Line data    Source code

       1             : /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
       2             :    Copyright (c) 2014-2023 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 "tools/SwitchingFunction.h"
      24             : #include "MultiColvarFilter.h"
      25             : 
      26             : //+PLUMEDOC MTRANSFORMS MTRANSFORM_MORE
      27             : /*
      28             : This action can be used to transform the colvar values calculated by a multicolvar using one minus a switching function
      29             : 
      30             : In this action each colvar, \f$s_i\f$, calculated by \ref mcolv is transformed by a \ref switchingfunction function that
      31             : is equal to one if the colvar is greater than a certain target value and which is equal to zero otherwise.
      32             : It is important to understand the distinction between what is done here and what is done by \ref MFILTER_MORE.
      33             : In \ref MFILTER_MORE a weight, \f$w_i\f$ for the colvar is calculated using the \ref histogrambead.  If one calculates the
      34             : MEAN for \ref MFILTER_MORE one is thus calculating:
      35             : 
      36             : \f[
      37             : \mu = \frac{ \sum_i [1 - \sigma(s_i) ] s_i }{\sum_i [1 - \sigma(s_i)] }
      38             : \f]
      39             : 
      40             : where \f$\sigma\f$ is the \ref switchingfunction.  In this action by contrast the colvar is being transformed by the \ref switchingfunction.
      41             : If one thus calculates a MEAN for this action one computes:
      42             : 
      43             : \f[
      44             : \mu = \frac{ \sum_{i=1}^N 1 - \sigma(s_i) }{ N }
      45             : \f]
      46             : 
      47             : In other words, you are calculating the mean for the transformed colvar.
      48             : 
      49             : \par Examples
      50             : 
      51             : The following input gives an example of how a MTRANSFORM_MORE action can be used to duplicate
      52             : functionality that is elsewhere in PLUMED.
      53             : 
      54             : \plumedfile
      55             : DISTANCES ...
      56             :  GROUPA=1-10 GROUPB=11-20
      57             :  LABEL=d1
      58             : ... DISTANCES
      59             : MTRANSFORM_MORE DATA=d1 SWITCH={GAUSSIAN D_0=1.5 R_0=0.00001}
      60             : \endplumedfile
      61             : 
      62             : In this case you can achieve the same result by using:
      63             : 
      64             : \plumedfile
      65             : DISTANCES ...
      66             :  GROUPA=1-10 GROUPB=11-20
      67             :  MORE_THAN={GAUSSIAN D_0=1.5 R_0=0.00001}
      68             : ... DISTANCES
      69             : \endplumedfile
      70             : (see \ref DISTANCES)
      71             : 
      72             : The advantage of MTRANSFORM_MORE comes, however, if you want to use transformed colvars as input
      73             : for \ref MULTICOLVARDENS
      74             : 
      75             : */
      76             : //+ENDPLUMEDOC
      77             : 
      78             : //+PLUMEDOC MFILTERS MFILTER_MORE
      79             : /*
      80             : This action can be used to filter the distribution of colvar values in a multicolvar so that one can compute the mean and so on for only those multicolvars more than a tolerance.
      81             : 
      82             : This action can be used to create a dynamic group of atom based on the value of a multicolvar.
      83             : In this action a multicolvar is within the dynamic group if its value is greater than a target.
      84             : In actuality a weight, \f$w_i\f$  is ascribed to each colvar, \f$s_i\f$ calculated by a multicolvar
      85             : and this weight measures the degree to which a colvar is a member of the group.  This weight is
      86             : calculated using a \ref switchingfunction , \f$\sigma\f$ so it is given by:
      87             : 
      88             : \f[
      89             : w_i = 1 - \sigma(s_i)
      90             : \f]
      91             : 
      92             : If one calculates a function of the set of multicolvars
      93             : these weights are included in the calculation.  As such if one calculates the MEAN, \f$\mu\f$ of a filtered
      94             : multicolvar what is computed is the following:
      95             : 
      96             : \f[
      97             : \mu = \frac{ \sum_i w_i s_i }{ \sum_i w_i}
      98             : \f]
      99             : 
     100             : One is thus calculating the mean for those colvars that are greater than the target.
