API Reference¶

This page documents every public symbol in the cyhighs package. The descriptions are generated from the docstrings in the source.

Solver interfaces¶

cyhighs.solve_linear_problem ¶

solve_linear_problem(
    objective_coefficients,
    *,
    constraint_matrix_values=None,
    constraint_matrix_row_indices=None,
    constraint_matrix_column_pointers=None,
    constraint_lower_bounds=None,
    constraint_upper_bounds=None,
    variable_lower_bounds=None,
    variable_upper_bounds=None,
    integrality=None,
    initial_column_values=None,
    objective_sense: ObjectiveSense | int = MINIMIZE,
    options=None,
) -> LinearProblemSolution

Solve minimize c @ x subject to linear constraints and bounds.

This is the thin low level interface. It maps almost one to one onto the HiGHS C API, so the constraint matrix must already be a single matrix in compressed sparse column form together with per row lower and upper bounds. The problem solved is:

optimize   objective_coefficients @ x
subject to constraint_lower_bounds <= A @ x <= constraint_upper_bounds
           variable_lower_bounds <= x <= variable_upper_bounds

Inequality and equality constraints are both expressed through the row bounds. An inequality row A_i @ x <= b_i uses a lower bound of negative infinity and an upper bound of b_i, while an equality row uses equal lower and upper bounds. The convenience wrapper solve_linear_problem_sparse builds this merged form from separate inequality and equality matrices.

Parameters:

Name	Type	Description	Default
`objective_coefficients`		The objective coefficient vector `c` of length `number_of_columns`.	required
`constraint_matrix_values`		The `A` matrix stored values in CSC form.	`None`
`constraint_matrix_row_indices`		The `A` matrix row indices in CSC form.	`None`
`constraint_matrix_column_pointers`		The `A` matrix column pointers in CSC form. The three constraint matrix arrays must be given together, or all omitted for a problem with no constraints.	`None`
`constraint_lower_bounds`		Per row lower bounds. Required when the constraint matrix is given.	`None`
`constraint_upper_bounds`		Per row upper bounds. Required when the constraint matrix is given, and must match the lower bounds length.	`None`
`variable_lower_bounds`		Per variable lower bounds. Default to negative infinity, meaning free below.	`None`
`variable_upper_bounds`		Per variable upper bounds. Default to positive infinity, meaning free above.	`None`
`integrality`		Per variable integrality using `VariableType` values. When omitted the problem is solved as a pure linear program.	`None`
`initial_column_values`		A warm start solution for the variables.	`None`
`objective_sense`	`ObjectiveSense \| int`	Whether to minimize (default) or maximize the objective.	`MINIMIZE`
`options`		Solver options as a mapping from `HighsOption` to a value. Unlike raw HiGHS, `HighsOption.OUTPUT_FLAG` defaults to `False` here, so the solver is silent unless it is explicitly set.	`None`

Returns:

Type	Description
`LinearProblemSolution`	The `LinearProblemSolution` record.

cyhighs.solve_linear_problem_sparse ¶

solve_linear_problem_sparse(
    objective_coefficients,
    *,
    inequality_constraint_matrix=None,
    inequality_upper_bounds=None,
    equality_constraint_matrix=None,
    equality_right_hand_sides=None,
    variable_lower_bounds=None,
    variable_upper_bounds=None,
    integrality=None,
    initial_column_values=None,
    objective_sense: ObjectiveSense | int = MINIMIZE,
    options=None,
) -> LinearProblemSolution

Solve a linear or mixed integer program with SciPy sparse matrices.

Minimizes c @ x subject to A_ub @ x <= b_ub, A_eq @ x == b_eq and the variable bounds. The two constraint matrices are merged internally into the row bounded form used by solve_linear_problem.

Parameters:

Name	Type	Description	Default
`objective_coefficients`		The objective coefficient vector `c`.	required
`inequality_constraint_matrix`		The inequality matrix `A_ub` in any sparse format.	`None`
`inequality_upper_bounds`		The right hand side vector `b_ub`. Required when `A_ub` is given.	`None`
`equality_constraint_matrix`		The equality matrix `A_eq` in any sparse format. When both matrices are given they must share the same sparse format.	`None`
`equality_right_hand_sides`		The right hand side vector `b_eq`. Required when `A_eq` is given.	`None`
`variable_lower_bounds`		Per variable lower bounds.	`None`
`variable_upper_bounds`		Per variable upper bounds.	`None`
`integrality`		Per variable integrality using `VariableType` values.	`None`
`initial_column_values`		A warm start solution.	`None`
`objective_sense`	`ObjectiveSense \| int`	Whether to minimize (default) or maximize.	`MINIMIZE`
`options`		Solver options as a mapping from `HighsOption` to a value.	`None`

Returns:

Type	Description
`LinearProblemSolution`	The `LinearProblemSolution` record.

cyhighs.linprog ¶

linprog(
    c,
    A_ub=None,
    b_ub=None,
    A_eq=None,
    b_eq=None,
    bounds=None,
    method: str = "highs",
    x0=None,
    integrality=None,
    options=None,
)

Solve a linear program with a SciPy linprog compatible interface.

Minimizes c @ x subject to A_ub @ x <= b_ub, A_eq @ x == b_eq and the given variable bounds. Constraint matrices may be dense array likes or SciPy sparse matrices.

Parameters:

Name	Type	Description	Default
`c`		Coefficients of the linear objective to minimize.	required
`A_ub`		Inequality constraint matrix.	`None`
`b_ub`		Inequality right hand side.	`None`
`A_eq`		Equality constraint matrix.	`None`
`b_eq`		Equality right hand side.	`None`
`bounds`		Bounds on the variables following the SciPy convention. Defaults to `(0, None)` for every variable.	`None`
`method`	`str`	Present for SciPy compatibility. Only HiGHS is used, so this is accepted and otherwise ignored.	`'highs'`
`x0`		A warm start solution, wired to the HiGHS warm start.	`None`
`integrality`		Per variable integrality using the SciPy convention, which matches the `VariableType` values.	`None`
`options`		Solver options as a mapping from `HighsOption` to a value.	`None`

Returns:

Type	Description
	An `OptimizeResult` with `x`, `fun`, `slack`,
	`con`, `status`, `success`, `message`, `nit` and `highs_solution`
	fields.

The solution record¶

cyhighs.LinearProblemSolution ¶

Bases: NamedTuple

The outcome of solving a linear or mixed integer program.

The field names follow HiGHS terminology. Columns are decision variables and rows are constraints. For mixed integer problems the dual value and row value fields are set to None, since dual information is not defined there.

Attributes:

Name	Type	Description
`model_status`	`ModelStatus`	The status HiGHS reported for the solve, as a `ModelStatus` value.
`column_values`	`ndarray`	The solution vector `x`, one value per variable.
`objective_value`	`float`	The objective value at the returned solution.
`column_dual_values`	`ndarray \| None`	The reduced costs, one per variable. `None` for mixed integer problems.
`row_dual_values`	`ndarray \| None`	The dual values of the constraints, one per row. `None` for mixed integer problems.
`row_values`	`ndarray \| None`	The activity of each constraint row, that is the left hand side value at the solution. `None` for mixed integer problems.
`simplex_iteration_count`	`int`	The number of simplex iterations performed, or -1 if not available.
`presolved_num_columns`	`int`	The number of columns in the presolved model. Equal to `column_values`'s length if presolve was off or made no reductions.
`presolved_num_rows`	`int`	The number of rows in the presolved model. Equal to `row_values`'s length if presolve was off or made no reductions.
`presolved_num_nonzeros`	`int`	The number of nonzeros in the constraint matrix of the presolved model.

is_optimal `property` ¶

is_optimal: bool

Return True if HiGHS proved the returned solution optimal.

cyhighs.OptimizeResult ¶

Bases: NamedTuple

SciPy linprog shaped result, returned by linprog.

