/usr/include/opengm/inference/trws/trws_decomposition.hxx is in libopengm-dev 2.3.6-2.
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 | #ifndef TRWS_DECOMPOSITION_HXX_
#define TRWS_DECOMPOSITION_HXX_
#include <iostream>
#include <list>
#include <vector>
#include <utility>
#include <stdexcept>
#include <algorithm>
#include <opengm/graphicalmodel/decomposition/graphicalmodeldecomposition.hxx>
#include <opengm/inference/trws/utilities2.hxx>
#ifdef TRWS_DEBUG_OUTPUT
#include <opengm/inference/trws/output_debug_utils.hxx>
#endif
namespace opengm {
namespace trws_base{
#ifdef TRWS_DEBUG_OUTPUT
using OUT::operator <<;
#endif
template<class GM>
class Decomposition
{
public:
typedef typename GM::IndexType IndexType;
typedef typename GM::LabelType LabelType;
typedef std::vector<IndexType> IndexList;
typedef opengm::GraphicalModelDecomposition::SubVariable SubVariable;
typedef opengm::GraphicalModelDecomposition::SubVariableListType SubVariableListType;
Decomposition(const GM& gm,IndexType numSubModels=0)//!< numSubModels - ESTIMATED number of submodels to optimize memory allocation
:_numberOfModels(0),_gm(gm)
{ // Reserve memory
_variableLists.reserve(numSubModels);
_pwFactorLists.reserve(numSubModels);
};
virtual ~Decomposition()=0;
virtual IndexType getNumberOfSubModels()const{return _numberOfModels;}
virtual const IndexList& getVariableList(IndexType subModelId)const {return _variableLists[subModelId];}
virtual const IndexList& getFactorList(IndexType subModelId)const {return _pwFactorLists[subModelId];}
#ifdef TRWS_DEBUG_OUTPUT
virtual void PrintTestData(std::ostream& fout);
#endif
virtual void ComputeVariableDecomposition(std::vector<SubVariableListType>* plist)const;
static void CheckUnaryFactors(const GM& gm);//!< checks whether all variables have corresp. unary factor with the same index and vice versa
static void CheckDuplicateUnaryFactors(const GM& gm);
/*static*/ void CheckForIsolatedNodes(const GM& gm);
protected:
typedef std::pair<IndexType,IndexType> Edge;//first=factorId, second=neighborNodeId
typedef std::list<Edge> EdgeList;
typedef std::vector<EdgeList> NodeList;
IndexType _numberOfModels;
std::vector<IndexList> _variableLists;
std::vector<IndexList> _pwFactorLists;
const GM& _gm;
IndexType _addSubModel();
void _addSubFactor(const IndexType& factorId);
void _addSubVariable(const IndexType& variableId);
static void _CreateNodeList(const GM& gm,NodeList* pnodeList);
};
/*
* So far oriented to 2-nd order factors only!
