edu.umd.cfar.lamp.viper.util
Class DataMatrices

java.lang.Object
  edu.umd.cfar.lamp.viper.util.DataMatrices

public class DataMatrices

extends java.lang.Object

A variety of static methods for manipulating data matrices, including a transpose function and a function for computing the optimal assignment via the Hungarian algorithm. The implementation of the Hungarian algorithm is as Knuth presented in his book of graph algorithms in CWEB.

Author:

davidm

Nested Class Summary

static interface	DataMatrices.GetCost This interface is for functor objects that uses the nodes of a DataMatrix2d as weighted edges of a bipartite graph.
static class	DataMatrices.PassThrough Assumes Object is an instance of Number, and calls its getLong() method to determine the node cost.

Constructor Summary

DataMatrices()

Method Summary

static java.util.List	assign(DataMatrix2d mtx, DataMatrices.GetCost c) Returns the data elements in the data matrix that minimize the total cost of the bipartite graph whose edges are the weighted using the objects in the matrix mtx using the cost function c.
static DataMatrix2d	transpose(DataMatrix2d mtx) Returns a version of the matrix, flipped.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

DataMatrices

public DataMatrices()

Method Detail

transpose

public static DataMatrix2d transpose(DataMatrix2d mtx)

Returns a version of the matrix, flipped.

Parameters:

mtx - the matrix to transpose

Returns:

a copy of the matrix, with m.get(x,y) == transpose(m).get(y,x)

assign

public static java.util.List assign(DataMatrix2d mtx,
                                    DataMatrices.GetCost c)

Returns the data elements in the data matrix that minimize the total cost of the bipartite graph whose edges are the weighted using the objects in the matrix mtx using the cost function c.

Parameters:

mtx - the matrix to be used as the edge list of the bigraph

c - the cost function, called on each element of mtx

Returns:

a list of nodes from mtx, at most one from each column or row, that minimizes the total cost