Carries the same field names scipy.optimize.OptimizeResult exposes for a linprog call, plus a highs_solution field with the full HiGHS solve result.

Attributes:

Name	Type	Description
`x`	`ndarray`	The solution vector.
`fun`	`float`	The objective value at the returned solution.
`slack`	`ndarray`	The slack in each inequality constraint, `b_ub - A_ub @ x`.
`con`	`ndarray`	The residual of each equality constraint, `A_eq @ x - b_eq`.
`status`	`int`	SciPy convention status code: 0 optimal, 1 iteration or time limit, 2 infeasible, 3 unbounded, 4 numerical or other failure.
`success`	`bool`	True if `status` is 0.
`message`	`str`	Human readable description of `status`.
`nit`	`int`	The number of solver iterations performed.
`highs_solution`	`LinearProblemSolution`	The full `LinearProblemSolution` HiGHS returned for this solve.

Enumerations¶

cyhighs.ObjectiveSense ¶

Bases: IntEnum

Direction in which the objective function is optimized.

Mirrors the kHighsObjSense constants.

MAXIMIZE `class-attribute` `instance-attribute` ¶

MAXIMIZE = -1

Maximize the objective.

MINIMIZE `class-attribute` `instance-attribute` ¶

MINIMIZE = 1

Minimize the objective. This is the default and the convention the package interface is documented against.

cyhighs.VariableType ¶

Bases: IntEnum

Integrality domain of a single decision variable.

Mirrors the kHighsVarType constants. This is what HiGHS calls integrality, and the values are used directly in the integrality array passed to Highs_passMip.

CONTINUOUS `class-attribute` `instance-attribute` ¶

CONTINUOUS = 0

The variable may take any real value within its bounds.

IMPLICIT_INTEGER `class-attribute` `instance-attribute` ¶

IMPLICIT_INTEGER = 4

The variable is continuous but known to take an integer value at the optimum. This is used internally by HiGHS and is rarely set by callers.

INTEGER `class-attribute` `instance-attribute` ¶

INTEGER = 1

The variable must take an integer value within its bounds.

SEMI_CONTINUOUS `class-attribute` `instance-attribute` ¶

SEMI_CONTINUOUS = 2

The variable is either zero or a real value within its bounds. A finite upper bound is required by HiGHS for semi continuous variables.

SEMI_INTEGER `class-attribute` `instance-attribute` ¶

SEMI_INTEGER = 3

The variable is either zero or an integer value within its bounds. A finite upper bound is required by HiGHS for semi integer variables.

cyhighs.ModelStatus ¶

Bases: IntEnum

Outcome reported by HiGHS after a solve.

Mirrors the kHighsModelStatus constants from HiGHS 1.15.1. The most commonly inspected values are OPTIMAL, INFEASIBLE and UNBOUNDED.

INFEASIBLE `class-attribute` `instance-attribute` ¶

INFEASIBLE = 8

The model has no feasible solution.

INTERRUPT `class-attribute` `instance-attribute` ¶

INTERRUPT = 17

The solve was interrupted by the user.

ITERATION_LIMIT `class-attribute` `instance-attribute` ¶

ITERATION_LIMIT = 14

The solve stopped because the iteration limit was reached.

LOAD_ERROR `class-attribute` `instance-attribute` ¶

LOAD_ERROR = 1

The model could not be loaded.

MODEL_EMPTY `class-attribute` `instance-attribute` ¶

MODEL_EMPTY = 6

The model contains no variables and no constraints.

MODEL_ERROR `class-attribute` `instance-attribute` ¶

MODEL_ERROR = 2

The model is invalid, for example because of inconsistent dimensions.

NOTSET `class-attribute` `instance-attribute` ¶

NOTSET = 0

No status has been set. Seen when a solve has not been run.

OBJECTIVE_BOUND `class-attribute` `instance-attribute` ¶

OBJECTIVE_BOUND = 11

The solve stopped because the objective reached a supplied bound.

OBJECTIVE_TARGET `class-attribute` `instance-attribute` ¶

OBJECTIVE_TARGET = 12

The solve stopped because the objective reached a supplied target.

OPTIMAL `class-attribute` `instance-attribute` ¶

OPTIMAL = 7

An optimal solution was found.

POSTSOLVE_ERROR `class-attribute` `instance-attribute` ¶

POSTSOLVE_ERROR = 5

An error occurred during postsolve.

PRESOLVE_ERROR `class-attribute` `instance-attribute` ¶

PRESOLVE_ERROR = 3

An error occurred during presolve.

SOLUTION_LIMIT `class-attribute` `instance-attribute` ¶

SOLUTION_LIMIT = 16

The solve stopped because the solution count limit was reached.

SOLVE_ERROR `class-attribute` `instance-attribute` ¶

SOLVE_ERROR = 4

An error occurred during the main solve.

TIME_LIMIT `class-attribute` `instance-attribute` ¶

TIME_LIMIT = 13

The solve stopped because the time limit was reached.

UNBOUNDED `class-attribute` `instance-attribute` ¶

UNBOUNDED = 10

The objective is unbounded, so no finite optimum exists.

UNBOUNDED_OR_INFEASIBLE `class-attribute` `instance-attribute` ¶

UNBOUNDED_OR_INFEASIBLE = 9

The model is either unbounded or infeasible. HiGHS could not distinguish the two, often because presolve detected the condition.

UNKNOWN `class-attribute` `instance-attribute` ¶

UNKNOWN = 15

The status could not be determined.

cyhighs.PresolveRule ¶

Bases: IntEnum

A presolve reduction rule, identified by its bit position.

Mirrors the PresolveRuleType constants from HiGHS 1.15.1. Each member's value is the bit position HiGHS uses for that rule in the HighsOption.PRESOLVE_RULE_OFF bitmask and in HighsOption.PRESOLVE_RULE_TEST. Use PresolveRule.mask to combine rules into a bitmask rather than shifting bits by hand, for example:

{HighsOption.PRESOLVE_RULE_OFF: PresolveRule.mask(PresolveRule.PROBING, PresolveRule.SPARSIFY)}

EMPTY_ROW through DOMINATED_COL are always active in HiGHS and cannot actually be turned off through the bitmask, only FORCING_ROW onward can.

AGGREGATOR `class-attribute` `instance-attribute` ¶

AGGREGATOR = 12

Substitute out columns by aggregating rows.

COL_STUFFING `class-attribute` `instance-attribute` ¶

COL_STUFFING = 18

Fix columns to their best bound when doing so cannot cause infeasibility.

DEPENDENT_EQUATIONS `class-attribute` `instance-attribute` ¶

DEPENDENT_EQUATIONS = 10

Remove equations that are linearly dependent on other equations.