*/
template<class GM>
class MonotoneChainsDecomposition : public Decomposition<GM>
{
public:
typedef Decomposition<GM> parent;
typedef typename parent::IndexType IndexType;
typedef typename parent::LabelType LabelType;
typedef typename parent::IndexList IndexList;
typedef typename parent::SubVariable SubVariable;
typedef typename parent::SubVariableListType SubVariableListType;
MonotoneChainsDecomposition(const GM& gm,IndexType numSubModels=0);//!< numSubModels - ESTIMATED number of submodels to optimize memory allocation
protected:
void _GetMaximalMonotoneSequence(typename parent::NodeList* pnodesList,IndexType start);
};
template<class GM>
class GridDecomposition : public Decomposition<GM>
{
public:
typedef Decomposition<GM> parent;
typedef typename parent::IndexType IndexType;
typedef typename parent::LabelType LabelType;
typedef typename parent::IndexList IndexList;
typedef typename parent::SubVariable SubVariable;
typedef typename parent::SubVariableListType SubVariableListType;
GridDecomposition(const GM& gm,IndexType numSubModels=0);//!< numSubModels - ESTIMATED number of submodels to optimize memory allocation
IndexType xsize()const{return _xsize;}
IndexType ysize()const{return _ysize;}
private:
IndexType _xsize, _ysize;
protected:
void _computeGridSizes();
void _CheckGridModel();
void _initDecompositionLists();
IndexType _xysize()const{return _xsize*_ysize;}
IndexType _pwrowsize()const{return 2*_xsize-1;}
IndexType _pwIndexRow(IndexType x,IndexType y)const;//!> returns an index of a row pairwise factor places to the right to var (x,y)
IndexType _pwIndexCol(IndexType x,IndexType y)const;//!> returns an index of a column pairwise factor places to the down to var (x,y)
IndexType _varIndex(IndexType x,IndexType y)const{return x+_xsize*y;}
void _getRow(IndexType y,IndexList* plist)const;//!> returns indexes of variables in the row <y>
void _getCol(IndexType x,IndexList* plist)const;//!> returns indexes of variables in the column <y>
void _getPWRow(IndexType y, IndexList* plist)const;//!> return indexes of pairwise factors in the row <y>
void _getPWCol(IndexType x,IndexList* plist)const;//!> return indexes of pairwise factors in the column <x>
};
template<class GM>
class EdgeDecomposition : public Decomposition<GM>
{
public:
typedef Decomposition<GM> parent;
typedef typename parent::IndexType IndexType;
typedef typename parent::LabelType LabelType;
typedef typename parent::IndexList IndexList;
typedef typename parent::SubVariable SubVariable;
typedef typename parent::SubVariableListType SubVariableListType;
EdgeDecomposition(const GM& gm):parent(gm)
{
//parent::CheckUnaryFactors(gm);
parent::CheckDuplicateUnaryFactors(gm);
parent::_numberOfModels=gm.numberOfFactors()-gm.numberOfVariables();//!> this should be a number of pairwise factors
//bild variable and factor lists
parent::_variableLists.resize(parent::_numberOfModels,IndexList(2,(IndexType)0));
parent::_pwFactorLists.resize(parent::_numberOfModels,IndexList(1,(IndexType)0));
IndexType pwFid=0;
for (IndexType fId=0;fId<gm.numberOfFactors();++fId)
{
if (gm[fId].numberOfVariables()==1) continue;
if ((gm[fId].numberOfVariables()>2) || (gm[fId].numberOfVariables()==0))
std::runtime_error("EdgeDecomposition<GM>::EdgeDecomposition(): Only factors of order 1 or 2 are supported!");
//factor of order 2:
parent::_variableLists[pwFid][0]=gm[fId].variableIndex(0);
