diff options
author | Georgey | 2017-07-05 11:40:43 +0530 |
---|---|---|
committer | Georgey | 2017-07-05 11:40:43 +0530 |
commit | 938fef4a37a7b7c61b4b6ff74cb4cfd2f100c427 (patch) | |
tree | b343c0ee5609433c80e0de1db8b6886c9126dc2d /thirdparty/linux/include/coin1/ClpDynamicMatrix.hpp | |
parent | 5b72577efe080c5294b32d804e4d26351fef30bc (diff) | |
download | FOSSEE-Optimization-toolbox-938fef4a37a7b7c61b4b6ff74cb4cfd2f100c427.tar.gz FOSSEE-Optimization-toolbox-938fef4a37a7b7c61b4b6ff74cb4cfd2f100c427.tar.bz2 FOSSEE-Optimization-toolbox-938fef4a37a7b7c61b4b6ff74cb4cfd2f100c427.zip |
Added linux shared libraries and header files for int and ecos functions
Diffstat (limited to 'thirdparty/linux/include/coin1/ClpDynamicMatrix.hpp')
-rw-r--r-- | thirdparty/linux/include/coin1/ClpDynamicMatrix.hpp | 381 |
1 files changed, 381 insertions, 0 deletions
diff --git a/thirdparty/linux/include/coin1/ClpDynamicMatrix.hpp b/thirdparty/linux/include/coin1/ClpDynamicMatrix.hpp new file mode 100644 index 0000000..da4e144 --- /dev/null +++ b/thirdparty/linux/include/coin1/ClpDynamicMatrix.hpp @@ -0,0 +1,381 @@ +/* $Id: ClpDynamicMatrix.hpp 1755 2011-06-28 18:24:31Z lou $ */ +// Copyright (C) 2004, International Business Machines +// Corporation and others. All Rights Reserved. +// This code is licensed under the terms of the Eclipse Public License (EPL). + +#ifndef ClpDynamicMatrix_H +#define ClpDynamicMatrix_H + + +#include "CoinPragma.hpp" + +#include "ClpPackedMatrix.hpp" +class ClpSimplex; +/** This implements a dynamic matrix when we have a limit on the number of + "interesting rows". This version inherits from ClpPackedMatrix and knows that + the real matrix is gub. A later version could use shortest path to generate columns. + +*/ + +class ClpDynamicMatrix : public ClpPackedMatrix { + +public: + /// enums for status of various sorts + enum DynamicStatus { + soloKey = 0x00, + inSmall = 0x01, + atUpperBound = 0x02, + atLowerBound = 0x03 + }; + /**@name Main functions provided */ + //@{ + /// Partial pricing + virtual void partialPricing(ClpSimplex * model, double start, double end, + int & bestSequence, int & numberWanted); + + /** + update information for a pivot (and effective rhs) + */ + virtual int updatePivot(ClpSimplex * model, double oldInValue, double oldOutValue); + /** Returns effective RHS offset if it is being used. This is used for long problems + or big dynamic or anywhere where going through full columns is + expensive. This may re-compute */ + virtual double * rhsOffset(ClpSimplex * model, bool forceRefresh = false, + bool check = false); + + using ClpPackedMatrix::times ; + /** Return <code>y + A * scalar *x</code> in <code>y</code>. + @pre <code>x</code> must be of size <code>numColumns()</code> + @pre <code>y</code> must be of size <code>numRows()</code> */ + virtual void times(double scalar, + const double * x, double * y) const; + /// Modifies rhs offset + void modifyOffset(int sequence, double amount); + /// Gets key value when none in small + double keyValue(int iSet) const; + /** + mode=0 - Set up before "updateTranspose" and "transposeTimes" for duals using extended + updates array (and may use other if dual values pass) + mode=1 - Update dual solution after "transposeTimes" using extended rows. + mode=2 - Compute all djs and compute key dual infeasibilities + mode=3 - Report on key dual infeasibilities + mode=4 - Modify before updateTranspose in partial pricing + */ + virtual void dualExpanded(ClpSimplex * model, CoinIndexedVector * array, + double * other, int mode); + /** + mode=0 - Create list of non-key basics in pivotVariable_ using + number as numberBasic in and out + mode=1 - Set all key variables as basic + mode=2 - return number extra rows needed, number gives maximum number basic + mode=3 - before replaceColumn + mode=4 - return 1 if can do primal, 2 if dual, 3 if both + mode=5 - save any status stuff (when in good state) + mode=6 - restore status stuff + mode=7 - flag given variable (normally sequenceIn) + mode=8 - unflag all variables + mode=9 - synchronize costs + mode=10 - return 1 if there may be changing bounds on variable (column generation) + mode=11 - make sure set is clean (used when a variable rejected - but not flagged) + mode=12 - after factorize but before permute stuff + mode=13 - at end of simplex to delete stuff + */ + virtual int generalExpanded(ClpSimplex * model, int mode, int & number); + /** Purely for column generation and similar ideas. Allows + matrix and any bounds or costs