/usr/include/ThePEG/Utilities/CompSelector.h is in libthepeg-dev 1.8.0-1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 | // -*- C++ -*-
//
// CompSelector.h is a part of ThePEG - Toolkit for HEP Event Generation
// Copyright (C) 1999-2011 Leif Lonnblad
//
// ThePEG is licenced under version 2 of the GPL, see COPYING for details.
// Please respect the MCnet academic guidelines, see GUIDELINES for details.
//
#ifndef THEPEG_CompSelector_H
#define THEPEG_CompSelector_H
//
// This is the declaration of the CompSelector class.
//
#include "ThePEG/Utilities/Selector.h"
namespace ThePEG {
/**
* The CompSelector class works like the Selector class in that it can
* be used to randomly select objects according to associated
* probabilities. In addition, the CompSelector class is able to
* handle the case where the associated probabilities are
* overestimates and the selected object will be discarded according
* to some weight. If then a weight above one is encountered, this
* means that the overestimated probability for the selected object
* was wrong and it should in fact have been higher. If this happens,
* the CompSelecteor will go into compensation mode, which means that
* the selected object will be oversampled a period after the
* violation to compensate for having been undersampled before. Also
* the associated probability is adjusted to reflect the new
* overestimate.
*
* The available functions are not as many as in Selector, and some of
* the works somewhat differently. Before starting sampling the
* objects should be added to a CompSelector object with the insert()
* function. To selct an object the select() function should be
* used. After that the weight with which the object should be
* accepted should be presented with the reweight() function which
* normally returns zero. If, however, the weight is larger than unity
* the new overestimated probability is returned and the CompSelector
* enters the compensating mode. Note that the weight is passed as a
* reference and may be changed in by the reweight function if in the
* compensating mode.
*/
template <typename T, typename WeightType = double>
class CompSelector {
public:
/** @name Standard constructors and destructors. */
//@{
/**
* The default constructor. The optional argument gives the margin
* used to get a new overestimated probability for an object when
* entering compensation mode.
*/
CompSelector(double newMargin = 1.1, double newTolerance = 1.0e-6)
: N(0), last(), theMargin(newMargin), theTolerance(newTolerance) {}
//@}
public:
/** @name The main function controlling the selection and compensation. */
//@{
/**
* Insert an object given a probability for this object. If the
* probability is zero or negative, the object will not be inserted
* and the probability itself is returned. Otherwise the sum of
* probabilities so far is returned. Note that if selection has
* already started and this CompSelector is in compensating mode, it
* will immediately leave this mode and the selection procedure will
* start from scratch.
*/
WeightType insert(WeightType d, const T & t) {
reset();
return selector.insert(d, t);
}
/**
* Selct an object randomly. Given a random number generator which
* generates flat random numbers in the interval ]0,1[ with the
* <code>operator()()</code> function, select an object according to
* the individual probabilities specified when they were
* inserted. If the generated number is outside the allowed range or
* the Selector is empty, a range_error will be thrown. The
* generator should have a push_back function which will be used
* push back a uniform random number in the interval ]0,1[
* calculated from the fraction of rnd which was in the range of the
* selected object.
*/
template <typename RNDGEN>
T & select(RNDGEN & rnd) throw(range_error) {
++N;
if ( !compensating() ) last = selector.select(rnd);
return last;
}
/**
* Report the weight associated with the last selected
* object. Returns the zero if weight was below unity, otherwise the
* compensation mode will be entered and the new overestimated
* probabilty for the last selected object will be returned.
*/
WeightType reweight(double & weight) {
if ( abs(weight) > 1.0 + tolerance() ) {
// Retrieve the old overestimate of the object by seing how much
// the summed weights are decreased when removing the object.
WeightType oldtot = selector.sum();
WeightType oldmax = oldtot - selector.erase(last);
WeightType newmax = oldmax*abs(weight)*margin();
WeightType newtot = selector.insert(newmax, last);
double rat = newmax/oldmax;
// Setup the new compensation level.
