Line data    Source code

       1             : /* +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
       2             :    Copyright (c) 2016-2023 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/ActionWithValue.h"
      23             : #include "core/ActionWithArguments.h"
      24             : #include "core/ActionPilot.h"
      25             : #include "core/ActionRegister.h"
      26             : #include "core/PlumedMain.h"
      27             : #include "core/ActionSet.h"
      28             : 
      29             : //+PLUMEDOC GRIDCALC ACCUMULATE
      30             : /*
      31             : Sum the elements of this value over the course of the trajectory
      32             : 
      33             : This action is used to sum the outputs from another action over the course of the trajectory.  This is useful
      34             : if you want to calculate the average value that a CV took over the course of a simulation.  As an example, the following
      35             : input can be used to calculate the average distance between atom 1 and atom 2.
      36             : 
      37             : ```plumed
      38             : c: CONSTANT VALUE=1
      39             : d: DISTANCE ATOMS=1,2
      40             : # This adds together the value of the distance on every step
      41             : s: ACCUMULATE ARG=d STRIDE=1
      42             : # This adds one every time we add a new distance to the value s
      43             : n: ACCUMULATE ARG=c STRIDE=1
      44             : # This is thus the average distance
      45             : a: CUSTOM ARG=s,n FUNC=x/y PERIODIC=NO
      46             : # This prints out the average over the whole trajectory (STRIDE=0 means print at end only)
      47             : PRINT ARG=a FILE=average.dat STRIDE=0
      48             : ```
      49             : 
      50             : You can use this action for block averaging by using the `CLEAR` keyword as shown below:
      51             : 
      52             : ```plumed
      53             : c: CONSTANT VALUE=1
      54             : d: DISTANCE ATOMS=1,2
      55             : # This adds together the value of the distance on every step
      56             : s: ACCUMULATE ARG=d STRIDE=1 CLEAR=1000
      57             : # This adds one every time we add a new distance to the value s
      58             : n: ACCUMULATE ARG=c STRIDE=1 CLEAR=1000
      59             : # This is thus the average distance
      60             : a: CUSTOM ARG=s,n FUNC=x/y PERIODIC=NO
      61             : # This prints out the average over the whole trajectory (STRIDE=0 means print at end only)
      62             : PRINT ARG=a FILE=average.dat STRIDE=1000
      63             : ```
      64             : 
      65             : The instructions `CLEAR=1000` in the above input tells PLUMED to set the values `s` and `n` back to
      66             : zero after 1000 new steps have been performed. The PRINT action will thus print a block average that
      67             : is taken from the first 1000 steps of the trajectory, a second block average from the second 1000 steps
      68             : of the trajectory and so on.  Notice that you can achieve a similar effect using UPDATE_FROM and UPDATE_UNTIL as
      69             : shown below:
      70             : 
      71             : ```plumed
      72             : c: CONSTANT VALUE=1
      73             : d: DISTANCE ATOMS=1,2
      74             : s1: ACCUMULATE ARG=d STRIDE=1 UPDATE_UNTIL=1000
      75             : n1: ACCUMULATE ARG=c STRIDE=1 UPDATE_UNTIL=1000
      76             : a1: CUSTOM ARG=s1,n1 FUNC=x/y PERIODIC=NO
      77             : s2: ACCUMULATE ARG=d STRIDE=1 UPDATE_FROM=1000
      78             : n2: ACCUMULATE ARG=c STRIDE=1 UPDATE_FROM=1000
      79             : a2: CUSTOM ARG=s2,n2 FUNC=x/y PERIODIC=NO
      80             : diff: CUSTOM ARG=a1,a2 FUNC=x-y PERIODIC=NO
      81             : PRINT ARG=a1,a2,diff FILE=colver STRIDE=2000
      82             : ```
      83             : 
      84             : This output calculates the average distance for the first 1000 frames of the trajectory and for the second 1000 frames of the trajectory.
      85             : These two averages are then output to a file called colvar as well as the difference between them.
      86             : 
      87             : ## Estimating histograms
      88             : 
      89             : We can also use this action to construct histograms. The example below shows how you can estimate the
      90             : distribution of distances between atoms 1 and 2 that are sampled over the course of the trajectory.
