Up  

Module Glpk


module Glpk: sig .. end
OCaml bindings to glpk. Please see the glpk manual for further explanations on the semantics of functions.

Warning: contrarily to the C version of glpk, all indexes are 0-based.


Author(s): Samuel Mimram


Types


type lp
A linear programmation problem.

type direction =
| Minimize
| Maximize
Direction of the optimization.

type aux_var_type =
| Free_var
| Lower_bounded_var
| Upper_bounded_var
| Double_bounded_var
| Fixed_var
Type of bounds of an auxiliary variable.

type prob_class =
| Linear_prog (*linear programming*)
| Mixed_integer_prog (*mixed integer programming*)
Class of a problem.

type var_kind =
| Continuous_var (*continuous variable*)
| Integer_var (*integer variable*)
Kind of a variable.

Exceptions


exception Fault
The problem has no rows/columns, or the initial basis is invalid, or the initial basis matrix is singular or ill-conditionned.
exception Empty
The problem has no rows and/or column.
exception Bad_basis
The LP basis is invalid beacause the number of basic variables is not the same as the number of rows.
exception Lower_limit
The objective function being minimized has reached its lower limit and continues decreasing.
exception Upper_limit
The objective function being maximized has reached its upper limit and continues increasing.
exception No_primal_feasible_solution
The problem has no primal feasible solution.
exception No_dual_feasible_solution
The problem has no dual feasible solution.
exception Iteration_limit
Iterations limit exceeded.
exception Time_limit
Time limit exceeded.
exception No_convergence
Very slow convergence or divergence.
exception Solver_failure
Failure of the solver (the current basis matrix got singular or ill-conditionned).
exception Unknown_error
Unknown error (this exception should disappear in future versions).

Functions



Creating, reading and saving problems


val new_problem : unit -> lp
Create a new linear programmation problem.
val make_problem : direction ->
       float array ->
       float array array ->
       (float * float) array -> (float * float) array -> lp
make_problem dir zcoefs constrs pbounds xbounds creates the new linear programmation problem where Z = Sum_i zcoefs.(i) * x_ i should be optimized in the direction dir under the constraints fst pbounds.(i) <= p_i <= snd pbounds.(i) and fst xbounds.(i) <= x_i <= snd xbounds.(i) where p_i = Sum_j constrs.(i).(j) * x_j. The bounds may be + / - infinity.
val read_cplex : string -> lp
Read problem data in CPLEX LP format from a file.
val write_cplex : lp -> string -> unit
Write prblem data in CPLEX LP format into a file.

Setting and retreiving paramters of a problem


val set_prob_name : lp -> string -> unit
Set the problem name.
val get_prob_name : lp -> string
Retrieve the problem name.
val set_class : lp -> prob_class -> unit
Set the problem class.
val get_class : lp -> prob_class
Retrieve the problem class.
val set_direction : lp -> direction -> unit
Set the direction of the optimization.
val get_direction : lp -> direction
Retrieve the direction of the optimization.
val set_obj_name : lp -> string -> unit
Set the objective name.
val get_obj_name : lp -> string
Retrieve the objective name.
val add_rows : lp -> int -> unit
Add rows.
val get_num_rows : lp -> int
Retreive the number of rows.
val set_row_name : lp -> int -> string -> unit
Set the name of a row.
val get_row_name : lp -> int -> string
Retrieve the name of a row.
val set_row_bounds : lp -> int -> aux_var_type -> float -> float -> unit
Set a row bound.
val add_columns : lp -> int -> unit
Add columns.
val get_num_cols : lp -> int
Retreive the number of columns.
val set_col_name : lp -> int -> string -> unit
Set the name of a column.
val get_col_name : lp -> int -> string
Retrieve the name of a column.
val set_col_kind : lp -> int -> var_kind -> unit
Set column kind.
val set_col_bounds : lp -> int -> aux_var_type -> float -> float -> unit
Set a column boudaries.
val set_obj_coef : lp -> int -> float -> unit
Set an objective coefficient.
val load_matrix : lp -> float array array -> unit
Load a constraint matrix.

Solving problems and retreiving solutions


val scale_problem : lp -> unit
Scale problem data.
val unscale_problem : lp -> unit
Unscale problem data.
val warm_up : lp -> unit
Warm up the LP basis for the specified problem object using current statuses assigned to rows and columns.
val simplex : lp -> unit
Solve an LP problem using the simplex method. You must use builtin presolver (see use_presolver) to get an exception if the problem has no feasible solution.
val interior : lp -> unit
Solve an LP problem using the primal-dual interior point method.
val branch_and_bound : lp -> unit
Solve a MIP proble using the branch-and-bound method.
val branch_and_bound_opt : lp -> unit
Solve a MIP proble using and optimized version of the branch-and-bound method.
val get_obj_val : lp -> float
Retrieve objective value.
val get_col_primal : lp -> int -> float
Get the primal value of the structural variable associated with a column.
val get_col_primals : lp -> float array
Get the primal values of the structural variables associated with each column.

Setting parameters of the solver


val set_message_level : lp -> int -> unit
Set the level of messages output by sover routines. The second argument might be: 
val use_presolver : lp -> bool -> unit
Use the builtin LP-presolver in simplex?
val set_simplex_iteration_count : lp -> int -> unit
Initialize the simplex iteration counter.
val reset_simplex_iteration_count : lp -> unit
Reset the simplex iteration counter.
val get_simplex_iteration_count : lp -> int
This number is incremented after each simplex iteration.
val set_simplex_iteration_limit : lp -> int -> unit
Set the maximum number of iterations that simplex should do.
val get_simplex_iteration_limit : lp -> int
Retrieve the maximum number of iterations that simplex should do.
val set_simplex_time_limit : lp -> float -> unit
Set the maximum amount of time that simplex should take.
val get_simplex_time_limit : lp -> float
Retrieve the maximum amount of time that simplex should take.