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 "SMACOF.h"
      23             : 
      24             : namespace PLMD {
      25             : namespace dimred {
      26             : 
      27           1 : void SMACOF::run( const Matrix<double>& Weights, const Matrix<double>& Distances, const double& tol, const unsigned& maxloops, Matrix<double>& InitialZ ) {
      28             :   unsigned M = Distances.nrows();
      29             : 
      30             :   // Calculate V
      31             :   Matrix<double> V(M,M);
      32             :   double totalWeight=0.;
      33         501 :   for(unsigned i=0; i<M; ++i) {
      34      250500 :     for(unsigned j=0; j<M; ++j) {
      35      250000 :       if(i==j) {
      36         500 :         continue;
      37             :       }
      38      249500 :       V(i,j)=-Weights(i,j);
      39      249500 :       if( j<i ) {
      40      124750 :         totalWeight+=Weights(i,j);
      41             :       }
      42             :     }
      43      250500 :     for(unsigned j=0; j<M; ++j) {
      44      250000 :       if(i==j) {
      45         500 :         continue;
      46             :       }
      47      249500 :       V(i,i)-=V(i,j);
      48             :     }
      49             :   }
      50             : 
      51             :   // And pseudo invert V
      52             :   Matrix<double> mypseudo(M, M);
      53           1 :   pseudoInvert(V, mypseudo);
      54             :   Matrix<double> dists( M, M );
      55           1 :   double myfirstsig = calculateSigma( Weights, Distances, InitialZ, dists ) / totalWeight;
      56             : 
      57             :   // initial sigma is made up of the original distances minus the distances between the projections all squared.
      58             :   Matrix<double> temp( M, M ), BZ( M, M ), newZ( M, InitialZ.ncols() );
      59          14 :   for(unsigned n=0; n<maxloops; ++n) {
      60          14 :     if(n==maxloops-1) {
      61           0 :       plumed_merror("ran out of steps in SMACOF algorithm");
      62             :     }
      63             : 
      64             :     // Recompute BZ matrix
      65        7014 :     for(unsigned i=0; i<M; ++i) {
      66     3507000 :       for(unsigned j=0; j<M; ++j) {
      67     3500000 :         if(i==j) {
      68        7000 :           continue;  //skips over the diagonal elements
      69             :         }
      70             : 
      71     3493000 :         if( dists(i,j)>0 ) {
      72     3493000 :           BZ(i,j) = -Weights(i,j)*Distances(i,j) / dists(i,j);
      73             :         } else {
      74           0 :           BZ(i,j)=0.;
      75             :         }
      76             :       }
      77             :       //the diagonal elements are -off diagonal elements BZ(i,i)-=BZ(i,j)   (Equation 8.25)
      78        7000 :       BZ(i,i)=0; //create the space memory for the diagonal elements which are scalars
      79     3507000 :       for(unsigned j=0; j<M; ++j) {
      80     3500000 :         if(i==j) {
      81        7000 :           continue;
      82             :         }
      83     3493000 :         BZ(i,i)-=BZ(i,j);
      84             :       }
      85             :     }
      86             : 
      87          14 :     mult( mypseudo, BZ, temp);
      88          14 :     mult(temp, InitialZ, newZ);
      89             :     //Compute new sigma
      90          14 :     double newsig = calculateSigma( Weights, Distances, newZ, dists ) / totalWeight;
      91             :     //Computing whether the algorithm has converged (has the mass of the potato changed
      92             :     //when we put it back in the oven!)
      93          14 :     if( std::fabs( newsig - myfirstsig )<tol ) {
      94             :       break;
      95             :     }
      96             :     myfirstsig=newsig;
      97             :     InitialZ = newZ;
      98             :   }
      99           1 : }
     100             : 
     101          15 : double SMACOF::calculateSigma( const Matrix<double>& Weights, const Matrix<double>& Distances, const Matrix<double>& InitialZ, Matrix<double>& dists ) {
     102             :   unsigned M = Distances.nrows();
     103             :   double sigma=0;
     104        7500 :   for(unsigned i=1; i<M; ++i) {
     105     1878735 :     for(unsigned j=0; j<i; ++j) {
     106             :       double dlow=0;
     107     5613750 :       for(unsigned k=0; k<InitialZ.ncols(); ++k) {
     108     3742500 :         double tmp=InitialZ(i,k) - InitialZ(j,k);
     109     3742500 :         dlow+=tmp*tmp;
     110             :       }
     111     1871250 :       dists(i,j)=dists(j,i)=std::sqrt(dlow);
     112     1871250 :       double tmp3 = Distances(i,j) - dists(i,j);
     113     1871250 :       sigma += Weights(i,j)*tmp3*tmp3;
     114             :     }
     115             :   }
     116          15 :   return sigma;
     117             : }
     118             : 
     119             : }
     120             : }