DEPENDENT_FREE_COLS `class-attribute` `instance-attribute` ¶

DEPENDENT_FREE_COLS = 11

Remove free columns that are linearly dependent on other columns.

DOMINATED_COL `class-attribute` `instance-attribute` ¶

DOMINATED_COL = 5

Fix columns whose cost and bounds mean one bound is always optimal.

DOUBLETON_EQUATION `class-attribute` `instance-attribute` ¶

DOUBLETON_EQUATION = 9

Substitute out one column of an equation with exactly two columns.

DUAL_FIXING `class-attribute` `instance-attribute` ¶

DUAL_FIXING = 17

Fix columns whose reduced cost sign forces them to a bound.

EMPTY_COL `class-attribute` `instance-attribute` ¶

EMPTY_COL = 3

Remove columns with no nonzero coefficients by fixing them.

EMPTY_ROW `class-attribute` `instance-attribute` ¶

EMPTY_ROW = 0

Remove rows with no nonzero coefficients.

ENUMERATION `class-attribute` `instance-attribute` ¶

ENUMERATION = 16

Solve small independent components of the model by enumeration.

FIXED_COL `class-attribute` `instance-attribute` ¶

FIXED_COL = 4

Substitute out columns whose bounds force a single value.

FORCING_COL `class-attribute` `instance-attribute` ¶

FORCING_COL = 7

Remove columns whose bounds force every row containing them.

FORCING_ROW `class-attribute` `instance-attribute` ¶

FORCING_ROW = 6

Remove rows whose bounds force every column in the row to a bound.

FREE_COL_SUBSTITUTION `class-attribute` `instance-attribute` ¶

FREE_COL_SUBSTITUTION = 8

Substitute out free columns using a row they appear in.

INITIAL_SWEEP `class-attribute` `instance-attribute` ¶

INITIAL_SWEEP = 19

The initial sweep that removes empty rows and columns.

PARALLEL_ROWS_AND_COLS `class-attribute` `instance-attribute` ¶

PARALLEL_ROWS_AND_COLS = 13

Merge rows or columns that are parallel to one another.

PROBING `class-attribute` `instance-attribute` ¶

PROBING = 15

Fix binary variables whose value can be deduced by probing.

REDUNDANT_ROW `class-attribute` `instance-attribute` ¶

REDUNDANT_ROW = 2

Remove rows implied by the bounds of their columns.

SINGLETON_ROW `class-attribute` `instance-attribute` ¶

SINGLETON_ROW = 1

Remove rows with a single nonzero coefficient by bounding the column.

SPARSIFY `class-attribute` `instance-attribute` ¶

SPARSIFY = 14

Reduce fill-in by combining rows to cancel out matrix entries.

mask `classmethod` ¶

mask(*rules: PresolveRule) -> int

Combine rules into the bitmask presolve_rule_off expects.

For example, PresolveRule.mask(PresolveRule.PROBING, PresolveRule.SPARSIFY) disables just those two rules.

Options¶

cyhighs.HighsOption ¶

Bases: Enum

A HiGHS option together with its name and value kind.

The enumeration value is a (name, kind) pair. name is the string the HiGHS C API expects and kind is one of the OPTION_KIND_* labels.

ALLOWED_COST_SCALE_FACTOR `class-attribute` `instance-attribute` ¶

ALLOWED_COST_SCALE_FACTOR = (
    "allowed_cost_scale_factor",
    OPTION_KIND_INT,
)

Largest power-of-two factor permitted when scaling the costs. Integer, default 0.

ALLOWED_MATRIX_SCALE_FACTOR `class-attribute` `instance-attribute` ¶

ALLOWED_MATRIX_SCALE_FACTOR = (
    "allowed_matrix_scale_factor",
    OPTION_KIND_INT,
)

Largest power-of-two factor permitted when scaling the constraint matrix. Integer, default 20.

ALLOW_UNBOUNDED_OR_INFEASIBLE `class-attribute` `instance-attribute` ¶

ALLOW_UNBOUNDED_OR_INFEASIBLE = (
    "allow_unbounded_or_infeasible",
    OPTION_KIND_BOOL,
)

Whether ModelStatus.UNBOUNDED_OR_INFEASIBLE may be reported when presolve cannot distinguish the two. Boolean, default False.

BLEND_MULTI_OBJECTIVES `class-attribute` `instance-attribute` ¶

BLEND_MULTI_OBJECTIVES = (
    "blend_multi_objectives",
    OPTION_KIND_BOOL,
)

Blend multiple objectives or apply lexicographically. Boolean, default True.

CENTRING_RATIO_TOLERANCE `class-attribute` `instance-attribute` ¶

CENTRING_RATIO_TOLERANCE = (
    "centring_ratio_tolerance",
    OPTION_KIND_DOUBLE,
)

Centring stops when the ratio max(x_js_j) / min(x_js_j) is below this tolerance. Double, default 100.

COST_SCALE_FACTOR `class-attribute` `instance-attribute` ¶

COST_SCALE_FACTOR = ('cost_scale_factor', OPTION_KIND_INT)

Scaling factor for costs. Integer, default 0.

DUAL_FEASIBILITY_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_FEASIBILITY_TOLERANCE = (
    "dual_feasibility_tolerance",
    OPTION_KIND_DOUBLE,
)

Tolerance for dual feasibility. Double, default 1e-7.

DUAL_RESIDUAL_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_RESIDUAL_TOLERANCE = (
    "dual_residual_tolerance",
    OPTION_KIND_DOUBLE,
)

Dual residual tolerance. Double, default 1e-7.

DUAL_SIMPLEX_COST_PERTURBATION_MULTIPLIER `class-attribute` `instance-attribute` ¶

DUAL_SIMPLEX_COST_PERTURBATION_MULTIPLIER = (
    "dual_simplex_cost_perturbation_multiplier",
    OPTION_KIND_DOUBLE,
)

Dual simplex cost perturbation multiplier. 0 means no perturbation. Double, default 1.0.

DUAL_SIMPLEX_PIVOT_GROWTH_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_SIMPLEX_PIVOT_GROWTH_TOLERANCE = (
    "dual_simplex_pivot_growth_tolerance",
    OPTION_KIND_DOUBLE,
)

Dual simplex pivot growth tolerance. Double, default 1e-9.

DUAL_STEEPEST_EDGE_WEIGHT_ERROR_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_STEEPEST_EDGE_WEIGHT_ERROR_TOLERANCE = (
    "dual_steepest_edge_weight_error_tolerance",
    OPTION_KIND_DOUBLE,
)

Tolerance on dual steepest edge weight errors. Double, default infinity.

DUAL_STEEPEST_EDGE_WEIGHT_LOG_ERROR_THRESHOLD `class-attribute` `instance-attribute` ¶

DUAL_STEEPEST_EDGE_WEIGHT_LOG_ERROR_THRESHOLD = (
    "dual_steepest_edge_weight_log_error_threshold",
    OPTION_KIND_DOUBLE,
)

Threshold on dual steepest edge weight errors for the Devex switch. Double, default 10.

FACTOR_PIVOT_THRESHOLD `class-attribute` `instance-attribute` ¶

FACTOR_PIVOT_THRESHOLD = (
    "factor_pivot_threshold",
    OPTION_KIND_DOUBLE,
)

Matrix factorization pivot threshold. Double, default 0.1.