      91             : 
      92             : ```plumed
      93             : c: CONSTANT VALUE=1
      94             : d: DISTANCE ATOMS=1,2
      95             : # Construct the instantaneous histogram from the instantaneous value of the distance
      96             : kde: KDE ARG=d BANDWIDTH=0.05 GRID_MIN=0 GRID_MAX=5 GRID_BIN=250
      97             : # Now add together all the instantaneous histograms
      98             : hist: ACCUMULATE ARG=kde STRIDE=1
      99             : # And normalise the histogram
     100             : n: ACCUMULATE ARG=c STRIDE=1
     101             : a: CUSTOM ARG=hist,n FUNC=x/y PERIODIC=NO
     102             : # And print out the final histogram
     103             : DUMPGRID ARG=a FILE=histo.grid
     104             : ```
     105             : 
     106             : At first glance the fact that we use a [KDE](KDE.md) action to construct an instantaneous histogram from a single
     107             : distance may appear odd.  The reason for doing this, however, is to introduce a clear distinction between
     108             : the syntaxes that are used for spatial and temporal averaging.  To see what I mean consider the following input:
     109             : 
     110             : 
     111             : ```plumed
     112             : c: CONSTANT VALUE=5
     113             : d: DISTANCE ATOMS1=1,2 ATOMS2=3,4 ATOMS3=5,6 ATOMS4=7,8 ATOMS5=9,10
     114             : # Construct the instantaneous histogram from the instantaneous value of the distance
     115             : kde: KDE ARG=d BANDWIDTH=0.05 GRID_MIN=0 GRID_MAX=5 GRID_BIN=250
     116             : # Now add together all the instantaneous histograms
     117             : hist: ACCUMULATE ARG=kde STRIDE=1
     118             : # And normalise the histogram
     119             : n: ACCUMULATE ARG=c STRIDE=1
     120             : a: CUSTOM ARG=hist,n FUNC=x/y PERIODIC=NO
     121             : # And print out the final histogram
     122             : DUMPGRID ARG=a FILE=histo.grid
     123             : ```
     124             : 
     125             : This input computes 5 distances. Kernels corresponding to all five of these distances are added to the instantaneous
     126             : histogram that is constructed using the [KDE](KDE.md) action.  When we call the accumulate action here we are thus
     127             : not simply adding a single kernel to the accumulated grid when we add the elements from `kde`.
     128             : 
     129             : */
     130             : //+ENDPLUMEDOC
     131             : 
     132             : namespace PLMD {
     133             : namespace generic {
     134             : 
     135             : class Accumulate :
     136             :   public ActionWithValue,
     137             :   public ActionWithArguments,
     138             :   public ActionPilot {
     139             : private:
     140             :   bool clearnextstep;
     141             :   unsigned clearstride;
     142             : public:
     143             :   static void registerKeywords( Keywords& keys );
     144             :   Accumulate( const ActionOptions& );
     145             :   unsigned getNumberOfDerivatives() override;
     146        5806 :   bool calculateOnUpdate() override {
     147        5806 :     return false;
     148             :   }
     149         174 :   bool calculateConstantValues( const bool& have_atoms ) override {
     150         174 :     return false;
     151             :   }
     152        2818 :   void calculate() override {}
     153        2818 :   void apply() override {}
     154             :   void update() override ;
     155             : };
     156             : 
     157             : PLUMED_REGISTER_ACTION(Accumulate,"ACCUMULATE")
     158             : 
     159         176 : void Accumulate::registerKeywords( Keywords& keys ) {
     160         176 :   Action::registerKeywords( keys );
     161         176 :   ActionWithValue::registerKeywords( keys );
     162         176 :   ActionWithArguments::registerKeywords( keys );
     163         176 :   ActionPilot::registerKeywords( keys );
     164         176 :   keys.use("UPDATE_FROM");
     165         176 :   keys.use("UPDATE_UNTIL");
     166         176 :   keys.remove("NUMERICAL_DERIVATIVES");
     167         352 :   keys.addInputKeyword("compulsory","ARG","scalar/grid","the label of the argument that is being added to on each timestep");
