimal non-fractional solution.

Constructor Summary

Tree(KnapsackItem[] items, java.util.HashMap<IntVar,TreeLeaf> varPositionMaping, TreeLeaf[] leaves, IntVar zero)
It constructs a tree out of the list of items and creates proper supporting structures.

Tree(Tree tree)
It creates a tree by making a shallow copy.

Tree(TreeNode node)
Create a single node tree.

Method Summary

int	computeIntrusionWeight(int weightOfItemChecked, int maxNoOfItems, int profitOfItemChecked, double efficiencyOfItemChecked, double profitSlack) It returns the amount of weight of a given item being checked which can be replaced by Right items given the amount of profitSlack.
int	computeMinProfit(int minWeight) It computes the minimum of capacity variable for knapsack constraint given the minimum requirement for profit.
int	computeMinWeight(int minProfit) It computes the minimum of capacity variable for knapsack constraint given the minimum requirement for profit.
int	computeReplacableWeight(int weightOfItemChecked, int maxNoOfItems, int profitOfItemChecked, double efficiencyOfItemChecked, double profitSlack) It returns the amount of weight of a given item being checked which can be replaced by Right items given the amount of profitSlack.
TreeLeaf	findNextLeafAtLeastOfWeight(TreeLeaf leaf, int weight) It finds next leaf of a maximum weight of at least weight, so it can have some parts of it mandatory.
TreeLeaf	findPreviousLeafAtLeastOfWeight(TreeLeaf leaf, int weight) It finds previous leaf of a maximum weight of at least weight, so it can have some parts of it forbidden.
int	getCriticalPosition(int capacity) It finds a leaf which reaches the limit of the given capacity.
TreeLeaf	getFirst() Used to search for mandatory
TreeLeaf	getLast() It returns the last (the least efficient) item in the tree.
void	initializeComputeForbidden() It initializes the private variables required by computation of how much weight we can replace for any Left item.
void	initializeComputeMandatory() It initializes the private variables required by computation of how much weight we can replace for any Left item.
Tree	merge(Tree that) A merge method for trees, it added a new root from the ancients
void	recompute() It recomputes all the attributes of the internal nodes of the knapsack tree.
java.lang.String	toString()
void	updateCritical(int capacity) It updates information about the critical item, as well as information about fraction of critical item which is not taken.
void	updateFromList(java.util.List<TreeLeaf> list, int startingPosition) Used for updating the tree using a list of nodes that have changed.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

root

public final TreeNode root

It specifies the root of the tree.

criticalLeaf

public TreeLeaf criticalLeaf

It specifies the leaf containing the critical item. The critical item is the one used partially by Linear Programming approach.

criticalRightLeaf

public int criticalRightLeaf

It specifies the leaf containing the last right item which is being used during computeMandatory(). if max() of the right item after criticalRight is changed then it can be safely ignored.

criticalLeftLeaf

public int criticalLeftLeaf

It specifies the leaf containing the left most item which is being used during computeForbidden(). if min() of the left item before criticalLeft is changed then it can be safely ignored.

first

public TreeLeaf first

It specifies the first (counting from left to right), the most efficient item in the tree.

last

public TreeLeaf last

It specifies the last (counting from left to right), the least efficient item in the tree.

optimalProfit

public double optimalProfit

It specifies the optimalProfit of possibly non-integral solution generated by LP relaxation.

availableWeightOfCriticalItem

public int availableWeightOfCriticalItem

It specifies the fraction of the critical item which has been not included in the optimal non-integral solution.

takenWeightOfCriticalItem

public int takenWeightOfCriticalItem

It specifies how much weight is used by an optimal non-fractional solution.

alreadyObtainedProfit

public int alreadyObtainedProfit

It specifies the already obtained profit due to the items which are already included in the solution.

alreadyUsedCapacity

public int alreadyUsedCapacity

It specifies the already used capacity due to the items which are already included in the solution.

exhaustedRightItems

public boolean exhaustedRightItems

It specifies if the mandatory check has run out of right items to complement mandatory items.

exhaustedLeftItems

public boolean exhaustedLeftItems

It specifies that computeForbidden part of the consistency function has run out of left mandatory items.

Constructor Detail

Tree

public Tree(TreeNode node)

Create a single node tree.

Parameters:

node - a root of this one-node tree.

Tree

public Tree(Tree tree)

It creates a tree by making a shallow copy.

Parameters:

tree -

Tree

public Tree(KnapsackItem[] items,
            java.util.HashMap<IntVar,TreeLeaf> varPositionMaping,
            TreeLeaf[] leaves,
            IntVar zero)

It constructs a tree out of the list of items and creates proper supporting structures. It assumes the list of items is greater than 1.

