ruby-cbc 0.1.9

data/ext/ruby-cbc/install/include/coin/ClpSimplexDual.hpp

@@ -0,0 +1,300 @@
+/* $Id: ClpSimplexDual.hpp 1761 2011-07-06 16:06:24Z forrest $ */
+// Copyright (C) 2002, International Business Machines
+// Corporation and others.  All Rights Reserved.
+// This code is licensed under the terms of the Eclipse Public License (EPL).
+/*
+   Authors
+
+   John Forrest
+
+ */
+#ifndef ClpSimplexDual_H
+#define ClpSimplexDual_H
+
+#include "ClpSimplex.hpp"
+
+/** This solves LPs using the dual simplex method
+
+    It inherits from ClpSimplex.  It has no data of its own and
+    is never created - only cast from a ClpSimplex object at algorithm time.
+
+*/
+
+class ClpSimplexDual : public ClpSimplex {
+
+public:
+
+     /**@name Description of algorithm */
+     //@{
+     /** Dual algorithm
+
+         Method
+
+        It tries to be a single phase approach with a weight of 1.0 being
+        given to getting optimal and a weight of updatedDualBound_ being
+        given to getting dual feasible.  In this version I have used the
+        idea that this weight can be thought of as a fake bound.  If the
+        distance between the lower and upper bounds on a variable is less
+        than the feasibility weight then we are always better off flipping
+        to other bound to make dual feasible.  If the distance is greater
+        then we make up a fake bound updatedDualBound_ away from one bound.
+        If we end up optimal or primal infeasible, we check to see if
+        bounds okay.  If so we have finished, if not we increase updatedDualBound_
+        and continue (after checking if unbounded). I am undecided about
+        free variables - there is coding but I am not sure about it.  At
+        present I put them in basis anyway.
+
+        The code is designed to take advantage of sparsity so arrays are
+        seldom zeroed out from scratch or gone over in their entirety.
+        The only exception is a full scan to find outgoing variable for
+        Dantzig row choice.  For steepest edge we keep an updated list
+        of infeasibilities (actually squares).
+        On easy problems we don't need full scan - just
+        pick first reasonable.
+
+        One problem is how to tackle degeneracy and accuracy.  At present
+        I am using the modification of costs which I put in OSL and some
+        of what I think is the dual analog of Gill et al.
+        I am still not sure of the exact details.
+
+        The flow of dual is three while loops as follows:
+
+        while (not finished) {
+
+          while (not clean solution) {
+
+             Factorize and/or clean up solution by flipping variables so
+         dual feasible.  If looks finished check fake dual bounds.
+         Repeat until status is iterating (-1) or finished (0,1,2)
+
+          }
+
+          while (status==-1) {
+
+            Iterate until no pivot in or out or time to re-factorize.
+
+            Flow is:
+
+            choose pivot row (outgoing variable).  if none then
+        we are primal feasible so looks as if done but we need to
+        break and check bounds etc.
+
+        Get pivot row in tableau
+
+            Choose incoming column.  If we don't find one then we look
+        primal infeasible so break and check bounds etc.  (Also the
+        pivot tolerance is larger after any iterations so that may be
+        reason)
+
+            If we do find incoming column, we may have to adjust costs to
+        keep going forwards (anti-degeneracy).  Check pivot will be stable
+        and if unstable throw away iteration and break to re-factorize.
+        If minor error re-factorize after iteration.
+
+        Update everything (this may involve flipping variables to stay
+        dual feasible.
+
+          }
+
+        }
+
+        TODO's (or maybe not)
+
+        At present we never check we are going forwards.  I overdid that in
+        OSL so will try and make a last resort.
+
+        Needs partial scan pivot out option.
+
+        May need other anti-degeneracy measures, especially if we try and use
+        loose tolerances as a way to solve in fewer iterations.
+
+        I like idea of dynamic scaling.  This gives opportunity to decouple
+        different implications of scaling for accuracy, iteration count and
+        feasibility tolerance.
