Line data    Source code

       1             : /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
       2             :    Copyright (c) 2015-2023 The plumed team
       3             :    (see the PEOPLE file at the root of the distribution for a list of names)
       4             : 
       5             :    See http://www.plumed.org for more information.
       6             : 
       7             :    This file is part of plumed, version 2.
       8             : 
       9             :    plumed is free software: you can redistribute it and/or modify
      10             :    it under the terms of the GNU Lesser General Public License as published by
      11             :    the Free Software Foundation, either version 3 of the License, or
      12             :    (at your option) any later version.
      13             : 
      14             :    plumed is distributed in the hope that it will be useful,
      15             :    but WITHOUT ANY WARRANTY; without even the implied warranty of
      16             :    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
      17             :    GNU Lesser General Public License for more details.
      18             : 
      19             :    You should have received a copy of the GNU Lesser General Public License
      20             :    along with plumed.  If not, see <http://www.gnu.org/licenses/>.
      21             : +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ */
      22             : #include "core/ActionRegister.h"
      23             : #include "SketchMapBase.h"
      24             : #include "SMACOF.h"
      25             : 
      26             : //+PLUMEDOC DIMRED SKETCHMAP_SMACOF
      27             : /*
      28             : Optimize the sketch-map stress function using the SMACOF algorithm.
      29             : 
      30             : \par Examples
      31             : 
      32             : */
      33             : //+ENDPLUMEDOC
      34             : 
      35             : namespace PLMD {
      36             : namespace dimred {
      37             : 
      38             : class SketchMapSmacof : public SketchMapBase {
      39             : private:
      40             :   unsigned max_smap, maxiter;
      41             :   double smap_tol, iter_tol, regulariser;
      42             :   double recalculateWeights( const Matrix<double>& projections, Matrix<double>& weights );
      43             : public:
      44             :   static void registerKeywords( Keywords& keys );
      45             :   explicit SketchMapSmacof( const ActionOptions& ao );
      46             :   void minimise( Matrix<double>& ) override;
      47             : };
      48             : 
      49       10421 : PLUMED_REGISTER_ACTION(SketchMapSmacof,"SKETCHMAP_SMACOF")
      50             : 
      51           2 : void SketchMapSmacof::registerKeywords( Keywords& keys ) {
      52           2 :   SketchMapBase::registerKeywords( keys );
      53           4 :   keys.add("compulsory","SMACOF_TOL","1E-4","the tolerance for each SMACOF cycle");
      54           4 :   keys.add("compulsory","SMACOF_MAXCYC","1000","maximum number of optimization cycles for SMACOF algorithm");
      55           4 :   keys.add("compulsory","SMAP_TOL","1E-4","the tolerance for sketch-map");
      56           4 :   keys.add("compulsory","SMAP_MAXCYC","100","maximum number of optimization cycles for iterative sketch-map algorithm");
      57           6 :   keys.add("compulsory","REGULARISE_PARAM","0.001","this is used to ensure that we don't divide by zero when updating weights");
      58           2 : }
      59             : 
      60           1 : SketchMapSmacof::SketchMapSmacof( const ActionOptions& ao ):
      61             :   Action(ao),
      62           1 :   SketchMapBase(ao)
      63             : {
      64           1 :   parse("REGULARISE_PARAM",regulariser);
      65           2 :   parse("SMACOF_MAXCYC",max_smap); parse("SMAP_MAXCYC",maxiter);
      66           2 :   parse("SMACOF_TOL",smap_tol); parse("SMAP_TOL",iter_tol);
      67           1 : }
      68             : 
      69           1 : void SketchMapSmacof::minimise( Matrix<double>& projections ) {
      70             :   Matrix<double> weights( distances.nrows(), distances.ncols() ); weights=0.;
      71           1 :   double filt = recalculateWeights( projections, weights );
      72             : 
      73           1 :   for(unsigned i=0; i<maxiter; ++i) {
      74           1 :     SMACOF::run( weights, distances, smap_tol, max_smap, projections );
      75             :     // Recalculate weights matrix and sigma
      76           1 :     double newsig = recalculateWeights( projections, weights );
      77             :     // Test whether or not the algorithm has converged
      78           1 :     if( std::fabs( newsig - filt )<iter_tol ) break;
      79             :     // Make initial sigma into new sigma so that the value of new sigma is used every time so that the error can be reduced
      80             :     filt=newsig;
      81             :   }
      82           1 : }
      83             : 
      84           2 : double SketchMapSmacof::recalculateWeights( const Matrix<double>& projections, Matrix<double>& weights ) {
      85             :   double filt=0, totalWeight=0.;; double dr;
      86        1000 :   for(unsigned i=1; i<weights.nrows(); ++i) {
      87      250498 :     for(unsigned j=0; j<i; ++j) {
      88      249500 :       double ninj=getWeight(i)*getWeight(j); totalWeight += ninj;
      89             : 
      90             :       double tempd=0;
      91      748500 :       for(unsigned k=0; k<projections.ncols(); ++k) {
      92      499000 :         double tmp = projections(i,k) - projections(j,k);
      93      499000 :         tempd += tmp*tmp;
      94             :       }
      95      249500 :       double dij=std::sqrt(tempd);
      96             : 
      97      249500 :       double fij = transformLowDimensionalDistance( dij, dr );
      98      249500 :       double filter=transformed(i,j)-fij;
      99      249500 :       double diff=distances(i,j) - dij;
     100             : 
     101      249500 :       if( std::fabs(diff)<regulariser ) weights(i,j)=weights(j,i)=0.0;
     102      249403 :       else weights(i,j)=weights(j,i) = ninj*( (1-mixparam)*( filter*dr )/diff + mixparam );
     103      249500 :       filt += ninj*( (1-mixparam)*filter*filter + mixparam*diff*diff );
     104             :     }
     105             :   }
     106           2 :   return filt / totalWeight;
     107             : }
     108             : 
     109             : }
     110             : }