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 ANALYSIS COLLECT 30 : /* 31 : Collect data from the trajectory for later analysis 32 : 33 : The way this command can be used is illustrated by the following example: 34 : 35 : ```plumed 36 : d: DISTANCE ATOMS=1,2 37 : c: COLLECT ARG=d STRIDE=1 38 : DUMPVECTOR ARG=c FILE=timeseries 39 : ``` 40 : 41 : The COLLECT command in the input above stores the time series of distances over the whole 42 : trajectory. The STRIDE keyword controls how frequently data is stored and the [DUMPVECTOR](DUMPVECTOR.md) 43 : command then outputs the full time series of `d` values at the end of the calculation. 44 : 45 : The above example is not particularly useful as you can achieve the same result by using simply 46 : a DISTANCE command and the [PRINT](PRINT.md) command. The COLLECT command is useful if you want 47 : use the whole trajectory to perform a analysis such as dimensionality reduction (see [dimred](module_dimred.md)). 48 : The shortcut command [COLLECT_FRAMES](COLLECT_FRAMES.md) uses this action heavily and allows one to easily 49 : deal with the fact that COLLECT can only collect the time series for one PLUMED Value at a time. 50 : 51 : ## Collecting a part of the trajectory 52 : 53 : You can use the CLEAR keyword to collect a subset of the trajectory as illustrated below: 54 : 55 : ```plumed 56 : d: DISTANCE ATOMS=1,2 57 : c: COLLECT ARG=d STRIDE=1 CLEAR=1000 58 : DUMPVECTOR ARG=c FILE=timeseries STRIDE=1000 59 : ``` 60 : 61 : This outputs files contain from this input contain 1000-frame blocks of the trajectory for the distance between atom 1 and 2. 62 : This type of input might proove useful if you wish to perform separate analysis on different parts of the trajectory for 63 : later comparison. 64 : 65 : ## Collecting vectors 66 : 67 : You can use the collect command even if the input value has a rank that is greater than zero. For example, the following 68 : input collects vectors of distances from the trajectory: 69 : 70 : ```plumed 71 : d: DISTANCE ATOMS1=1,2 ATOMS2=3,4 ATOMS3=5,6 ATOMS4=7,8 72 : c: COLLECT ARG=d TYPE=vector STRIDE=1 CLEAR=100 73 : DUMPVECTOR ARG=c FILE=timeseries STRIDE=100 74 : ``` 75 : 76 : Notice that if the input to the collect action is a value with rank>0 you __must__ use the TYPE keyword to specify whether 77 : the output Value is a vector or matrix. In the above input we are storing a vector so the DUMPVECTOR command outputs 78 : a list of 400 distances - 4 distances for each frame. The assumption in the input above is that the four distances that have 79 : been computed by the DISTANCES command are indistinguishable. 80 : 81 : By using the following input we can ensure the the four distances are treated as distinguishable when we do any analysis: 82 : 83 : ```plumed 84 : d: DISTANCE ATOMS1=1,2 ATOMS2=3,4 ATOMS3=5,6 ATOMS4=7,8 85 : c: COLLECT ARG=d TYPE=matrix STRIDE=1 CLEAR=100 86 : DUMPVECTOR ARG=c FILE=timeseries STRIDE=100 87 : ``` 88 : 89 : The value `c` is now a $100\times 4$ matrix. Furthermore, when we use DUMPVECTOR to output the output file will contain 90 : four columns of data. 91 : 92 : ## Collecting matrices 93 : 94 : You can also use this action to collect a matrix as illustrated by the following example: 95 : 96 : ```plumed 97 : d: DISTANCE_MATRIX GROUPA=1,2 GROUPB=3,4,5 98 : c: COLLECT ARG=d TYPE=vector STRIDE=1 CLEAR=100 99 : DUMPVECTOR ARG=c FILE=timeseries STRIDE=100 100 : ``` 101 : 102 : The value `c` here is a vector with 600 elements as the input matrix is converted to a vector. These vectors are then stored in 103 : one contiguous object. 