+
+        for use of exotic parameter startFinishoptions see Clpsimplex.hpp
+     */
+
+     int dual(int ifValuesPass, int startFinishOptions = 0);
+     /** For strong branching.  On input lower and upper are new bounds
+         while on output they are change in objective function values
+         (>1.0e50 infeasible).
+         Return code is 0 if nothing interesting, -1 if infeasible both
+         ways and +1 if infeasible one way (check values to see which one(s))
+         Solutions are filled in as well - even down, odd up - also
+         status and number of iterations
+     */
+     int strongBranching(int numberVariables, const int * variables,
+                         double * newLower, double * newUpper,
+                         double ** outputSolution,
+                         int * outputStatus, int * outputIterations,
+                         bool stopOnFirstInfeasible = true,
+                         bool alwaysFinish = false,
+                         int startFinishOptions = 0);
+     /// This does first part of StrongBranching
+     ClpFactorization * setupForStrongBranching(char * arrays, int numberRows,
+               int numberColumns, bool solveLp = false);
+     /// This cleans up after strong branching
+     void cleanupAfterStrongBranching(ClpFactorization * factorization);
+     //@}
+
+     /**@name Functions used in dual */
+     //@{
+     /** This has the flow between re-factorizations
+         Broken out for clarity and will be used by strong branching
+
+         Reasons to come out:
+         -1 iterations etc
+         -2 inaccuracy
+         -3 slight inaccuracy (and done iterations)
+         +0 looks optimal (might be unbounded - but we will investigate)
+         +1 looks infeasible
+         +3 max iterations
+
+         If givenPi not NULL then in values pass
+      */
+     int whileIterating(double * & givenPi, int ifValuesPass);
+     /** The duals are updated by the given arrays.
+         Returns number of infeasibilities.
+         After rowArray and columnArray will just have those which
+         have been flipped.
+         Variables may be flipped between bounds to stay dual feasible.
+         The output vector has movement of primal
+         solution (row length array) */
+     int updateDualsInDual(CoinIndexedVector * rowArray,
+                           CoinIndexedVector * columnArray,
+                           CoinIndexedVector * outputArray,
+                           double theta,
+                           double & objectiveChange,
+                           bool fullRecompute);
+     /** The duals are updated by the given arrays.
+         This is in values pass - so no changes to primal is made
+     */
+     void updateDualsInValuesPass(CoinIndexedVector * rowArray,
+                                  CoinIndexedVector * columnArray,
+                                  double theta);
+     /** While updateDualsInDual sees what effect is of flip
+         this does actual flipping.
+     */
+     void flipBounds(CoinIndexedVector * rowArray,
+                     CoinIndexedVector * columnArray);
+     /**
+         Row array has row part of pivot row
+         Column array has column part.
+         This chooses pivot column.
+         Spare arrays are used to save pivots which will go infeasible
+         We will check for basic so spare array will never overflow.
+         If necessary will modify costs
+         For speed, we may need to go to a bucket approach when many
+         variables are being flipped.
+         Returns best possible pivot value
+     */
+     double dualColumn(CoinIndexedVector * rowArray,
+                       CoinIndexedVector * columnArray,
+                       CoinIndexedVector * spareArray,
+                       CoinIndexedVector * spareArray2,
+                       double accpetablePivot,
+                       CoinBigIndex * dubiousWeights);
+     /// Does first bit of dualColumn
+     int dualColumn0(const CoinIndexedVector * rowArray,
+                     const CoinIndexedVector * columnArray,
+                     CoinIndexedVector * spareArray,
+                     double acceptablePivot,
+                     double & upperReturn, double &bestReturn, double & badFree);
+     /**
+         Row array has row part of pivot row
+         Column array has column part.
+         This sees what is best thing to do in dual values pass
+         if sequenceIn==sequenceOut can change dual on chosen row and leave variable in basis
+     */
+     void checkPossibleValuesMove(CoinIndexedVector * rowArray,
+                                  CoinIndexedVector * columnArray,
+                                  double acceptablePivot);
+     /**
+         Row array has row part of pivot row
+         Column array has column part.