Level level;
level.weight = 1.0/rat;
level.lastN = long(N*newtot/oldtot);
// If we are already compensating, reweight the previous
// compensation levels.
for ( int i = 0, M = levels.size(); i < M; ++i ) {
levels[i].lastN = long(levels[i].lastN*newtot/oldtot);
levels[i].weight /= rat;
}
levels.push_back(level);
weight /= rat;
return newmax;
}
// If we are compensating we should only accept the selection if the
// weight is above the previous overestimate.
if ( compensating() ) if ( abs(weight) < levels.back().weight ) weight = 0.0;
return WeightType();
}
/**
* Exit compensation mode and start selection procedure from
* scratch.
*/
void reset() {
N = 0;
levels.clear();
last = T();
}
/**
* Erases all objects.
*/
void clear() {
selector.clear();
reset();
}
/**
* Set the margin used to get a new overestimated probability for an
* object when entering compensation mode.
*/
void margin(double m) { theMargin = m; }
/**
* Set the tolerance for how much a weight is allowed to be
* larger than unity before starting the compensation.
*/
void tolerance(double t) { theTolerance = t; }
//@}
/** @name Simple access functions. */
//@{
/**
* Return true if this CompSelector is in a compensating state.
*/
bool compensating() {
// Leave all levels which has reached there 'expiry date'.
while ( levels.size() && levels.back().lastN < N ) levels.pop_back();
return !levels.empty();
}
/**
* If in a compensating mode, return the number of selection needed
* before exiting this mode.
*/
long compleft() const { return levels.empty()? 0: levels.back().lastN - N; }
/**
* Return the sum of probabilities of the objects inserted. Note
* that probabilities specified when objects are inserted are
* rescaled with this number to give unit probability for
* 'select()'.
*/
WeightType sum() const { return selector.sum(); }
/**
* Return the margin used to get a new overestimated probability for an
* object when entering compensation mode.
*/
double margin() const { return theMargin; }
/**
* Return the tolerance for how much a weight is allowed to be
* larger than unity before starting the compensation.
*/
double tolerance() const { return theTolerance; }
//@}
/** @name I/O functions. */
//@{
/**
* Output to a stream.
*/
template <typename OStream>
void output(OStream & os) const {
os << selector << N << last << theMargin << theTolerance << levels.size();
for ( int i = 0, M = levels.size(); i < M; ++i )
os << levels[i].lastN << levels[i].weight;
}
/**
* Input from a stream.
*/
template <typename IStream>
void input(IStream & is) {
long M;
is >> selector >> N >> last >> theMargin >> theTolerance >> M;
levels.resize(M);
for ( int i = 0; i < M; ++i ) is >> levels[i].lastN >> levels[i].weight;
}
//@}
private:
/**
* Internal struct used for bookkeeping when compensating.
*/
struct Level {
/**
* The selection number at which point this level of compensation
* is ended.
*/
long lastN;
/**
* The minimum weight allowed when compensating on this level.
*/
double weight;
};
private:
/**
* The underlying selector
*/
Selector<T,WeightType> selector;
/**
* The number of selections so far.
*/
long N;
/**
* The last selected object.
*/
T last;
/**
* The margin used to get a new overestimated probability for an
* object when entering compensation mode.
*/
double theMargin;
/**
* Set the tolerance for how much a weight is allowed to be
* larger than unity before starting the compensation.
*/
double theTolerance;
/**
* The currently active compensation levels.
*/
vector<Level> levels;
};
/**
* Output a Selector to a stream.
*/
template <typename OStream, typename T, typename WeightType>
inline OStream & operator<<(OStream & os,
const CompSelector<T,WeightType> & s) {
s.output(os);
return os;
}
/**
* Input a Selector from a stream.
*/
template <typename IStream, typename T, typename WeightType>
inline IStream & operator>>(IStream & is,
CompSelector<T,WeightType> & s) {
s.input(is);
return is;
}
}
#endif /* THEPEG_CompSelector_H */
|