FACTOR_PIVOT_TOLERANCE `class-attribute` `instance-attribute` ¶

FACTOR_PIVOT_TOLERANCE = (
    "factor_pivot_tolerance",
    OPTION_KIND_DOUBLE,
)

Matrix factorization pivot tolerance. Double, default 1e-10.

GLPSOL_COST_ROW_LOCATION `class-attribute` `instance-attribute` ¶

GLPSOL_COST_ROW_LOCATION = (
    "glpsol_cost_row_location",
    OPTION_KIND_INT,
)

Location of cost row for Glpsol file: -2 => Last; -1 => None; 0 => None if empty, otherwise data file location; 1 <= n <= num_row => Location n; n > num_row => Last. Integer, default 0.

HIGHS_ANALYSIS_LEVEL `class-attribute` `instance-attribute` ¶

HIGHS_ANALYSIS_LEVEL = (
    "highs_analysis_level",
    OPTION_KIND_INT,
)

Analysis level in HiGHS. Integer, default 0.

HIGHS_DEBUG_LEVEL `class-attribute` `instance-attribute` ¶

HIGHS_DEBUG_LEVEL = ('highs_debug_level', OPTION_KIND_INT)

Internal debugging verbosity from 0 to 3. Integer, default 0.

HIPO_BLOCK_SIZE `class-attribute` `instance-attribute` ¶

HIPO_BLOCK_SIZE = ('hipo_block_size', OPTION_KIND_INT)

Block size for dense linear algebra within HiPO. Integer, default 128.

HIPO_ORDERING `class-attribute` `instance-attribute` ¶

HIPO_ORDERING = ('hipo_ordering', OPTION_KIND_STRING)

HiPO matrix reordering: "choose", "metis", "amd" or "rcm". String, default "choose".

HIPO_PARALLEL_TYPE `class-attribute` `instance-attribute` ¶

HIPO_PARALLEL_TYPE = (
    "hipo_parallel_type",
    OPTION_KIND_STRING,
)

HiPO parallelism: "tree", "node" or "both". String, default "both".

HIPO_SYSTEM `class-attribute` `instance-attribute` ¶

HIPO_SYSTEM = ('hipo_system', OPTION_KIND_STRING)

HiPO Newton system: "choose", "augmented" or "normaleq". String, default "choose".

ICRASH `class-attribute` `instance-attribute` ¶

ICRASH = ('icrash', OPTION_KIND_BOOL)

Run iCrash. Boolean, default False.

ICRASH_APPROX_ITER `class-attribute` `instance-attribute` ¶

ICRASH_APPROX_ITER = ("icrash_approx_iter", OPTION_KIND_INT)

iCrash approximate minimization iterations. Integer, default 50.

ICRASH_BREAKPOINTS `class-attribute` `instance-attribute` ¶

ICRASH_BREAKPOINTS = (
    "icrash_breakpoints",
    OPTION_KIND_BOOL,
)

Exact subproblem solution for iCrash. Boolean, default False.

ICRASH_DUALIZE `class-attribute` `instance-attribute` ¶

ICRASH_DUALIZE = ('icrash_dualize', OPTION_KIND_BOOL)

Dualize strategy for iCrash. Boolean, default False.

ICRASH_EXACT `class-attribute` `instance-attribute` ¶

ICRASH_EXACT = ('icrash_exact', OPTION_KIND_BOOL)

Exact subproblem solution for iCrash. Boolean, default False.

ICRASH_ITERATIONS `class-attribute` `instance-attribute` ¶

ICRASH_ITERATIONS = ('icrash_iterations', OPTION_KIND_INT)

iCrash iterations. Integer, default 30.

ICRASH_STARTING_WEIGHT `class-attribute` `instance-attribute` ¶

ICRASH_STARTING_WEIGHT = (
    "icrash_starting_weight",
    OPTION_KIND_DOUBLE,
)

iCrash starting weight. Double, default 1e-3.

ICRASH_STRATEGY `class-attribute` `instance-attribute` ¶

ICRASH_STRATEGY = ('icrash_strategy', OPTION_KIND_STRING)

Strategy for iCrash. String, default "ICA".

IIS_STRATEGY `class-attribute` `instance-attribute` ¶

IIS_STRATEGY = ('iis_strategy', OPTION_KIND_INT)

Strategy for IIS calculation: 0 => Light test; 1 => Try dual ray; 2 => Try elastic LP; 4 => Prioritise columns; 8 => Find true IIS; 16 => Find relaxation IIS for MIP. Integer, default 0.

IIS_TIME_LIMIT `class-attribute` `instance-attribute` ¶

IIS_TIME_LIMIT = ('iis_time_limit', OPTION_KIND_DOUBLE)

Time limit for computing IIS (seconds). Double, default infinity.

INFINITE_BOUND `class-attribute` `instance-attribute` ¶

INFINITE_BOUND = ('infinite_bound', OPTION_KIND_DOUBLE)

Limit on |constraint bound|: values greater than or equal to this will be treated as infinite. Double, default 1e20.

INFINITE_COST `class-attribute` `instance-attribute` ¶

INFINITE_COST = ('infinite_cost', OPTION_KIND_DOUBLE)

Limit on |cost coefficient|: values greater than or equal to this will be treated as infinite. Double, default 1e20.

IPM_ITERATION_LIMIT `class-attribute` `instance-attribute` ¶

IPM_ITERATION_LIMIT = (
    "ipm_iteration_limit",
    OPTION_KIND_INT,
)

Iteration limit for IPM solver. Integer, default a very large value.

IPM_OPTIMALITY_TOLERANCE `class-attribute` `instance-attribute` ¶

IPM_OPTIMALITY_TOLERANCE = (
    "ipm_optimality_tolerance",
    OPTION_KIND_DOUBLE,
)

IPM optimality tolerance. Double, default 1e-8.

IPX_DUALIZE_STRATEGY `class-attribute` `instance-attribute` ¶

IPX_DUALIZE_STRATEGY = (
    "ipx_dualize_strategy",
    OPTION_KIND_INT,
)

Strategy for dualizing before IPX. Integer, default 2.

KEEP_N_ROWS `class-attribute` `instance-attribute` ¶

KEEP_N_ROWS = ('keep_n_rows', OPTION_KIND_INT)

For multiple N-rows in MPS files: delete rows / delete entries / keep rows (-1/0/1). Integer, default -1.

KKT_TOLERANCE `class-attribute` `instance-attribute` ¶

KKT_TOLERANCE = ('kkt_tolerance', OPTION_KIND_DOUBLE)

If changed from its default value, this tolerance is used for all feasibility and optimality (KKT) measures. Double, default 1e-7.

LARGE_MATRIX_VALUE `class-attribute` `instance-attribute` ¶

LARGE_MATRIX_VALUE = (
    "large_matrix_value",
    OPTION_KIND_DOUBLE,
)

Upper limit on |matrix entries|: values greater than or equal to this will be treated as infinite. Double, default 1e15.

LESS_INFEASIBLE_DSE_CHECK `class-attribute` `instance-attribute` ¶

LESS_INFEASIBLE_DSE_CHECK = (
    "less_infeasible_DSE_check",
    OPTION_KIND_BOOL,
)

Check whether the LP is a candidate for less infeasible dual steepest edge (LiDSE). Boolean, default True.