     101             : 
     102             : \par Examples
     103             : 
     104             : The example shown below calculates the mean for those distances that greater than 1.5 nm in length
     105             : 
     106             : \plumedfile
     107             : DISTANCES GROUPA=1 GROUPB=2-50 MEAN LABEL=d1
     108             : MFILTER_MORE DATA=d1 SWITCH={GAUSSIAN D_0=1.5 R_0=0.00001} MEAN LABEL=d4
     109             : \endplumedfile
     110             : 
     111             : More complicated things can be done by using the label of a filter as input to a new multicolvar as shown
     112             : in the example below.  Here the coordination numbers of all atoms are computed.  The atoms with a coordination
     113             : number greater than 2 are then identified using the filter.  This reduced list of atoms is then used as input
     114             : to a second coordination number calculation.  This second coordination number thus measures the number of
     115             : two-coordinated atoms that each of the two-coordinated atoms is bound to.
     116             : 
     117             : \plumedfile
     118             : c1: COORDINATIONNUMBER SPECIES=1-150 SWITCH={EXP D_0=4.0 R_0=0.5 D_MAX=6.0}
     119             : cf: MFILTER_MORE DATA=c1 SWITCH={RATIONAL D_0=2.0 R_0=0.1} LOWMEM
     120             : c2: COORDINATIONNUMBER SPECIES=cf SWITCH={EXP D_0=4.0 R_0=0.5 D_MAX=6.0} MORE_THAN={RATIONAL D_0=2.0 R_0=0.1}
     121             : \endplumedfile
     122             : 
     123             : */
     124             : //+ENDPLUMEDOC
     125             : 
     126             : namespace PLMD {
     127             : namespace multicolvar {
     128             : 
     129             : class FilterMore : public MultiColvarFilter {
     130             : private:
     131             :   SwitchingFunction sf;
     132             : public:
     133             :   static void registerKeywords( Keywords& keys );
     134             :   explicit FilterMore(const ActionOptions& ao);
     135             :   double applyFilter( const double& val, double& df ) const override;
     136             : };
     137             : 
     138       13797 : PLUMED_REGISTER_ACTION(FilterMore,"MFILTER_MORE")
     139       13787 : PLUMED_REGISTER_ACTION(FilterMore,"MTRANSFORM_MORE")
     140             : 
     141          15 : void FilterMore::registerKeywords( Keywords& keys ) {
     142          15 :   MultiColvarFilter::registerKeywords( keys );
     143          30 :   keys.add("compulsory","NN","6","The n parameter of the switching function ");
     144          30 :   keys.add("compulsory","MM","0","The m parameter of the switching function; 0 implies 2*NN");
     145          30 :   keys.add("compulsory","D_0","0.0","The d_0 parameter of the switching function");
     146          30 :   keys.add("compulsory","R_0","The r_0 parameter of the switching function");
     147          30 :   keys.add("optional","SWITCH","This keyword is used if you want to employ an alternative to the continuous switching function defined above. "
     148             :            "The following provides information on the \\ref switchingfunction that are available. "
     149             :            "When this keyword is present you no longer need the NN, MM, D_0 and R_0 keywords.");
     150          15 : }
     151             : 
     152           7 : FilterMore::FilterMore(const ActionOptions& ao):
     153             :   Action(ao),
     154           7 :   MultiColvarFilter(ao) {
     155             :   // Read in the switching function
     156             :   std::string sw, errors;
     157          14 :   parse("SWITCH",sw);
     158           7 :   if(sw.length()>0) {
     159           7 :     sf.set(sw,errors);
     160           7 :     if( errors.length()!=0 ) {
     161           0 :       error("problem reading SWITCH keyword : " + errors );
     162             :     }
     163             :   } else {
     164           0 :     double r_0=-1.0, d_0;
     165             :     int nn, mm;
     166           0 :     parse("NN",nn);
     167           0 :     parse("MM",mm);
     168           0 :     parse("R_0",r_0);
     169           0 :     parse("D_0",d_0);
     170           0 :     if( r_0<0.0 ) {
     171           0 :       error("you must set a value for R_0");
     172             :     }
     173           0 :     sf.set(nn,mm,r_0,d_0);
     174             :   }
     175           7 :   log.printf("  filtering colvar values and focussing only on those more than %s\n",( sf.description() ).c_str() );
     176             : 
     177           7 :   checkRead();
     178           7 : }
     179             : 
     180       16806 : double FilterMore::applyFilter( const double& val, double& df ) const {
     181       16806 :   double f = 1.0 - sf.calculate( val, df );
     182       16806 :   df*=-val;
     183       16806 :   return f;
     184             : }
     185             : 
     186             : }
     187             : }