104 : 105 : If by contrast we use `TYPE=matrix` as shown below: 106 : 107 : ```plumed 108 : d: DISTANCE_MATRIX GROUPA=1,2 GROUPB=3,4,5 109 : c: COLLECT ARG=d TYPE=matrix STRIDE=1 CLEAR=100 110 : DUMPVECTOR ARG=c FILE=timeseries STRIDE=100 111 : ``` 112 : 113 : A $100 \times 6$ matrix is stored. Each row of this matrix contains a vectorized version of the input matrix. 114 : There is currently no way to store the collected data in a way that recognises that each of the input PLUMED Value 115 : was a matrix. You also cannot use this action to store functions evaluated on a grid. 116 : 117 : */ 118 : //+ENDPLUMEDOC 119 : 120 : namespace PLMD { 121 : namespace generic { 122 : 123 : class Collect : 124 : public ActionWithValue, 125 : public ActionWithArguments, 126 : public ActionPilot { 127 : private: 128 : bool usefirstconf; 129 : unsigned clearstride; 130 : public: 131 : static void registerKeywords( Keywords& keys ); 132 : Collect( const ActionOptions& ); 133 : unsigned getNumberOfDerivatives(); 134 35558 : bool calculateOnUpdate() override { 135 35558 : return false; 136 : } 137 182 : bool calculateConstantValues( const bool& have_atoms ) override { 138 182 : return false; 139 : } 140 17687 : void calculate() override {} 141 17687 : void apply() override {} 142 : void update() override ; 143 : }; 144 : 145 : PLUMED_REGISTER_ACTION(Collect,"COLLECT") 146 : 147 192 : void Collect::registerKeywords( Keywords& keys ) { 148 192 : Action::registerKeywords( keys ); 149 192 : ActionWithValue::registerKeywords( keys ); 150 192 : ActionWithArguments::registerKeywords( keys ); 151 192 : ActionPilot::registerKeywords( keys ); 152 192 : keys.use("UPDATE_FROM"); 153 192 : keys.use("UPDATE_UNTIL"); 154 384 : keys.addInputKeyword("compulsory","ARG","scalar/vector/matrix","the label of the value whose time series is being stored for later analysis"); 155 192 : keys.add("compulsory","STRIDE","1","the frequency with which the data should be collected and added to the quantity being averaged"); 156 192 : keys.add("compulsory","CLEAR","0","the frequency with which to clear all the accumulated data. The default value " 157 : "of 0 implies that all the data will be used and that the grid will never be cleared"); 158 192 : keys.add("compulsory","TYPE","auto","required if you are collecting an object with rank>0. Should be vector/matrix and determines how data is stored. If rank==0 then data has to be stored as a vector"); 159 384 : keys.setValueDescription("vector/matrix","the time series for the input quantity"); 160 192 : } 161 : 162 94 : Collect::Collect( const ActionOptions& ao ): 163 : Action(ao), 164 : ActionWithValue(ao), 165 : ActionWithArguments(ao), 166 : ActionPilot(ao), 167 94 : usefirstconf(false) { 168 94 : if( getNumberOfArguments()!