LESS_INFEASIBLE_DSE_CHOOSE_ROW `class-attribute` `instance-attribute` ¶

LESS_INFEASIBLE_DSE_CHOOSE_ROW = (
    "less_infeasible_DSE_choose_row",
    OPTION_KIND_BOOL,
)

Use LiDSE if the LP has the right properties. Boolean, default True.

LOG_DEV_LEVEL `class-attribute` `instance-attribute` ¶

LOG_DEV_LEVEL = ('log_dev_level', OPTION_KIND_INT)

Output development messages: 0 => none; 1 => info; 2 => detailed; 3 => verbose. Integer, default 0.

LOG_FILE `class-attribute` `instance-attribute` ¶

LOG_FILE = ('log_file', OPTION_KIND_STRING)

Log file. String, default "".

LOG_GITHASH `class-attribute` `instance-attribute` ¶

LOG_GITHASH = ('log_githash', OPTION_KIND_BOOL)

Log the githash. Boolean, default True.

LOG_TO_CONSOLE `class-attribute` `instance-attribute` ¶

LOG_TO_CONSOLE = ('log_to_console', OPTION_KIND_BOOL)

Whether log messages are written to the console. Boolean, default True.

LP_PRESOLVE_REQUIRES_BASIS_POSTSOLVE `class-attribute` `instance-attribute` ¶

LP_PRESOLVE_REQUIRES_BASIS_POSTSOLVE = (
    "lp_presolve_requires_basis_postsolve",
    OPTION_KIND_BOOL,
)

Prevents LP presolve steps for which postsolve cannot maintain a basis. Boolean, default True.

MAX_CENTRING_STEPS `class-attribute` `instance-attribute` ¶

MAX_CENTRING_STEPS = ("max_centring_steps", OPTION_KIND_INT)

Maximum number of steps to use when computing the analytic centre. Integer, default 5.

MAX_DUAL_SIMPLEX_CLEANUP_LEVEL `class-attribute` `instance-attribute` ¶

MAX_DUAL_SIMPLEX_CLEANUP_LEVEL = (
    "max_dual_simplex_cleanup_level",
    OPTION_KIND_INT,
)

Max level of dual simplex cleanup. Integer, default 1.

MAX_DUAL_SIMPLEX_PHASE1_CLEANUP_LEVEL `class-attribute` `instance-attribute` ¶

MAX_DUAL_SIMPLEX_PHASE1_CLEANUP_LEVEL = (
    "max_dual_simplex_phase1_cleanup_level",
    OPTION_KIND_INT,
)

Max level of dual simplex phase 1 cleanup. Integer, default 2.

MIP_ABS_GAP `class-attribute` `instance-attribute` ¶

MIP_ABS_GAP = ('mip_abs_gap', OPTION_KIND_DOUBLE)

Absolute optimality gap at which a mixed integer solve stops. Double, default 1e-6.

MIP_ALLOW_CUT_SEPARATION_AT_NODES `class-attribute` `instance-attribute` ¶

MIP_ALLOW_CUT_SEPARATION_AT_NODES = (
    "mip_allow_cut_separation_at_nodes",
    OPTION_KIND_BOOL,
)

Whether cut separation at nodes other than the root node is permitted. Boolean, default True.

MIP_ALLOW_RESTART `class-attribute` `instance-attribute` ¶

MIP_ALLOW_RESTART = ('mip_allow_restart', OPTION_KIND_BOOL)

Whether MIP restart is permitted. Boolean, default True.

MIP_DETECT_SYMMETRY `class-attribute` `instance-attribute` ¶

MIP_DETECT_SYMMETRY = (
    "mip_detect_symmetry",
    OPTION_KIND_BOOL,
)

Whether MIP symmetry should be detected. Boolean, default True.

MIP_FEASIBILITY_TOLERANCE `class-attribute` `instance-attribute` ¶

MIP_FEASIBILITY_TOLERANCE = (
    "mip_feasibility_tolerance",
    OPTION_KIND_DOUBLE,
)

Feasibility tolerance for integer variables. Double, default 1e-6.

MIP_HEURISTIC_EFFORT `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_EFFORT = (
    "mip_heuristic_effort",
    OPTION_KIND_DOUBLE,
)

Effort spent for MIP heuristics. Double, default 0.05.

MIP_HEURISTIC_RUN_FEASIBILITY_JUMP `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_RUN_FEASIBILITY_JUMP = (
    "mip_heuristic_run_feasibility_jump",
    OPTION_KIND_BOOL,
)

Use the feasibility jump heuristic. Boolean, default True.

MIP_HEURISTIC_RUN_RENS `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_RUN_RENS = (
    "mip_heuristic_run_rens",
    OPTION_KIND_BOOL,
)

Use the RENS heuristic. Boolean, default True.

MIP_HEURISTIC_RUN_RINS `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_RUN_RINS = (
    "mip_heuristic_run_rins",
    OPTION_KIND_BOOL,
)

Use the RINS heuristic. Boolean, default True.

MIP_HEURISTIC_RUN_ROOT_REDUCED_COST `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_RUN_ROOT_REDUCED_COST = (
    "mip_heuristic_run_root_reduced_cost",
    OPTION_KIND_BOOL,
)

Use the rootReducedCost heuristic. Boolean, default True.

MIP_HEURISTIC_RUN_SHIFTING `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_RUN_SHIFTING = (
    "mip_heuristic_run_shifting",
    OPTION_KIND_BOOL,
)

Use the Shifting heuristic. Boolean, default False.

MIP_HEURISTIC_RUN_ZI_ROUND `class-attribute` `instance-attribute` ¶

MIP_HEURISTIC_RUN_ZI_ROUND = (
    "mip_heuristic_run_zi_round",
    OPTION_KIND_BOOL,
)

Use the ZI Round heuristic. Boolean, default False.

MIP_IMPROVING_SOLUTION_FILE `class-attribute` `instance-attribute` ¶

MIP_IMPROVING_SOLUTION_FILE = (
    "mip_improving_solution_file",
    OPTION_KIND_STRING,
)

File for reporting improving MIP solutions. Not reported for an empty string. String, default "".

MIP_IMPROVING_SOLUTION_REPORT_SPARSE `class-attribute` `instance-attribute` ¶

MIP_IMPROVING_SOLUTION_REPORT_SPARSE = (
    "mip_improving_solution_report_sparse",
    OPTION_KIND_BOOL,
)

Whether improving MIP solutions should be reported in sparse format. Boolean, default False.

MIP_IMPROVING_SOLUTION_SAVE `class-attribute` `instance-attribute` ¶

MIP_IMPROVING_SOLUTION_SAVE = (
    "mip_improving_solution_save",
    OPTION_KIND_BOOL,
)

Whether improving MIP solutions should be saved. Boolean, default False.

MIP_IPM_SOLVER `class-attribute` `instance-attribute` ¶

MIP_IPM_SOLVER = ('mip_ipm_solver', OPTION_KIND_STRING)

MIP IPM solver: "choose", "ipx" or "hipo". String, default "choose".

MIP_LIFTING_FOR_PROBING `class-attribute` `instance-attribute` ¶

MIP_LIFTING_FOR_PROBING = (
    "mip_lifting_for_probing",
    OPTION_KIND_INT,
)

Level of lifting for probing that is used. Integer, default -1.