     168         176 :   keys.add("compulsory","STRIDE","1","the frequency with which the data should be collected and added to the quantity being averaged");
     169         176 :   keys.add("compulsory","CLEAR","0","the frequency with which to clear all the accumulated data.  The default value "
     170             :            "of 0 implies that all the data will be used and that the grid will never be cleared");
     171         352 :   keys.setValueDescription("scalar/grid","a sum calculated from the time series of the input quantity");
     172         176 : }
     173             : 
     174          88 : Accumulate::Accumulate( const ActionOptions& ao ):
     175             :   Action(ao),
     176             :   ActionWithValue(ao),
     177             :   ActionWithArguments(ao),
     178             :   ActionPilot(ao),
     179          88 :   clearnextstep(true) {
     180          88 :   if( getNumberOfArguments()!=1 ) {
     181           0 :     error("there should only be one argument to this action");
     182             :   }
     183          88 :   if( !getPntrToArgument(0)->hasDerivatives() && getPntrToArgument(0)->getRank()!=0 ) {
     184           0 :     error("input to the accumulate action should be a scalar or a grid");
     185             :   }
     186             : 
     187          88 :   if( getStride()==0 ) {
     188           1 :     error("STRIDE should be a positive number of steps");
     189             :   }
     190          87 :   parse("CLEAR",clearstride);
     191          87 :   if( clearstride>0 ) {
     192          11 :     if( clearstride%getStride()!=0 ) {
     193           1 :       error("CLEAR parameter must be a multiple of STRIDE");
     194             :     }
     195          11 :     log.printf("  clearing average every %u steps \n",clearstride);
     196             :   }
     197          87 :   std::vector<std::size_t> shape( getPntrToArgument(0)->getShape() );
     198          87 :   addValueWithDerivatives( shape );
     199          87 :   setNotPeriodic();
     200          87 :   if( getPntrToArgument(0)->isPeriodic() ) {
     201           0 :     error("you cannot accumulate a periodic quantity");
     202             :   }
     203          90 : }
     204             : 
     205          48 : unsigned Accumulate::getNumberOfDerivatives() {
     206          48 :   if( getPntrToArgument(0)->getRank()>0 ) {
     207          48 :     return getPntrToArgument(0)->getNumberOfGridDerivatives();
     208             :   }
     209           0 :   return getPntrToArgument(0)->getNumberOfDerivatives();
     210             : }
     211             : 
     212        2814 : void Accumulate::update() {
     213        2814 :   if( clearnextstep ) {
     214          94 :     if( getPntrToComponent(0)->getNumberOfValues()!=getPntrToArgument(0)->getNumberOfValues() ) {
     215           0 :       getPntrToComponent(0)->setShape( getPntrToArgument(0)->getShape() );
     216             :     }
     217          94 :     clearnextstep=false;
     218          94 :     getPntrToComponent(0)->set(0,0.0);
     219          94 :     getPntrToComponent(0)->clearDerivatives(true);
     220             :   }
     221        2814 :   if( getStep()==0 ) {
     222             :     return;
     223             :   }
     224             : 
     225             :   Value* myarg=getPntrToArgument(0);
     226        2728 :   Value* myout = getPntrToComponent(0);
     227        2728 :   if( getPntrToArgument(0)->getRank()>0 ) {
     228         351 :     unsigned nvals = myarg->getNumberOfValues(), nder = myarg->getNumberOfGridDerivatives();
     229      277070 :     for(unsigned i=0; i<nvals; ++i) {
     230      276719 :       myout->set( i, myout->get(i) + myarg->get(i) );
     231      829866 :       for(unsigned j=0; j<nder; ++j) {
     232      553147 :         myout->addGridDerivatives( i, j, myarg->getGridDerivative( i, j ) );
     233             :       }
     234             :     }
     235             :   } else {
     236        2377 :     getPntrToComponent(0)->add( getPntrToArgument(0)->get() );
     237             :   }
     238             : 
     239             :   // Clear if required
     240        2728 :   if( clearstride>0 && getStep()%clearstride==0 ) {
     241          18 :     clearnextstep=true;
     242             :   }
     243             : }
     244             : 
     245             : }
     246             : }