+         This sees what is best thing to do in branch and bound cleanup
+         If sequenceIn_ < 0 then can't do anything
+     */
+     void checkPossibleCleanup(CoinIndexedVector * rowArray,
+                               CoinIndexedVector * columnArray,
+                               double acceptablePivot);
+     /**
+         This sees if we can move duals in dual values pass.
+         This is done before any pivoting
+     */
+     void doEasyOnesInValuesPass(double * givenReducedCosts);
+     /**
+         Chooses dual pivot row
+         Would be faster with separate region to scan
+         and will have this (with square of infeasibility) when steepest
+         For easy problems we can just choose one of the first rows we look at
+
+         If alreadyChosen >=0 then in values pass and that row has been
+         selected
+     */
+     void dualRow(int alreadyChosen);
+     /** Checks if any fake bounds active - if so returns number and modifies
+         updatedDualBound_ and everything.
+         Free variables will be left as free
+         Returns number of bounds changed if >=0
+         Returns -1 if not initialize and no effect
+         Fills in changeVector which can be used to see if unbounded
+         and cost of change vector
+         If 2 sets to original (just changed)
+     */
+     int changeBounds(int initialize, CoinIndexedVector * outputArray,
+                      double & changeCost);
+     /** As changeBounds but just changes new bounds for a single variable.
+         Returns true if change */
+     bool changeBound( int iSequence);
+     /// Restores bound to original bound
+     void originalBound(int iSequence);
+     /** Checks if tentative optimal actually means unbounded in dual
+         Returns -3 if not, 2 if is unbounded */
+     int checkUnbounded(CoinIndexedVector * ray, CoinIndexedVector * spare,
+                        double changeCost);
+     /**  Refactorizes if necessary
+          Checks if finished.  Updates status.
+          lastCleaned refers to iteration at which some objective/feasibility
+          cleaning too place.
+
+          type - 0 initial so set up save arrays etc
+               - 1 normal -if good update save
+           - 2 restoring from saved
+     */
+     void statusOfProblemInDual(int & lastCleaned, int type,
+                                double * givenDjs, ClpDataSave & saveData,
+                                int ifValuesPass);
+     /** Perturbs problem (method depends on perturbation())
+         returns nonzero if should go to dual */
+     int perturb();
+     /** Fast iterations.  Misses out a lot of initialization.
+         Normally stops on maximum iterations, first re-factorization
+         or tentative optimum.  If looks interesting then continues as
+         normal.  Returns 0 if finished properly, 1 otherwise.
+     */
+     int fastDual(bool alwaysFinish = false);
+     /** Checks number of variables at fake bounds.  This is used by fastDual
+         so can exit gracefully before end */
+     int numberAtFakeBound();
+
+     /** Pivot in a variable and choose an outgoing one.  Assumes dual
+         feasible - will not go through a reduced cost.  Returns step length in theta
+         Return codes as before but -1 means no acceptable pivot
+     */
+     int pivotResultPart1();
+     /** Get next free , -1 if none */
+     int nextSuperBasic();
+     /** Startup part of dual (may be extended to other algorithms)
+         returns 0 if good, 1 if bad */
+     int startupSolve(int ifValuesPass, double * saveDuals, int startFinishOptions);
+     void finishSolve(int startFinishOptions);
+     void gutsOfDual(int ifValuesPass, double * & saveDuals, int initialStatus,
+                     ClpDataSave & saveData);
+     //int dual2(int ifValuesPass,int startFinishOptions=0);
+     void resetFakeBounds(int type);
+
+     //@}
+};
+#endif

data/ext/ruby-cbc/install/include/coin/ClpSimplexNonlinear.hpp

@@ -0,0 +1,117 @@
+/* $Id: ClpSimplexNonlinear.hpp 2025 2014-03-19 12:49:55Z forrest $ */
+// 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).