parent::_variableLists[pwFid][1]=gm[fId].variableIndex(1);
parent::_pwFactorLists[pwFid][0]=fId;
++pwFid;
}
}
};
#ifdef TRWS_DEBUG_OUTPUT
template <class SubFactor>
struct printSubFactor
{
printSubFactor(std::ostream& out):_out(out){};
void operator()(const SubFactor& a)
{
_out << "("<<a.subModelId_ <<","<< a.subFactorId_ <<")"<<", ";
}
private:
std::ostream& _out;
};
#endif
#ifdef TRWS_DEBUG_OUTPUT
template <class SubVariable>
struct printSubVariable
{
printSubVariable(std::ostream& out):_out(out){};
void operator()(const SubVariable& a)
{
_out << "("<<a.subModelId_ <<","<< a.subVariableId_ <<")"<<", ";
}
private:
std::ostream& _out;
};
#endif
//-------------------------IMPLEMENTATION------------------------------------------------
template<class GM>
Decomposition<GM>::~Decomposition<GM>()
{}
#ifdef TRWS_DEBUG_OUTPUT
template<class GM>
void Decomposition<GM>::PrintTestData(std::ostream& fout)
{
fout <<"_numberOfModels;" << _numberOfModels<<std::endl;
fout <<"_variableLists:"<<_variableLists<<std::endl;
fout <<"_pwFactorLists:"<<_pwFactorLists<<std::endl;
}
#endif
template<class GM>
MonotoneChainsDecomposition<GM>::MonotoneChainsDecomposition(const GM& gm,IndexType numSubModels)
:parent(gm,numSubModels)
{ parent::CheckDuplicateUnaryFactors(gm);
parent::CheckForIsolatedNodes(gm);
typename parent::NodeList nodeList(gm.numberOfVariables());
parent::_CreateNodeList(gm,&nodeList);
for (IndexType start=0;start<nodeList.size();++start)
while (!nodeList[start].empty())
{ parent::_addSubModel();
_GetMaximalMonotoneSequence(&nodeList,(IndexType)start);
}
}
template<class GM>
GridDecomposition<GM>::GridDecomposition(const GM& gm,IndexType numSubModels)
:parent(gm,numSubModels)
{
//estimate xsize and ysize
_computeGridSizes();
parent::_numberOfModels=_xsize+_ysize;
//bild variable and factor lists
_initDecompositionLists();
}
template<class Factor>
bool dependsOnVariable(const Factor& f,typename Factor::IndexType varId)
{
return (std::find(f.variableIndicesBegin(),f.variableIndicesEnd(),varId) != f.variableIndicesEnd());
}
template<class GM>
void GridDecomposition<GM>::_computeGridSizes()
{
IndexType numberOfVars=parent::_gm.numberOfVariables();
IndexType numTotal=parent::_gm.numberOfFactors();
std::vector<IndexType> ind;
for (IndexType f=numberOfVars;f<numTotal;++f)
{
std::vector<IndexType> ind(parent::_gm[f].numberOfVariables());
if (ind.size()!=2)
throw std::runtime_error("GridDecomposition<GM>::_computeGridSizes():Incorrect grid structure! : only pairwise factors are supported !=0");
parent::_gm[f].variableIndices(ind.begin());
if (ind[1]<=ind[0])
throw std::runtime_error("GridDecomposition<GM>::_computeGridSizes():Incorrect grid structure! : pairwise factors should be oriented from smaller to larger variable indices !=0");
if (ind[1]-ind[0]!=1)
{
_xsize=ind[1]-ind[0];
_ysize=numberOfVars/_xsize;
if (numberOfVars%_xsize !=0)
throw std::runtime_error("GridDecomposition<GM>::_computeGridSizes():Incorrect grid structure! : numberOfVars%xsize !=0");
break;
}else if (f==numTotal-1)
{
_xsize=numberOfVars;
_ysize=1;
break;
};
};
_CheckGridModel();
};
template<class GM>
void GridDecomposition<GM>::_CheckGridModel()
{
bool incorrect=false;
//check vertical structure
for (IndexType y=0;y<_ysize;++y)
for (IndexType x=0;x<_xsize;++x)
{
if (y<_ysize-1)
{
IndexType ind=_pwIndexCol(x,y);