to be updated (sensibly). + Returns non-zero if any changes. + */ + virtual int refresh(ClpSimplex * model); + /** Creates a variable. This is called after partial pricing and will modify matrix. + Will update bestSequence. + */ + virtual void createVariable(ClpSimplex * model, int & bestSequence); + /// Returns reduced cost of a variable + virtual double reducedCost( ClpSimplex * model, int sequence) const; + /// Does gub crash + void gubCrash(); + /// Writes out model (without names) + void writeMps(const char * name); + /// Populates initial matrix from dynamic status + void initialProblem(); + /** Adds in a column to gub structure (called from descendant) and returns sequence */ + int addColumn(int numberEntries, const int * row, const double * element, + double cost, double lower, double upper, int iSet, + DynamicStatus status); + /** If addColumn forces compression then this allows descendant to know what to do. + If >=0 then entry stayed in, if -1 then entry went out to lower bound.of zero. + Entries at upper bound (really nonzero) never go out (at present). + */ + virtual void packDown(const int * , int ) {} + /// Gets lower bound (to simplify coding) + inline double columnLower(int sequence) const { + if (columnLower_) return columnLower_[sequence]; + else return 0.0; + } + /// Gets upper bound (to simplify coding) + inline double columnUpper(int sequence) const { + if (columnUpper_) return columnUpper_[sequence]; + else return COIN_DBL_MAX; + } + + //@} + + + + /**@name Constructors, destructor */ + //@{ + /** Default constructor. */ + ClpDynamicMatrix(); + /** This is the real constructor. + It assumes factorization frequency will not be changed. + This resizes model !!!! + The contents of original matrix in model will be taken over and original matrix + will be sanitized so can be deleted (to avoid a very small memory leak) + */ + ClpDynamicMatrix(ClpSimplex * model, int numberSets, + int numberColumns, const int * starts, + const double * lower, const double * upper, + const CoinBigIndex * startColumn, const int * row, + const double * element, const double * cost, + const double * columnLower = NULL, const double * columnUpper = NULL, + const unsigned char * status = NULL, + const unsigned char * dynamicStatus = NULL); + + /** Destructor */ + virtual ~ClpDynamicMatrix(); + //@} + + /**@name Copy method */ + //@{ + /** The copy constructor. */ + ClpDynamicMatrix(const ClpDynamicMatrix&); + /** The copy constructor from an CoinPackedMatrix. */ + ClpDynamicMatrix(const CoinPackedMatrix&); + + ClpDynamicMatrix& operator=(const ClpDynamicMatrix&); + /// Clone + virtual ClpMatrixBase * clone() const ; + //@} + /**@name gets and sets */ + //@{ + /// Status of row slacks + inline ClpSimplex::Status getStatus(int sequence) const { + return static_cast<ClpSimplex::Status> (status_[sequence] & 7); + } + inline void setStatus(int sequence, ClpSimplex::Status status) { + unsigned char & st_byte = status_[sequence]; + st_byte = static_cast<unsigned char>(st_byte & ~7); + st_byte = static_cast<unsigned char>(st_byte | status); + } + /// Whether flagged slack + inline bool flaggedSlack(int i) const { + return (status_[i] & 8) != 0; + } + inline void setFlaggedSlack(int i) { + status_[i] = static_cast<unsigned char>(status_[i] | 8); + } + inline void unsetFlaggedSlack(int i) { + status_[i] = static_cast<unsigned char>(status_[i] & ~8); + } + /// Number of sets (dynamic rows) + inline int numberSets() const { + return numberSets_; + } + /// Number of possible gub variables + inline int numberGubEntries() const + { return startSet_[numberSets_];} + /// Sets + inline int * startSets() const + { return startSet_;} + /// Whether flagged + inline bool flagged(int i) const { + return (dynamicStatus_[i] & 8) != 0; + } + inline void setFlagged(int i) { + dynamicStatus_[i] = static_cast<unsigned char>(dynamicStatus_[i] | 8); + } + inline void unsetFlagged(int i) { + dynamicStatus_[i] = static_cast<unsigned char>(dynamicStatus_[i] & ~8); + } + inline void setDynamicStatus(int sequence, DynamicStatus status) { + unsigned char & st_byte = dynamicStatus_[sequence]; + st_byte = static_cast<unsigned char>(st_byte & ~7); + st_byte = static_cast<unsigned char>(st_byte | status); + } + inline DynamicStatus getDynamicStatus(int sequence) const { + return static_cast<DynamicStatus> (dynamicStatus_[sequence] & 7); + } + /// Saved value of objective offset + inline double objectiveOffset() const { + return objectiveOffset_; + } + /// Starts of each