MIP_LP_AGE_LIMIT `class-attribute` `instance-attribute` ¶

MIP_LP_AGE_LIMIT = ('mip_lp_age_limit', OPTION_KIND_INT)

Maximal age of dynamic LP rows before they are removed from the LP relaxation in the MIP solver. Integer, default 10.

MIP_LP_SOLVER `class-attribute` `instance-attribute` ¶

MIP_LP_SOLVER = ('mip_lp_solver', OPTION_KIND_STRING)

MIP LP solver: "choose", "simplex", "ipm", "ipx" or "hipo". String, default "choose".

MIP_MAX_IMPROVING_SOLS `class-attribute` `instance-attribute` ¶

MIP_MAX_IMPROVING_SOLS = (
    "mip_max_improving_sols",
    OPTION_KIND_INT,
)

Limit on the number of improving solutions found to stop the MIP solver prematurely. Integer, default a very large value.

MIP_MAX_LEAVES `class-attribute` `instance-attribute` ¶

MIP_MAX_LEAVES = ('mip_max_leaves', OPTION_KIND_INT)

MIP solver max number of leaf nodes. Integer, default a very large value.

MIP_MAX_NODES `class-attribute` `instance-attribute` ¶

MIP_MAX_NODES = ('mip_max_nodes', OPTION_KIND_INT)

Maximum number of branch and bound nodes. Integer, default a very large value.

MIP_MAX_STALL_NODES `class-attribute` `instance-attribute` ¶

MIP_MAX_STALL_NODES = (
    "mip_max_stall_nodes",
    OPTION_KIND_INT,
)

MIP solver max number of nodes where estimate is above cutoff bound. Integer, default a very large value.

MIP_MAX_START_NODES `class-attribute` `instance-attribute` ¶

MIP_MAX_START_NODES = (
    "mip_max_start_nodes",
    OPTION_KIND_INT,
)

MIP solver max number of nodes when completing a partial MIP start. Integer, default 500.

MIP_MIN_CLIQUETABLE_ENTRIES_FOR_PARALLELISM `class-attribute` `instance-attribute` ¶

MIP_MIN_CLIQUETABLE_ENTRIES_FOR_PARALLELISM = (
    "mip_min_cliquetable_entries_for_parallelism",
    OPTION_KIND_INT,
)

Minimal number of entries in the MIP solver cliquetable before neighbourhood queries of the conflict graph use parallel processing. Integer, default 100000.

MIP_MIN_LOGGING_INTERVAL `class-attribute` `instance-attribute` ¶

MIP_MIN_LOGGING_INTERVAL = (
    "mip_min_logging_interval",
    OPTION_KIND_DOUBLE,
)

MIP minimum logging interval. Double, default 5.

MIP_POOL_AGE_LIMIT `class-attribute` `instance-attribute` ¶

MIP_POOL_AGE_LIMIT = ("mip_pool_age_limit", OPTION_KIND_INT)

Maximal age of rows in the MIP solver cutpool before they are deleted. Integer, default 30.

MIP_POOL_SOFT_LIMIT `class-attribute` `instance-attribute` ¶

MIP_POOL_SOFT_LIMIT = (
    "mip_pool_soft_limit",
    OPTION_KIND_INT,
)

Soft limit on the number of rows in the MIP solver cutpool for dynamic age adjustment. Integer, default 10000.

MIP_PSCOST_MINRELIABLE `class-attribute` `instance-attribute` ¶

MIP_PSCOST_MINRELIABLE = (
    "mip_pscost_minreliable",
    OPTION_KIND_INT,
)

Minimal number of observations before MIP solver pseudo costs are considered reliable. Integer, default 8.

MIP_REL_GAP `class-attribute` `instance-attribute` ¶

MIP_REL_GAP = ('mip_rel_gap', OPTION_KIND_DOUBLE)

Relative optimality gap at which a mixed integer solve stops. Double, default 1e-4.

MIP_REPORT_LEVEL `class-attribute` `instance-attribute` ¶

MIP_REPORT_LEVEL = ('mip_report_level', OPTION_KIND_INT)

MIP solver reporting level. Integer, default 1.

MIP_ROOT_PRESOLVE_ONLY `class-attribute` `instance-attribute` ¶

MIP_ROOT_PRESOLVE_ONLY = (
    "mip_root_presolve_only",
    OPTION_KIND_BOOL,
)

Whether MIP presolve is only applied at the root node. Boolean, default False.

MIP_SEARCH_SIMULATE_CONCURRENCY `class-attribute` `instance-attribute` ¶

MIP_SEARCH_SIMULATE_CONCURRENCY = (
    "mip_search_simulate_concurrency",
    OPTION_KIND_BOOL,
)

Simulate MIP search concurrency on a single thread. Boolean, default False.

MPS_PARSER_TYPE_FREE `class-attribute` `instance-attribute` ¶

MPS_PARSER_TYPE_FREE = (
    "mps_parser_type_free",
    OPTION_KIND_BOOL,
)

Use the free format MPS file reader. Boolean, default True.

NO_UNNECESSARY_REBUILD_REFACTOR `class-attribute` `instance-attribute` ¶

NO_UNNECESSARY_REBUILD_REFACTOR = (
    "no_unnecessary_rebuild_refactor",
    OPTION_KIND_BOOL,
)

No unnecessary refactorization on simplex rebuild. Boolean, default True.

OBJECTIVE_BOUND `class-attribute` `instance-attribute` ¶

OBJECTIVE_BOUND = ('objective_bound', OPTION_KIND_DOUBLE)

Bound on the objective that stops the solve when reached. Double, default infinity.

OBJECTIVE_TARGET `class-attribute` `instance-attribute` ¶

OBJECTIVE_TARGET = ('objective_target', OPTION_KIND_DOUBLE)

Target objective value that stops the solve when reached. Double, default negative infinity.

OPTIMALITY_TOLERANCE `class-attribute` `instance-attribute` ¶

OPTIMALITY_TOLERANCE = (
    "optimality_tolerance",
    OPTION_KIND_DOUBLE,
)

Optimality tolerance. Double, default 1e-7.

OUTPUT_FLAG `class-attribute` `instance-attribute` ¶

OUTPUT_FLAG = ('output_flag', OPTION_KIND_BOOL)

Master switch for all HiGHS logging. Set to True to let the solver print its log. Boolean; HiGHS itself defaults to True, but solve_linear_problem defaults it to False so the solver is silent unless requested.

PARALLEL `class-attribute` `instance-attribute` ¶

PARALLEL = ('parallel', OPTION_KIND_STRING)

Parallelism control. One of "off", "choose" or "on". String, default "choose".

PDLP_CUPDLPC_RESTART_METHOD `class-attribute` `instance-attribute` ¶

PDLP_CUPDLPC_RESTART_METHOD = (
    "pdlp_cupdlpc_restart_method",
    OPTION_KIND_INT,
)

Restart mode for cuPDLP-C solver: 0 => none; 1 => GPU (default); 2 => CPU. Integer, default 1.

PDLP_ITERATION_LIMIT `class-attribute` `instance-attribute` ¶

PDLP_ITERATION_LIMIT = (
    "pdlp_iteration_limit",
    OPTION_KIND_INT,
)

Iteration limit for PDLP solver. Integer, default a very large value.