Parameters:

items - knapsack items used to create the tree.

varPositionMaping - mapping of variables into positions within the tree.

leaves - array of leaves of the created tree.

zero - it specifies a variable equal to value 0.

Method Detail

merge

public Tree merge(Tree that)

A merge method for trees, it added a new root from the ancients

Parameters:

that - A tree that is being merged with this tree.

Returns:

The tree resulting of the merge of this and that

updateCritical

public void updateCritical(int capacity)

It updates information about the critical item, as well as information about fraction of critical item which is not taken.

Parameters:

capacity - available capacity to be used by knapsack.

getCriticalPosition

public int getCriticalPosition(int capacity)

It finds a leaf which reaches the limit of the given capacity. Items weight is added from the most efficient to the least efficient.

Parameters:

capacity - available capacity to be used by knapsack.

Returns:

the position of the item.

getFirst

public TreeLeaf getFirst()

Used to search for mandatory

Returns:

The first item

getLast

public TreeLeaf getLast()

It returns the last (the least efficient) item in the tree. It is a starting leaf for computeForbidden() part.

Returns:

The last item

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

updateFromList

public void updateFromList(java.util.List<TreeLeaf> list,
                           int startingPosition)

Used for updating the tree using a list of nodes that have changed.

Parameters:

list - list of leaves that needs to be updated.

startingPosition - it specifies the first leaf in the array which has not been updated before.

recompute

public void recompute()

It recomputes all the attributes of the internal nodes of the knapsack tree.

initializeComputeMandatory

public void initializeComputeMandatory()

It initializes the private variables required by computation of how much weight we can replace for any Left item.

computeReplacableWeight

public int computeReplacableWeight(int weightOfItemChecked,
                                   int maxNoOfItems,
                                   int profitOfItemChecked,
                                   double efficiencyOfItemChecked,
                                   double profitSlack)

It returns the amount of weight of a given item being checked which can be replaced by Right items given the amount of profitSlack.

Parameters:

weightOfItemChecked - the weight of item being checked.

maxNoOfItems - the maximum number which can be taken of checked items.

profitOfItemChecked - the profit of the item being checked.

efficiencyOfItemChecked - the efficiency of the item being checked.

profitSlack - the amount of reserve profit which can be sacrificed before violating the constraint.

Returns:

the amount of weight of a given item that can be replaced by Right items without violating the constraint.

findNextLeafAtLeastOfWeight

public TreeLeaf findNextLeafAtLeastOfWeight(TreeLeaf leaf,
                                            int weight)

It finds next leaf of a maximum weight of at least weight, so it can have some parts of it mandatory.

Parameters:

leaf - starting leaf, the result must be to the right of this leaf.

weight - weight condition which must be satisfied by the found leaf.

Returns:

tree leaf on the right to the supplied leaf with at least specified weight.

initializeComputeForbidden

public void initializeComputeForbidden()

It initializes the private variables required by computation of how much weight we can replace for any Left item.

computeIntrusionWeight

public int computeIntrusionWeight(int weightOfItemChecked,
                                  int maxNoOfItems,
                                  int profitOfItemChecked,
                                  double efficiencyOfItemChecked,
                                  double profitSlack)

It returns the amount of weight of a given item being checked which can be replaced by Right items given the amount of profitSlack.

Parameters:

weightOfItemChecked - the weight of item being checked.

maxNoOfItems - the maximum number which can be taken of checked items.

profitOfItemChecked - the profit of the item being checked.

efficiencyOfItemChecked - the efficiency of the item being checked.

profitSlack - the amount of reserve profit which can be sacrificed before violating the constraint.

Returns:

the amount of weight of a given item that can be replaced by Right items without violating the constraint.

findPreviousLeafAtLeastOfWeight

public TreeLeaf findPreviousLeafAtLeastOfWeight(TreeLeaf leaf,
                                                int weight)

It finds previous leaf of a maximum weight of at least weight, so it can have some parts of it forbidden.

Parameters:

leaf - starting leaf, the result must be to the left of this leaf.

weight - weight condition which must be satisfied by the found leaf.

Returns:

tree leaf on the left to the supplied leaf with at least specified weight.

computeMinWeight

public int computeMinWeight(int minProfit)

It computes the minimum of capacity variable for knapsack constraint given the minimum requirement for profit.

Parameters:

minProfit - minimum profit obtained by the knapsack.

Returns:

it returns the minimum required weight to satisfy min profit requirements.

computeMinProfit

public int computeMinProfit(int minWeight)

It computes the minimum of capacity variable for knapsack constraint given the minimum requirement for profit.

Parameters:

minWeight - - the minimum of weight within a knapsack.

Returns:

it returns the minimum required weight to satisfy min profit requirements.