column + inline CoinBigIndex * startColumn() const { + return startColumn_; + } + /// rows + inline int * row() const { + return row_; + } + /// elements + inline double * element() const { + return element_; + } + /// costs + inline double * cost() const { + return cost_; + } + /// ids of active columns (just index here) + inline int * id() const { + return id_; + } + /// Optional lower bounds on columns + inline double * columnLower() const { + return columnLower_; + } + /// Optional upper bounds on columns + inline double * columnUpper() const { + return columnUpper_; + } + /// Lower bounds on sets + inline double * lowerSet() const { + return lowerSet_; + } + /// Upper bounds on sets + inline double * upperSet() const { + return upperSet_; + } + /// size + inline int numberGubColumns() const { + return numberGubColumns_; + } + /// first free + inline int firstAvailable() const { + return firstAvailable_; + } + /// first dynamic + inline int firstDynamic() const { + return firstDynamic_; + } + /// number of columns in dynamic model + inline int lastDynamic() const { + return lastDynamic_; + } + /// number of rows in original model + inline int numberStaticRows() const { + return numberStaticRows_; + } + /// size of working matrix (max) + inline int numberElements() const { + return numberElements_; + } + inline int * keyVariable() const { + return keyVariable_; + } + /// Switches off dj checking each factorization (for BIG models) + void switchOffCheck(); + /// Status region for gub slacks + inline unsigned char * gubRowStatus() const { + return status_; + } + /// Status region for gub variables + inline unsigned char * dynamicStatus() const { + return dynamicStatus_; + } + /// Returns which set a variable is in + int whichSet (int sequence) const; + //@} + + +protected: + /**@name Data members + The data members are protected to allow access for derived classes. */ + //@{ + /// Sum of dual infeasibilities + double sumDualInfeasibilities_; + /// Sum of primal infeasibilities + double sumPrimalInfeasibilities_; + /// Sum of Dual infeasibilities using tolerance based on error in duals + double sumOfRelaxedDualInfeasibilities_; + /// Sum of Primal infeasibilities using tolerance based on error in primals + double sumOfRelaxedPrimalInfeasibilities_; + /// Saved best dual on gub row in pricing + double savedBestGubDual_; + /// Saved best set in pricing + int savedBestSet_; + /// Backward pointer to pivot row !!! + int * backToPivotRow_; + /// Key variable of set (only accurate if none in small problem) + mutable int * keyVariable_; + /// Backward pointer to extra row + int * toIndex_; + // Reverse pointer from index to set + int * fromIndex_; + /// Number of sets (dynamic rows) + int numberSets_; + /// Number of active sets + int numberActiveSets_; + /// Saved value of objective offset + double objectiveOffset_; + /// Lower bounds on sets + double * lowerSet_; + /// Upper bounds on sets + double * upperSet_; + /// Status of slack on set + unsigned char * status_; + /// Pointer back to model + ClpSimplex * model_; + /// first free + int firstAvailable_; + /// first free when iteration started + int firstAvailableBefore_; + /// first dynamic + int firstDynamic_; + /// number of columns in dynamic model + int lastDynamic_; + /// number of rows in original model + int numberStaticRows_; + /// size of working matrix (max) + int numberElements_; + /// Number of dual infeasibilities + int numberDualInfeasibilities_; + /// Number of primal infeasibilities + int numberPrimalInfeasibilities_; + /** If pricing will declare victory (i.e. no check every factorization). + -1 - always check + 0 - don't check + 1 - in don't check mode but looks optimal + */ + int noCheck_; + /// Infeasibility weight when last full pass done + double infeasibilityWeight_; + /// size + int numberGubColumns_; + /// current maximum number of columns (then compress) + int maximumGubColumns_; + /// current maximum number of elemnts (then compress) + int maximumElements_; + /// Start of each set + int * startSet_; + /// next in chain + int * next_; + /// Starts of each column + CoinBigIndex * startColumn_; + /// rows + int * row_; + /// elements + double * element_; + /// costs + double * cost_; + /// ids of active columns (just index here) + int * id_; + /// for status and which bound + unsigned char * dynamicStatus_; + /// Optional lower bounds on columns + double * columnLower_; + /// Optional upper bounds on columns + double * columnUpper_; + //@} +}; + +#endif |