+/*
+   Authors
+
+   John Forrest
+
+ */
+#ifndef ClpSimplexNonlinear_H
+#define ClpSimplexNonlinear_H
+
+class ClpNonlinearInfo;
+class ClpQuadraticObjective;
+class ClpConstraint;
+
+#include "ClpSimplexPrimal.hpp"
+
+/** This solves non-linear LPs using the primal simplex method
+
+    It inherits from ClpSimplexPrimal.  It has no data of its own and
+    is never created - only cast from a ClpSimplexPrimal object at algorithm time.
+    If needed create new class and pass around
+
+*/
+
+class ClpSimplexNonlinear : public ClpSimplexPrimal {
+
+public:
+
+     /**@name Description of algorithm */
+     //@{
+     /** Primal algorithms for reduced gradient
+         At present we have two algorithms:
+
+     */
+     /// A reduced gradient method.
+     int primal();
+     /** Primal algorithm for quadratic
+         Using a semi-trust region approach as for pooling problem
+         This is in because I have it lying around
+     */
+     int primalSLP(int numberPasses, double deltaTolerance,
+		int otherOptions=0);
+     /// May use a cut approach for solving any LP
+     int primalDualCuts(char * rowsIn, int startUp, int algorithm);
+     /** Primal algorithm for nonlinear constraints
+         Using a semi-trust region approach as for pooling problem
+         This is in because I have it lying around
+
+     */
+     int primalSLP(int numberConstraints, ClpConstraint ** constraints,
+                   int numberPasses, double deltaTolerance);
+
+     /** Creates direction vector.  note longArray is long enough
+         for rows and columns.  If numberNonBasic 0 then is updated
+         otherwise mode is ignored and those are used.
+         Norms are only for those > 1.0e3*dualTolerance
+         If mode is nonzero then just largest dj */
+     void directionVector (CoinIndexedVector * longArray,
+                           CoinIndexedVector * spare1, CoinIndexedVector * spare2,
+                           int mode,
+                           double & normFlagged, double & normUnflagged,
+                           int & numberNonBasic);
+     /// Main part.
+     int whileIterating (int & pivotMode);
+     /**
+         longArray has direction
+         pivotMode -
+               0 - use all dual infeasible variables
+           1 - largest dj
+           while >= 10 trying startup phase
+         Returns 0 - can do normal iteration (basis change)
+         1 - no basis change
+         2 - if wants singleton
+         3 - if time to re-factorize
+         If sequenceIn_ >=0 then that will be incoming variable
+     */
+     int pivotColumn(CoinIndexedVector * longArray,
+                     CoinIndexedVector * rowArray,
+                     CoinIndexedVector * columnArray,
+                     CoinIndexedVector * spare,
+                     int & pivotMode,
+                     double & solutionError,
+                     double * array1);
+     /**  Refactorizes if necessary
+          Checks if finished.  Updates status.
+          lastCleaned refers to iteration at which some objective/feasibility
+          cleaning too place.
+
+          type - 0 initial so set up save arrays etc
+               - 1 normal -if good update save
+           - 2 restoring from saved
+     */
+     void statusOfProblemInPrimal(int & lastCleaned, int type,
+                                  ClpSimplexProgress * progress,
+                                  bool doFactorization,
+                                  double & bestObjectiveWhenFlagged);
+     /** Do last half of an iteration.
+         Return codes
+         Reasons to come out normal mode
+         -1 normal
+         -2 factorize now - good iteration
+         -3 slight inaccuracy - refactorize - iteration done
+         -4 inaccuracy - refactorize - no iteration
+         -5 something flagged - go round again
+         +2 looks unbounded
+         +3 max iterations (iteration done)
+
+     */
+     int pivotNonlinearResult();
+     //@}
+
+};
+#endif
+

data/ext/ruby-cbc/install/include/coin/ClpSimplexOther.hpp

@@ -0,0 +1,277 @@
+/* $Id: ClpSimplexOther.hpp 2070 2014-11-18 11:12:54Z forrest $ */
+// 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).