if (!dependsOnVariable(parent::_gm[ind],_varIndex(x,y)) || !dependsOnVariable(parent::_gm[ind],_varIndex(x,y+1)) )
incorrect=true;
};
if ((x<_xsize-1))
{
IndexType ind=_pwIndexRow(x,y);
if (!dependsOnVariable(parent::_gm[ind],_varIndex(x,y)) || !dependsOnVariable(parent::_gm[ind],_varIndex(x+1,y)))
incorrect=true;
}
if (incorrect)
throw std::runtime_error("GridDecomposition::_CheckGridModel():Incorrect grid structure!");
};
};
template<class GM>
void GridDecomposition<GM>::_initDecompositionLists()
{
parent::_variableLists.resize(parent::_numberOfModels);
parent::_pwFactorLists.resize(parent::_numberOfModels);
for (IndexType x=0;x<_xsize;++x)
{
_getCol(x,&parent::_variableLists[x]);
_getPWCol(x,&parent::_pwFactorLists[x]);
}
for (IndexType y=0;y<_ysize;++y)
{
_getRow(y,&parent::_variableLists[_xsize+y]);
_getPWRow(y,&parent::_pwFactorLists[_xsize+y]);
};
}
//make the vector of nodes with lists of edges. Each edge is present only once - in the list of the node with the smaller index
template<class GM>
void Decomposition<GM>::_CreateNodeList(const GM & gm,NodeList* pnodeList)
{
NodeList& varList=*pnodeList;
varList.resize(gm.numberOfVariables());
for (IndexType factorId=0;factorId<gm.numberOfFactors();++factorId)
{
if (gm[factorId].numberOfVariables()>2)
throw std::runtime_error("CreateEdgeList(): Only factors up to order 2 are supported!");
if (gm[factorId].numberOfVariables()==1) continue;
std::vector<IndexType> varIndices(gm[factorId].variableIndicesBegin(),gm[factorId].variableIndicesEnd());
if (varIndices[0] < varIndices[1])
varList[varIndices[0]].push_back(std::make_pair(factorId,varIndices[1]));
else
varList[varIndices[1]].push_back(std::make_pair(factorId,varIndices[0]));
}
}
template<class GM>
typename Decomposition<GM>::IndexType Decomposition<GM>::_addSubModel()
{
_variableLists.push_back(IndexList());
_pwFactorLists.push_back(IndexList());
_numberOfModels++;
return IndexType(_numberOfModels-1);
};
template<class GM>
void Decomposition<GM>::_addSubFactor(const IndexType& factorId)
{
_pwFactorLists[_numberOfModels-1].push_back(factorId);
}
template<class GM>
void Decomposition<GM>::_addSubVariable(const IndexType& variableId)
{
_variableLists[_numberOfModels-1].push_back(variableId);
}
template<class GM>
void MonotoneChainsDecomposition<GM>::_GetMaximalMonotoneSequence(typename parent::NodeList* pnodeList,IndexType start)
{
assert(start < pnodeList->size());
typename parent::NodeList& nodeList=*pnodeList;
if (!nodeList[start].empty())
parent::_addSubVariable(start);
else return;
while ( !nodeList[start].empty() )
{
typename parent::EdgeList::iterator it= nodeList[start].begin();
parent::_addSubVariable(it->second);
parent::_addSubFactor(it->first);
IndexType tmp=it->second;
nodeList[start].erase(it);
start=tmp;
}
}
template<class GM>
void Decomposition<GM>::CheckUnaryFactors(const GM& gm)
{
bool error=false;
for (IndexType factorId=0;factorId<gm.numberOfFactors();++factorId)
{
std::vector<IndexType> varIndices(gm[factorId].variableIndicesBegin(),gm[factorId].variableIndicesEnd());
if (gm[factorId].numberOfVariables()==1)
{
if ( (factorId < gm.numberOfVariables()) && (varIndices[0]==factorId))
continue;
else error=true;
}else if (factorId < gm.numberOfVariables())
error=true;
if (error)
throw std::runtime_error("Decomposition<GM>::CheckUnaryFactors(): Each variable has to have a unique unary factor, which moreover has the same index!");
}
}
