Class MILPSolver

Nested Relationships

Nested Types

Inheritance Relationships

Derived Types

Class Documentation

class MILPSolver

Subclassed by MILPSolverCLP, MILPSolverCPX

Public Types

enum ColumnType

Values:

enumerator C_REAL
enumerator C_INT
enumerator C_BOOL

Public Functions

inline virtual ~MILPSolver()
virtual MILPSolver *clone() = 0
virtual double getInfinity() = 0
virtual void addRow(const vector<pair<int, double>> &entries, const double &lb, const double &ub) = 0

Functions for the rows (constraints) of the model.

virtual void setRowName(const int &cons, const string &asString) = 0
virtual string getRowName(const int &cons) = 0
virtual double getRowUpper(const int &var) = 0
virtual void setRowUpper(const int &c, const double &b) = 0
virtual double getRowLower(const int &var) = 0
virtual void setRowLower(const int &c, const double &b) = 0
virtual int getNumRows() = 0
virtual void addCol(const vector<pair<int, double>> &entries, const double &lb, const double &ub, const ColumnType &type) = 0

Functions for the columns (variables) of the model.

virtual bool isColumnInteger(const int &c) = 0

Ascertain whether the given column is an integer.

Parameters:

c – A column index

Returns:

true if column c is an integer.

inline virtual bool isColumnBinary(const int &c)

Ascertain whether the given column is binary (i.e. a [0,1] integer).

Parameters:

c – A column index

Returns:

true if column c is binary.

virtual void setColName(const int &var, const string &asString) = 0
virtual string getColName(const int &var) = 0
virtual double getColUpper(const int &var) = 0
virtual void setColUpper(const int &var, const double &b) = 0
virtual double getColLower(const int &var) = 0
virtual void setColLower(const int &var, const double &b) = 0
virtual void setColBounds(const int &var, const double &lb, const double &ub) = 0
virtual int getNumCols() = 0
inline virtual void addEmptyRealCols(const int &n)

Add empty columns to the LP, for real-valued variables. Default bounds are 0 to getInfinity().

Parameters:

n – The number of columns to add

virtual void setMaximiseObjective(const bool &maxim) = 0

Objective function calls.

Specify that the calls to solve() should maximise the given objective, rather than minimising.

Parameters:

maxim – If true, the objective will be maximised.

virtual void setObjective(double *const entries) = 0

Set the objective function according to the vector of weights given.

Parameters:

entries – An array, size equal to getNumCols() containing the linear coefficient for each column, to use in the objective.

inline virtual void setQuadraticObjective(const Objective &o)

Specify a (possibly quadratic) objective to use when solving the LP.

Parameters:

o – The objective function to use

virtual void setObjCoeff(const int &var, const double &w) = 0

Set the linear coefficient in the objective function of the given column.

Parameters:

  • var – The column whose coefficient is to be changed

  • w – The coefficient for column <codevar

virtual void clearObjective() = 0

Clear the objective coefficients (set each to 0).

virtual bool solve(const bool &skipPresolve) = 0

Solving, and accessing solutions.

virtual bool quadraticPreSolve()

Find the best combination of the settings to the binary variables in the quadratic objective passed to setQuadraticObjective(). The default implementation branches over the settings to the integer objective terms, fixing the bounds on the variables to the best solution found. Note that after calling this a call to solve() is still to be able to access the best solution.

Returns:

true if the MIQCP could be solved.

virtual const double *getSolution() = 0
virtual const double *getPartialSolution(const int &from, const int &to) = 0

Obtain the solution values of a subset of the columns in the LP.

Parameters:

  • from – The start column index of the range to return (inclusive)

  • to – The end column index of the range to return (exclusive)

Returns:

A pointer to an array, where index 0 contains the value of the variable from etc.

inline virtual double getSingleSolutionVariableValue(const int &col)

Obtain the solution values of the specified variable.

Parameters:

col – The index of the variable (column)

Returns:

The value that variable takes in the most recent solution to the LP

virtual const double *getSolutionRows() = 0

Obtain the value of the row in the solution.

Returns:

An array containing getNumRows() values, one for each row.

inline virtual double getSingleSolutionRowValue(const int &row)

Obtain the value of the specified row in the solution.

Parameters:

row – A row index

Returns:

The value it takes in the solution

virtual double getObjValue() = 0

Return the computed value of the objective function.

virtual void writeLp(const string &filename) = 0

Miscellaneous functions.

virtual void getRow(const int &i, vector<pair<int, double>> &entries) = 0

Get a row from the LP, returning the coefficients it contains.

Parameters:

  • i – The index of the row to get

  • entries – A reference to a vector of int,double pairs which, after calling the method, will contain the column indices and coefficients of row i of the LP.

virtual void hush() = 0

Suppress all output from the solver

Protected Functions

inline MILPSolver()

Protected Attributes

Objective quadraticObjective
class Objective

Public Types

typedef map<int, Coefficient>::const_iterator const_iterator

Public Functions

inline Objective(const bool maxObj = false)
inline void setMaximise(const bool &m)
inline const bool &maximise() const
inline Coefficient &getTerm(const int &v)
inline const_iterator begin() const
inline const_iterator end() const
inline size_t size() const

Protected Attributes

map<int, Coefficient> terms
bool _maximise
struct Coefficient

Public Functions

inline Coefficient()

Public Members

double linearCoefficient
map<int, double> nonLinearCoefficients