wGmdh.jGmdh.hybrid
Class SseRegressionEquations

java.lang.Object
  extended by wGmdh.jGmdh.hybrid.SseRegressionEquations

public class SseRegressionEquations
extends java.lang.Object


Constructor Summary
SseRegressionEquations()
           
 
Method Summary
static java.util.List<java.util.List<java.lang.Integer>> combinations(java.util.List<java.lang.Integer> set, int order)
          Generates combinations without repetition of elements that belong to set
static java.util.List<java.util.List<java.lang.Integer>> combinationsRepeating(java.util.List<java.lang.Integer> set, int order)
          Generates combinations with repetition of elements that belong to set
static LinearEqSystem contractSystem(LinearEqSystem system, java.util.List<java.lang.Integer> toRemove)
          Provided with a model, this method gives matrix and vectors related to regression of a smaller model, whose links are a subset of original model's links.
static LinearEqSystem extendSystemByOne(LinearEqSystem system, double[][] inputs, double[] regressTo)
           
static java.util.List<java.util.List<java.util.List<java.lang.Integer>>> partitionCombinations(java.util.List<java.lang.Integer> list)
           
static int[] partitioningToCoordinates(java.util.List<java.util.List<java.lang.Integer>> partition, int nrVariables)
           
static java.util.List<java.lang.Integer> removeCR2List(int setSize, java.util.List<java.lang.Integer> n)
          Generates a list of all locations that need to be erased to have all the combinations that contain n-th element from a list of 1-combinations and 2-combinations removed.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Constructor Detail

SseRegressionEquations

public SseRegressionEquations()
Method Detail

combinationsRepeating

public static java.util.List<java.util.List<java.lang.Integer>> combinationsRepeating(java.util.List<java.lang.Integer> set,
                                                                                      int order)
                                                                               throws TooBig
Generates combinations with repetition of elements that belong to set

Parameters:
set -
order -
Returns:
Throws:
TooBig

combinations

public static java.util.List<java.util.List<java.lang.Integer>> combinations(java.util.List<java.lang.Integer> set,
                                                                             int order)
                                                                      throws TooBig
Generates combinations without repetition of elements that belong to set

Parameters:
set -
order -
Returns:
Throws:
TooBig

partitionCombinations

public static java.util.List<java.util.List<java.util.List<java.lang.Integer>>> partitionCombinations(java.util.List<java.lang.Integer> list)

removeCR2List

public static java.util.List<java.lang.Integer> removeCR2List(int setSize,
                                                              java.util.List<java.lang.Integer> n)
                                                       throws TooBig
Generates a list of all locations that need to be erased to have all the combinations that contain n-th element from a list of 1-combinations and 2-combinations removed.

Parameters:
setSize -
n -
Returns:
Throws:
TooBig

contractSystem

public static LinearEqSystem contractSystem(LinearEqSystem system,
                                            java.util.List<java.lang.Integer> toRemove)
Provided with a model, this method gives matrix and vectors related to regression of a smaller model, whose links are a subset of original model's links. The matrix and vector get deep copied.

Parameters:
system - system of linear regression equations for a polynomial network
toRemove - list of rows/clos we want to exclude from the original model. Suggested use: with removeCR2Array or removeCR2List
Returns:
contracted system of linear equations. does not return the vector of unknowns

extendSystemByOne

public static LinearEqSystem extendSystemByOne(LinearEqSystem system,
                                               double[][] inputs,
                                               double[] regressTo)
                                        throws TooBig
Parameters:
system - old system that will get extended
inputs - input values in points: inputs[0][] to inputs[inputs.length-1][] are related to old system; double[inputs.length-1][] are the additional input values
regressTo -
Returns:
Throws:
TooBig

partitioningToCoordinates

public static int[] partitioningToCoordinates(java.util.List<java.util.List<java.lang.Integer>> partition,
                                              int nrVariables)