=1 ) { 169 0 : error("there should only be one argument to this action"); 170 : } 171 94 : if( getPntrToArgument(0)->getRank()>0 && getPntrToArgument(0)->hasDerivatives() ) { 172 0 : error("input to the collect argument cannot be a grid"); 173 : } 174 : 175 : std::string type; 176 188 : parse("TYPE",type); 177 165 : if( getPntrToArgument(0)->getNumberOfValues()==1 && (type=="auto" || type=="vector") ) { 178 : type="vector"; 179 24 : } else if( getPntrToArgument(0)->getNumberOfValues()==1 && type=="matrix" ) { 180 0 : error("invalid type specified. Cannot construct a matrix by collecting scalars"); 181 48 : } else if( getPntrToArgument(0)->getNumberOfValues()!=1 && type=="auto" ) { 182 0 : error("missing TYPE keyword. TYPE should specify whether data is to be stored as a vector or a matrix"); 183 36 : } else if( type!="vector" && type!="matrix" ) { 184 0 : error("invalid TYPE specified. Should be matrix/scalar found " + type); 185 : } 186 : 187 94 : if( type=="vector" ) { 188 82 : log.printf(" adding %d elements to stored vector each time we collect\n", getPntrToArgument(0)->getNumberOfValues() ); 189 : } else { 190 12 : log.printf(" constructing matrix with rows of length %d from input data\n", getPntrToArgument(0)->getNumberOfValues() ); 191 : } 192 : 193 94 : parse("CLEAR",clearstride); 194 : unsigned nvals=0; 195 94 : if( clearstride==getStride() ) { 196 : nvals=1; 197 6 : usefirstconf=(getStride()==0); 198 88 : } else if( clearstride>0 ) { 199 15 : if( clearstride%getStride()!=0 ) { 200 0 : error("CLEAR parameter must be a multiple of STRIDE"); 201 : } 202 15 : log.printf(" clearing collected data every %u steps \n",clearstride); 203 15 : nvals=(clearstride/getStride()); 204 : } 205 : 206 94 : std::vector<unsigned> shape(1); 207 94 : shape[0]=nvals; 208 94 : getPntrToArgument(0)->buildDataStore(); 209 94 : if( type=="matrix" ) { 210 12 : shape.resize(2); 211 12 : shape[1] = getPntrToArgument(0)->getNumberOfValues(); 212 : } 213 94 : if( type=="vector" ) { 214 82 : shape[0] = nvals*getPntrToArgument(0)->getNumberOfValues(); 215 : } 216 94 : addValue( shape ); 217 94 : if( shape.size()==2 ) { 218 12 : getPntrToComponent(0)->reshapeMatrixStore( shape[1] ); 219 : } 220 94 : if( getPntrToArgument(0)->isPeriodic() ) { 221 : std::string min, max; 222 16 : getPntrToArgument(0)->getDomain( min, max ); 223 16 : setPeriodic( min, max ); 224 : } else { 225 78 : setNotPeriodic(); 226 : } 227 94 : } 228 : 229 0 : unsigned Collect::getNumberOfDerivatives() { 230 0 : return 0; 231 : } 232 : 233 17687 : void Collect::update() { 234 17687 : if( getStep()==0 || (!onStep() && !usefirstconf) ) { 235 671 : return ; 236 : } 237 17016 : usefirstconf=false; 238 : 239 : Value* myin=getPntrToArgument(0); 240 17016 : Value* myout=getPntrToComponent(0); 241 17016 : unsigned nargs=myin->getNumberOfValues(); 242 17016 : if( clearstride==getStride() ) { 243 339 : for(unsigned i=0; i<nargs; ++i) { 244 333 : myout->set( i, myin->get(i) ); 245 : } 246 17010 : } else if( clearstride>0 ) { 247 1125 : unsigned step = getStep() - clearstride*std::floor( getStep() / clearstride ); 248 1125 : if( getStep()%clearstride==0 ) { 249 25 : step = step + clearstride; 250 : } 251 1125 : unsigned base = (step/getStride()-1)*nargs; 252 2250 : for(unsigned i=0; i<nargs; ++i) { 253 1125 : myout->set( base+i, myin->get(i) ); 254 : } 255 : } else { 256 90800 : for(unsigned i=0; i<nargs; ++i) { 257 74915 : myout->push_back( myin->get(i) ); 258 : } 259 15885 : if( myout->getRank()==2 ) { 260 589 : myout->reshapeMatrixStore( nargs ); 261 : } 262 : } 263 : } 264 : 265 : } 266 : }