+/*
+   Authors
+
+   John Forrest
+
+ */
+#ifndef ClpSimplexOther_H
+#define ClpSimplexOther_H
+
+#include "ClpSimplex.hpp"
+
+/** This is for Simplex stuff which is neither dual nor primal
+
+    It inherits from ClpSimplex.  It has no data of its own and
+    is never created - only cast from a ClpSimplex object at algorithm time.
+
+*/
+
+class ClpSimplexOther : public ClpSimplex {
+
+public:
+
+     /**@name Methods */
+     //@{
+     /** Dual ranging.
+         This computes increase/decrease in cost for each given variable and corresponding
+         sequence numbers which would change basis.  Sequence numbers are 0..numberColumns
+         and numberColumns.. for artificials/slacks.
+         For non-basic variables the information is trivial to compute and the change in cost is just minus the
+         reduced cost and the sequence number will be that of the non-basic variables.
+         For basic variables a ratio test is between the reduced costs for non-basic variables
+         and the row of the tableau corresponding to the basic variable.
+         The increase/decrease value is always >= 0.0
+
+         Up to user to provide correct length arrays where each array is of length numberCheck.
+         which contains list of variables for which information is desired.  All other
+         arrays will be filled in by function.  If fifth entry in which is variable 7 then fifth entry in output arrays
+         will be information for variable 7.
+
+         If valueIncrease/Decrease not NULL (both must be NULL or both non NULL) then these are filled with
+         the value of variable if such a change in cost were made (the existing bounds are ignored)
+
+         When here - guaranteed optimal
+     */
+     void dualRanging(int numberCheck, const int * which,
+                      double * costIncrease, int * sequenceIncrease,
+                      double * costDecrease, int * sequenceDecrease,
+                      double * valueIncrease = NULL, double * valueDecrease = NULL);
+     /** Primal ranging.
+         This computes increase/decrease in value for each given variable and corresponding
+         sequence numbers which would change basis.  Sequence numbers are 0..numberColumns
+         and numberColumns.. for artificials/slacks.
+         This should only be used for non-basic variabls as otherwise information is pretty useless
+         For basic variables the sequence number will be that of the basic variables.
+
+         Up to user to provide correct length arrays where each array is of length numberCheck.
+         which contains list of variables for which information is desired.  All other
+         arrays will be filled in by function.  If fifth entry in which is variable 7 then fifth entry in output arrays
+         will be information for variable 7.
+
+         When here - guaranteed optimal
+     */
+     void primalRanging(int numberCheck, const int * which,
+                        double * valueIncrease, int * sequenceIncrease,
+                        double * valueDecrease, int * sequenceDecrease);
+     /** Parametrics
+         This is an initial slow version.
+         The code uses current bounds + theta * change (if change array not NULL)
+         and similarly for objective.
+         It starts at startingTheta and returns ending theta in endingTheta.
+         If reportIncrement 0.0 it will report on any movement
+         If reportIncrement >0.0 it will report at startingTheta+k*reportIncrement.
+         If it can not reach input endingTheta return code will be 1 for infeasible,
+         2 for unbounded, if error on ranges -1,  otherwise 0.
+         Normal report is just theta and objective but
+         if event handler exists it may do more
+         On exit endingTheta is maximum reached (can be used for next startingTheta)
+     */
+     int parametrics(double startingTheta, double & endingTheta, double reportIncrement,
+                     const double * changeLowerBound, const double * changeUpperBound,
+                     const double * changeLowerRhs, const double * changeUpperRhs,
+                     const double * changeObjective);
+     /** Version of parametrics which reads from file
+	 See CbcClpParam.cpp for details of format
+	 Returns -2 if unable to open file */
+     int parametrics(const char * dataFile);
+     /** Parametrics
+         This is an initial slow version.
+         The code uses current bounds + theta * change (if change array not NULL)
+         It starts at startingTheta and returns ending theta in endingTheta.