PDLP_OPTIMALITY_TOLERANCE `class-attribute` `instance-attribute` ¶

PDLP_OPTIMALITY_TOLERANCE = (
    "pdlp_optimality_tolerance",
    OPTION_KIND_DOUBLE,
)

PDLP optimality tolerance. Double, default 1e-7.

PDLP_RESTART_STRATEGY `class-attribute` `instance-attribute` ¶

PDLP_RESTART_STRATEGY = (
    "pdlp_restart_strategy",
    OPTION_KIND_INT,
)

Restart strategy for PDLP solver: 0 => off; 1 => fixed; 2 => adaptive; 3 => Halpern. Integer, default 2.

PDLP_RUIZ_ITERATIONS `class-attribute` `instance-attribute` ¶

PDLP_RUIZ_ITERATIONS = (
    "pdlp_ruiz_iterations",
    OPTION_KIND_INT,
)

Number of Ruiz scaling iteraitons for PDLP solver. Integer, default 10.

PDLP_SCALING_MODE `class-attribute` `instance-attribute` ¶

PDLP_SCALING_MODE = ('pdlp_scaling_mode', OPTION_KIND_INT)

Scaling mode for PDLP solver (default = 5): 1 => Ruiz; 2 => L2; 4 => PC. Integer, default 5.

PDLP_STEP_SIZE_STRATEGY `class-attribute` `instance-attribute` ¶

PDLP_STEP_SIZE_STRATEGY = (
    "pdlp_step_size_strategy",
    OPTION_KIND_INT,
)

Step size strategy for PDLP solver: 0 => fixed; 1 => adaptive; 2 => Malitsky-Pock; 3 => PID. Integer, default 1.

PRESOLVE `class-attribute` `instance-attribute` ¶

PRESOLVE = ('presolve', OPTION_KIND_STRING)

Presolve control. One of "off", "choose" or "on". String, default "choose".

PRESOLVE_PIVOT_THRESHOLD `class-attribute` `instance-attribute` ¶

PRESOLVE_PIVOT_THRESHOLD = (
    "presolve_pivot_threshold",
    OPTION_KIND_DOUBLE,
)

Matrix factorization pivot threshold for substitutions in presolve. Double, default 0.01.

PRESOLVE_REDUCTION_LIMIT `class-attribute` `instance-attribute` ¶

PRESOLVE_REDUCTION_LIMIT = (
    "presolve_reduction_limit",
    OPTION_KIND_INT,
)

Limit on number of presolve reductions. -1 means no limit. Integer, default -1.

PRESOLVE_REMOVE_SLACKS `class-attribute` `instance-attribute` ¶

PRESOLVE_REMOVE_SLACKS = (
    "presolve_remove_slacks",
    OPTION_KIND_BOOL,
)

Remove slacks after presolve. Boolean, default False.

PRESOLVE_RULE_LOGGING `class-attribute` `instance-attribute` ¶

PRESOLVE_RULE_LOGGING = (
    "presolve_rule_logging",
    OPTION_KIND_BOOL,
)

Log effectiveness of presolve rules for LP. Boolean, default False.

PRESOLVE_RULE_OFF `class-attribute` `instance-attribute` ¶

PRESOLVE_RULE_OFF = ('presolve_rule_off', OPTION_KIND_INT)

Bit mask of presolve rules that are not allowed. Build the mask with PresolveRule.mask, for example PresolveRule.mask(PresolveRule.PROBING, PresolveRule.SPARSIFY). Integer, default 0 (every rule allowed).

PRESOLVE_RULE_TEST `class-attribute` `instance-attribute` ¶

PRESOLVE_RULE_TEST = ("presolve_rule_test", OPTION_KIND_INT)

A single PresolveRule value to test in isolation. Development and debugging use only. Integer, default 0.

PRESOLVE_SUBSTITUTION_MAXFILLIN `class-attribute` `instance-attribute` ¶

PRESOLVE_SUBSTITUTION_MAXFILLIN = (
    "presolve_substitution_maxfillin",
    OPTION_KIND_INT,
)

Maximal fillin allowed for substitutions in presolve. Integer, default 10.

PRIMAL_FEASIBILITY_TOLERANCE `class-attribute` `instance-attribute` ¶

PRIMAL_FEASIBILITY_TOLERANCE = (
    "primal_feasibility_tolerance",
    OPTION_KIND_DOUBLE,
)

Tolerance for primal feasibility. Double, default 1e-7.

PRIMAL_RESIDUAL_TOLERANCE `class-attribute` `instance-attribute` ¶

PRIMAL_RESIDUAL_TOLERANCE = (
    "primal_residual_tolerance",
    OPTION_KIND_DOUBLE,
)

Primal residual tolerance. Double, default 1e-7.

PRIMAL_SIMPLEX_BOUND_PERTURBATION_MULTIPLIER `class-attribute` `instance-attribute` ¶

PRIMAL_SIMPLEX_BOUND_PERTURBATION_MULTIPLIER = (
    "primal_simplex_bound_perturbation_multiplier",
    OPTION_KIND_DOUBLE,
)

Primal simplex bound perturbation multiplier. 0 means no perturbation. Double, default 1.0.

QP_ALLOW_HOT_START `class-attribute` `instance-attribute` ¶

QP_ALLOW_HOT_START = (
    "qp_allow_hot_start",
    OPTION_KIND_BOOL,
)

Allow the active set QP solver to hot start. Boolean, default False.

QP_ITERATION_LIMIT `class-attribute` `instance-attribute` ¶

QP_ITERATION_LIMIT = ("qp_iteration_limit", OPTION_KIND_INT)

Iteration limit for the active set QP solver. Integer, default a very large value.

QP_NULLSPACE_LIMIT `class-attribute` `instance-attribute` ¶

QP_NULLSPACE_LIMIT = ("qp_nullspace_limit", OPTION_KIND_INT)

Nullspace limit for the active set QP solver. Integer, default 4000.

QP_REGULARIZATION_VALUE `class-attribute` `instance-attribute` ¶

QP_REGULARIZATION_VALUE = (
    "qp_regularization_value",
    OPTION_KIND_DOUBLE,
)

Regularization value added to the Hessian in the active set QP solver. Double, default 1e-7.

RANDOM_SEED `class-attribute` `instance-attribute` ¶

RANDOM_SEED = ('random_seed', OPTION_KIND_INT)

Seed for the random number generator. Integer, default 0.

RANGING `class-attribute` `instance-attribute` ¶

RANGING = ('ranging', OPTION_KIND_STRING)

Compute cost, bound, RHS and basic solution ranging: "off" or "on". String, default "off".

READ_BASIS_FILE `class-attribute` `instance-attribute` ¶

READ_BASIS_FILE = ('read_basis_file', OPTION_KIND_STRING)

Read basis file. String, default "".

READ_SOLUTION_FILE `class-attribute` `instance-attribute` ¶

READ_SOLUTION_FILE = (
    "read_solution_file",
    OPTION_KIND_STRING,
)

Read solution file. String, default "".

REBUILD_REFACTOR_SOLUTION_ERROR_TOLERANCE `class-attribute` `instance-attribute` ¶

REBUILD_REFACTOR_SOLUTION_ERROR_TOLERANCE = (
    "rebuild_refactor_solution_error_tolerance",
    OPTION_KIND_DOUBLE,
)

Tolerance on solution error when considering refactorization on simplex rebuild. Double, default 1e-8.

