vtkPAutoCorrelativeStatistics.cxx

/*=========================================================================

  Program:   Visualization Toolkit
  Module:    vtkPAutoCorrelativeStatistics.cxx

  Copyright (c) Ken Martin, Will Schroeder, Bill Lorensen
  All rights reserved.
  See Copyright.txt or http://www.kitware.com/Copyright.htm for details.

     This software is distributed WITHOUT ANY WARRANTY; without even
     the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
     PURPOSE.  See the above copyright notice for more information.

=========================================================================*/
#include "vtkToolkits.h"

#include "vtkPAutoCorrelativeStatistics.h"

#include "vtkCommunicator.h"
#include "vtkInformation.h"
#include "vtkInformationVector.h"
#include "vtkObjectFactory.h"
#include "vtkMultiBlockDataSet.h"
#include "vtkMultiProcessController.h"
#include "vtkTable.h"
#include "vtkVariant.h"

vtkStandardNewMacro(vtkPAutoCorrelativeStatistics);
vtkCxxSetObjectMacro(vtkPAutoCorrelativeStatistics, Controller, vtkMultiProcessController);
//-----------------------------------------------------------------------------
vtkPAutoCorrelativeStatistics::vtkPAutoCorrelativeStatistics()
{
  this->Controller = 0;
  this->SetController( vtkMultiProcessController::GetGlobalController() );
}

//-----------------------------------------------------------------------------
vtkPAutoCorrelativeStatistics::~vtkPAutoCorrelativeStatistics()
{
  this->SetController( 0 );
}

//-----------------------------------------------------------------------------
void vtkPAutoCorrelativeStatistics::PrintSelf( ostream& os, vtkIndent indent )
{
  this->Superclass::PrintSelf(os, indent);
  os << indent << "Controller: " << this->Controller << endl;
}

// ----------------------------------------------------------------------
void vtkPAutoCorrelativeStatistics::Learn( vtkTable* inData,
                                           vtkTable* inParameters,
                                           vtkMultiBlockDataSet* outMeta )
{
  if ( ! outMeta )
    {
    return;
    }

  // First calculate correlative statistics on local data set
  this->Superclass::Learn( inData, inParameters, outMeta );

  vtkTable* primaryTab = vtkTable::SafeDownCast( outMeta->GetBlock( 0 ) );
  if ( ! primaryTab )
    {
    return;
    }

  vtkIdType nRow = primaryTab->GetNumberOfRows();
  if ( ! nRow )
    {
    // No statistics were calculated.
    return;
    }

  // Make sure that parallel updates are needed, otherwise leave it at that.
  int np = this->Controller->GetNumberOfProcesses();
  if ( np < 2 )
    {
    return;
    }

  // Now get ready for parallel calculations
  vtkCommunicator* com = this->Controller->GetCommunicator();
  if ( ! com )
    {
    vtkErrorMacro("No parallel communicator.");
    }

  // (All) gather all sample sizes
  int n_l = primaryTab->GetValueByName( 0, "Cardinality" ).ToInt(); // Cardinality
  int* n_g = new int[np];
  com->AllGather( &n_l, n_g, 1 );

  // Iterate over all parameter rows
  for ( int r = 0; r < nRow; ++ r )
    {
    // (All) gather all local M statistics
    double M_l[5];
    M_l[0] = primaryTab->GetValueByName( r, "Mean Xs" ).ToDouble();
    M_l[1] = primaryTab->GetValueByName( r, "Mean Xt" ).ToDouble();
    M_l[2] = primaryTab->GetValueByName( r, "M2 Xs" ).ToDouble();
    M_l[3] = primaryTab->GetValueByName( r, "M2 Xt" ).ToDouble();
    M_l[4] = primaryTab->GetValueByName( r, "M XsXt" ).ToDouble();
    double* M_g = new double[5 * np];
    com->AllGather( M_l, M_g, 5 );

    // Aggregate all local quintuples of M statistics into global ones
    int ns = n_g[0];
    double meanXs = M_g[0];
    double meanXt = M_g[1];
    double mom2Xs = M_g[2];
    double mom2Xt = M_g[3];
    double momXsXt = M_g[4];

    for ( int i = 1; i < np; ++ i )
      {
      int ns_l = n_g[i];
      int N = ns + ns_l;

      int o = 5 * i;
      double meanXs_part = M_g[o];
      double meanXt_part = M_g[o + 1];
      double mom2Xs_part = M_g[o + 2];
      double mom2Xt_part = M_g[o + 3];
      double momXsXt_part = M_g[o + 4];

      double invN = 1. / static_cast<double>( N );

      double deltaXs = meanXs_part - meanXs;
      double deltaXs_sur_N = deltaXs * invN;

      double deltaXt = meanXt_part - meanXt;
      double deltaXt_sur_N = deltaXt * invN;

      int prod_ns = ns * ns_l;

      mom2Xs += mom2Xs_part
        + prod_ns * deltaXs * deltaXs_sur_N;

      mom2Xt += mom2Xt_part
        + prod_ns * deltaXt * deltaXt_sur_N;

      momXsXt += momXsXt_part
        + prod_ns * deltaXs * deltaXt_sur_N;

      meanXs += ns_l * deltaXs_sur_N;

      meanXt += ns_l * deltaXt_sur_N;

      ns = N;
      }

    primaryTab->SetValueByName( r, "Mean Xs", meanXs );
    primaryTab->SetValueByName( r, "Mean Xt", meanXt );
    primaryTab->SetValueByName( r, "M2 Xs", mom2Xs );
    primaryTab->SetValueByName( r, "M2 Xt", mom2Xt );
    primaryTab->SetValueByName( r, "M XsXt", momXsXt );

    // Set global statistics
    primaryTab->SetValueByName( r, "Cardinality", ns );

    // Clean-up
    delete [] M_g;
    }
  delete [] n_g;
}

// ----------------------------------------------------------------------
void vtkPAutoCorrelativeStatistics::Test( vtkTable* inData,
                                          vtkMultiBlockDataSet* inMeta,
                                          vtkTable* outMeta )
{
  if ( this->Controller->GetNumberOfProcesses() > 1 )
    {
    vtkWarningMacro( "Parallel correlative statistics: Hypothesis testing not implemented for more than 1 process." );
    return;
    }

  this->Superclass::Test( inData, inMeta, outMeta );
}