+         If it can not reach input endingTheta return code will be 1 for infeasible,
+         2 for unbounded, if error on ranges -1,  otherwise 0.
+         Event handler may do more
+         On exit endingTheta is maximum reached (can be used for next startingTheta)
+     */
+     int parametrics(double startingTheta, double & endingTheta, 
+                     const double * changeLowerBound, const double * changeUpperBound,
+                     const double * changeLowerRhs, const double * changeUpperRhs);
+     int parametricsObj(double startingTheta, double & endingTheta, 
+			const double * changeObjective);
+    /// Finds best possible pivot
+    double bestPivot(bool justColumns=false);
+  typedef struct {
+    double startingTheta;
+    double endingTheta;
+    double maxTheta;
+    double acceptableMaxTheta; // if this far then within tolerances
+    double * lowerChange; // full array of lower bound changes
+    int * lowerList; // list of lower bound changes
+    double * upperChange; // full array of upper bound changes
+    int * upperList; // list of upper bound changes
+    char * markDone; // mark which ones looked at
+    int * backwardBasic; // from sequence to pivot row
+    int * lowerActive;
+    double * lowerGap;
+    double * lowerCoefficient;
+    int * upperActive;
+    double * upperGap;
+    double * upperCoefficient;
+    int unscaledChangesOffset; 
+    bool firstIteration; // so can update rhs for accuracy
+  } parametricsData;
+
+private:
+     /** Parametrics - inner loop
+         This first attempt is when reportIncrement non zero and may
+         not report endingTheta correctly
+         If it can not reach input endingTheta return code will be 1 for infeasible,
+         2 for unbounded,  otherwise 0.
+         Normal report is just theta and objective but
+         if event handler exists it may do more
+     */
+     int parametricsLoop(parametricsData & paramData, double reportIncrement,
+                         const double * changeLower, const double * changeUpper,
+                         const double * changeObjective, ClpDataSave & data,
+                         bool canTryQuick);
+     int parametricsLoop(parametricsData & paramData,
+                         ClpDataSave & data,bool canSkipFactorization=false);
+     int parametricsObjLoop(parametricsData & paramData,
+                         ClpDataSave & data,bool canSkipFactorization=false);
+     /**  Refactorizes if necessary
+          Checks if finished.  Updates status.
+
+          type - 0 initial so set up save arrays etc
+               - 1 normal -if good update save
+           - 2 restoring from saved
+     */
+     void statusOfProblemInParametrics(int type, ClpDataSave & saveData);
+     void statusOfProblemInParametricsObj(int type, ClpDataSave & saveData);
+     /** This has the flow between re-factorizations
+
+         Reasons to come out:
+         -1 iterations etc
+         -2 inaccuracy
+         -3 slight inaccuracy (and done iterations)
+         +0 looks optimal (might be unbounded - but we will investigate)
+         +1 looks infeasible
+         +3 max iterations
+      */
+     int whileIterating(parametricsData & paramData, double reportIncrement,
+                        const double * changeObjective);
+     /** Computes next theta and says if objective or bounds (0= bounds, 1 objective, -1 none).
+         theta is in theta_.
+         type 1 bounds, 2 objective, 3 both.
+     */
+     int nextTheta(int type, double maxTheta, parametricsData & paramData,
+                   const double * changeObjective);
+     int whileIteratingObj(parametricsData & paramData);
+     int nextThetaObj(double maxTheta, parametricsData & paramData);
+     /// Restores bound to original bound
+     void originalBound(int iSequence, double theta, const double * changeLower,
+		     const double * changeUpper);
+     /// Compute new rowLower_ etc (return negative if infeasible - otherwise largest change)
+     double computeRhsEtc(parametricsData & paramData);
+     /// Redo lower_ from rowLower_ etc
+     void redoInternalArrays();
+     /**
+         Row array has row part of pivot row
+         Column array has column part.