template<class GM>
void Decomposition<GM>::CheckDuplicateUnaryFactors(const GM& gm)
{
std::vector<IndexType> numOfunaryFactors(gm.numberOfVariables(),(IndexType)0);
for (IndexType factorId=0;factorId<gm.numberOfFactors();++factorId)
{
if (gm[factorId].numberOfVariables()!=1)
continue;
numOfunaryFactors[gm[factorId].variableIndex(0)]++;
}
IndexType oneCount=std::count(numOfunaryFactors.begin(),numOfunaryFactors.end(),(IndexType)1);
exception_check(oneCount==numOfunaryFactors.size(),"Decomposition::CheckDuplicateUnaryFactors: all variables must have a unique associated unary factor!");
}
template<class GM>
void Decomposition<GM>::CheckForIsolatedNodes(const GM& gm)
{
for (IndexType varId=0;varId<gm.numberOfVariables();++varId)
{
bool isolatedNode=true;
for (IndexType localId=0;localId<gm.numberOfFactors(varId);++localId)
{
if (gm[gm.factorOfVariable(varId,localId)].numberOfVariables()>1)
isolatedNode=false;
}
if (isolatedNode==true)
{
_addSubModel();
_addSubVariable(varId);
//TODO:TEST throw std::runtime_error("Decomposition<GM>::CheckForIsolatedNodes(): Procesing of isolated nodes is not supported!");
}
}
}
template<class GM>
void Decomposition<GM>::ComputeVariableDecomposition(std::vector<SubVariableListType>* plist)const
{
plist->resize(_gm.numberOfVariables());
for (IndexType modelId=0;modelId<_numberOfModels;++modelId)
for (IndexType varId=0;varId<_variableLists[modelId].size();++varId)
(*plist)[_variableLists[modelId][varId]].push_back(SubVariable(modelId,varId));
}
template<class GM>
typename GridDecomposition<GM>::IndexType
GridDecomposition<GM>::_pwIndexRow(IndexType x,IndexType y)const//!> returns an index of a row pairwise factor places to the right to var (x,y)
{
assert(x<_xsize-1);
assert(y<_ysize);
if ((y==_ysize-1) && (x!=0)) return _pwIndexRow(0,y) + x;
return _xysize()+y*_pwrowsize()+2*x;
};
template<class GM>
typename GridDecomposition<GM>::IndexType
GridDecomposition<GM>::_pwIndexCol(IndexType x,IndexType y)const//!> returns an index of a column pairwise factor places to the down to var (x,y)
{
if (x==_xsize-1) return _pwIndexCol(x-1,y)+1;
return _pwIndexRow(x,y)+1;
};
template<class GM>
void GridDecomposition<GM>::
_getRow(IndexType y,IndexList* plist)const//!> returns indexes of variables in the row <y>
{
plist->resize(_xsize);
(*plist)[0]=_varIndex(0,y);
for (IndexType i=1;i<_xsize;++i)
(*plist)[i]=(*plist)[i-1]+1;
};
template<class GM>
void GridDecomposition<GM>::
_getCol(IndexType x,IndexList* plist)const//!> returns indexes of variables in the column <y>
{
plist->resize(_ysize);
(*plist)[0]=_varIndex(x,0);
for (IndexType i=1;i<_ysize;++i)
(*plist)[i]=(*plist)[i-1]+_xsize;
};
template<class GM>
void GridDecomposition<GM>::
_getPWRow(IndexType y, IndexList* plist)const//!> return indexes of pairwise factors in the row <y>
{
plist->resize(_xsize-1);
if (_xsize<=1)
return;
(*plist)[0]=_pwIndexRow(0,y);
IndexType step=2;
if (y==_ysize-1) step=1;
for (IndexType i=1;i<_xsize-1;++i)
(*plist)[i]=(*plist)[i-1]+step;
};
template<class GM>
void GridDecomposition<GM>::
_getPWCol(IndexType x,IndexList* plist)const//!> return indexes of pairwise factors in the column <x>
{
plist->resize(_ysize-1);
if (_ysize<=1)
return;
(*plist)[0]=_pwIndexCol(x,0);
for (IndexType i=1;i<_ysize-1;++i)
(*plist)[i]=(*plist)[i-1]+_pwrowsize();
};
}//namespace trws_base
}//namespace opengm
#endif /* DECOMPOSITIONTRWS_H_ */
|