+         This is used in dual ranging
+     */
+     void checkDualRatios(CoinIndexedVector * rowArray,
+                          CoinIndexedVector * columnArray,
+                          double & costIncrease, int & sequenceIncrease, double & alphaIncrease,
+                          double & costDecrease, int & sequenceDecrease, double & alphaDecrease);
+     /**
+         Row array has pivot column
+         This is used in primal ranging
+     */
+     void checkPrimalRatios(CoinIndexedVector * rowArray,
+                            int direction);
+     /// Returns new value of whichOther when whichIn enters basis
+     double primalRanging1(int whichIn, int whichOther);
+
+public:
+     /** Write the basis in MPS format to the specified file.
+     If writeValues true writes values of structurals
+     (and adds VALUES to end of NAME card)
+
+     Row and column names may be null.
+     formatType is
+     <ul>
+       <li> 0 - normal
+       <li> 1 - extra accuracy
+       <li> 2 - IEEE hex (later)
+     </ul>
+
+     Returns non-zero on I/O error
+     */
+     int writeBasis(const char *filename,
+                    bool writeValues = false,
+                    int formatType = 0) const;
+     /// Read a basis from the given filename
+     int readBasis(const char *filename);
+     /** Creates dual of a problem if looks plausible
+         (defaults will always create model)
+         fractionRowRanges is fraction of rows allowed to have ranges
+         fractionColumnRanges is fraction of columns allowed to have ranges
+     */
+     ClpSimplex * dualOfModel(double fractionRowRanges = 1.0, double fractionColumnRanges = 1.0) const;
+     /** Restores solution from dualized problem
+         non-zero return code indicates minor problems
+     */
+     int restoreFromDual(const ClpSimplex * dualProblem,
+			 bool checkAccuracy=false);
+     /** Sets solution in dualized problem
+         non-zero return code indicates minor problems
+     */
+     int setInDual(ClpSimplex * dualProblem);
+     /** Does very cursory presolve.
+         rhs is numberRows, whichRows is 3*numberRows and whichColumns is 2*numberColumns.
+     */
+     ClpSimplex * crunch(double * rhs, int * whichRows, int * whichColumns,
+                         int & nBound, bool moreBounds = false, bool tightenBounds = false);
+     /** After very cursory presolve.
+         rhs is numberRows, whichRows is 3*numberRows and whichColumns is 2*numberColumns.
+     */
+     void afterCrunch(const ClpSimplex & small,
+                      const int * whichRows, const int * whichColumns,
+                      int nBound);
+     /** Returns gub version of model or NULL
+	 whichRows has to be numberRows
+	 whichColumns has to be numberRows+numberColumns */
+     ClpSimplex * gubVersion(int * whichRows, int * whichColumns,
+			     int neededGub,
+			     int factorizationFrequency=50);
+     /// Sets basis from original
+     void setGubBasis(ClpSimplex &original,const int * whichRows,
+		      const int * whichColumns);
+     /// Restores basis to original
+     void getGubBasis(ClpSimplex &original,const int * whichRows,
+		      const int * whichColumns) const;
+     /// Quick try at cleaning up duals if postsolve gets wrong
+     void cleanupAfterPostsolve();
+     /** Tightens integer bounds - returns number tightened or -1 if infeasible
+     */
+     int tightenIntegerBounds(double * rhsSpace);
+     /** Expands out all possible combinations for a knapsack
+         If buildObj NULL then just computes space needed - returns number elements
+         On entry numberOutput is maximum allowed, on exit it is number needed or
+         -1 (as will be number elements) if maximum exceeded.  numberOutput will have at
+         least space to return values which reconstruct input.
+         Rows returned will be original rows but no entries will be returned for
+         any rows all of whose entries are in knapsack.  So up to user to allow for this.
+         If reConstruct >=0 then returns number of entrie which make up item "reConstruct"
+         in expanded knapsack.  Values in buildRow and buildElement;
+     */
+     int expandKnapsack(int knapsackRow, int & numberOutput,
+                        double * buildObj, CoinBigIndex * buildStart,
+                        int * buildRow, double * buildElement, int reConstruct = -1) const;
+     //@}
+};
+#endif