RESTART_PRESOLVE_REDUCTION_LIMIT `class-attribute` `instance-attribute` ¶

RESTART_PRESOLVE_REDUCTION_LIMIT = (
    "restart_presolve_reduction_limit",
    OPTION_KIND_INT,
)

Limit on number of further presolve reductions on restart in the MIP solver. -1 means no limit. Integer, default -1.

RUN_CENTRING `class-attribute` `instance-attribute` ¶

RUN_CENTRING = ('run_centring', OPTION_KIND_BOOL)

Perform centring steps to compute the analytic centre before crossover. Boolean, default False.

RUN_CROSSOVER `class-attribute` `instance-attribute` ¶

RUN_CROSSOVER = ('run_crossover', OPTION_KIND_STRING)

Whether to run crossover after an interior point solve. One of "off", "choose" or "on". String, default "on".

SIMPLEX_CRASH_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_CRASH_STRATEGY = (
    "simplex_crash_strategy",
    OPTION_KIND_INT,
)

Strategy for simplex crash: off / LTSSF / Bixby (0/½). Integer, default 0.

SIMPLEX_DUALIZE_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_DUALIZE_STRATEGY = (
    "simplex_dualize_strategy",
    OPTION_KIND_INT,
)

Strategy for dualizing before simplex: off / on (-1/1). Integer, default -1.

SIMPLEX_DUAL_EDGE_WEIGHT_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_DUAL_EDGE_WEIGHT_STRATEGY = (
    "simplex_dual_edge_weight_strategy",
    OPTION_KIND_INT,
)

Strategy for simplex dual edge weights: Choose / Dantzig / Devex / Steepest Edge (-1/0/½). Integer, default -1.

SIMPLEX_ITERATION_LIMIT `class-attribute` `instance-attribute` ¶

SIMPLEX_ITERATION_LIMIT = (
    "simplex_iteration_limit",
    OPTION_KIND_INT,
)

Iteration limit for simplex solver when solving LPs, but not subproblems in the MIP solver. Integer, default a very large value.

SIMPLEX_MAX_CONCURRENCY `class-attribute` `instance-attribute` ¶

SIMPLEX_MAX_CONCURRENCY = (
    "simplex_max_concurrency",
    OPTION_KIND_INT,
)

Maximum level of concurrency in parallel simplex. Integer, default 8.

SIMPLEX_MIN_CONCURRENCY `class-attribute` `instance-attribute` ¶

SIMPLEX_MIN_CONCURRENCY = (
    "simplex_min_concurrency",
    OPTION_KIND_INT,
)

Minimum level of concurrency in parallel simplex. Integer, default 1.

SIMPLEX_PERMUTE_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_PERMUTE_STRATEGY = (
    "simplex_permute_strategy",
    OPTION_KIND_INT,
)

Strategy for permuting before simplex: off / on (-1/1). Integer, default -1.

SIMPLEX_PRICE_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_PRICE_STRATEGY = (
    "simplex_price_strategy",
    OPTION_KIND_INT,
)

Strategy for PRICE in simplex. Integer, default 3.

SIMPLEX_PRIMAL_EDGE_WEIGHT_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_PRIMAL_EDGE_WEIGHT_STRATEGY = (
    "simplex_primal_edge_weight_strategy",
    OPTION_KIND_INT,
)

Strategy for simplex primal edge weights: Choose / Dantzig / Devex / Steepest Edge (-1/0/½). Integer, default -1.

SIMPLEX_SCALE_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_SCALE_STRATEGY = (
    "simplex_scale_strategy",
    OPTION_KIND_INT,
)

Simplex scaling strategy: off / choose / equilibration (default) / forced equilibration / max value (0/½/¾). Integer, default 2.

SIMPLEX_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_STRATEGY = ('simplex_strategy', OPTION_KIND_INT)

Strategy for simplex solver 0 => Choose; 1 => Dual (serial); 2 => Dual (SIP); 3 => Dual (PAMI); 4 => Primal. Integer, default 1.

SIMPLEX_UNSCALED_SOLUTION_STRATEGY `class-attribute` `instance-attribute` ¶

SIMPLEX_UNSCALED_SOLUTION_STRATEGY = (
    "simplex_unscaled_solution_strategy",
    OPTION_KIND_INT,
)

Strategy for solving the unscaled LP in simplex. Integer, default 1.

SIMPLEX_UPDATE_LIMIT `class-attribute` `instance-attribute` ¶

SIMPLEX_UPDATE_LIMIT = (
    "simplex_update_limit",
    OPTION_KIND_INT,
)

Limit on the number of simplex UPDATE operations. Integer, default 5000.

SMALL_MATRIX_VALUE `class-attribute` `instance-attribute` ¶

SMALL_MATRIX_VALUE = (
    "small_matrix_value",
    OPTION_KIND_DOUBLE,
)

Lower limit on |matrix entries|: values less than or equal to this will be treated as zero. Double, default 1e-9.

SOLUTION_FILE `class-attribute` `instance-attribute` ¶

SOLUTION_FILE = ('solution_file', OPTION_KIND_STRING)

Write solution file. String, default "".

SOLVER `class-attribute` `instance-attribute` ¶

SOLVER = ('solver', OPTION_KIND_STRING)

Which algorithm to use. One of "choose", "simplex", "ipm", "pdlp" or "hipo". String, default "choose".

SOLVE_RELAXATION `class-attribute` `instance-attribute` ¶

SOLVE_RELAXATION = ('solve_relaxation', OPTION_KIND_BOOL)

Solve the relaxation of discrete model components instead of the discrete model itself. Boolean, default False.

START_CROSSOVER_TOLERANCE `class-attribute` `instance-attribute` ¶

START_CROSSOVER_TOLERANCE = (
    "start_crossover_tolerance",
    OPTION_KIND_DOUBLE,
)

Tolerance to be satisfied before IPM crossover will start. Double, default 1e-8.

THREADS `class-attribute` `instance-attribute` ¶

THREADS = ('threads', OPTION_KIND_INT)

Number of threads to use. Zero means let HiGHS choose. Integer, default 0.

TIMELESS_LOG `class-attribute` `instance-attribute` ¶

TIMELESS_LOG = ('timeless_log', OPTION_KIND_BOOL)

Suppression of time-based data in logging. Boolean, default False.

TIME_LIMIT `class-attribute` `instance-attribute` ¶

TIME_LIMIT = ('time_limit', OPTION_KIND_DOUBLE)

Wall clock time limit for the solve in seconds. Double, default infinity.

USER_BOUND_SCALE `class-attribute` `instance-attribute` ¶

USER_BOUND_SCALE = ('user_bound_scale', OPTION_KIND_INT)

Exponent of power-of-two bound scaling for model. Integer, default 0.

USER_OBJECTIVE_SCALE `class-attribute` `instance-attribute` ¶

USER_OBJECTIVE_SCALE = (
    "user_objective_scale",
    OPTION_KIND_INT,
)

Exponent of power-of-two objective scaling for model. Integer, default 0.

USE_IMPLIED_BOUNDS_FROM_PRESOLVE `class-attribute` `instance-attribute` ¶

USE_IMPLIED_BOUNDS_FROM_PRESOLVE = (
    "use_implied_bounds_from_presolve",
    OPTION_KIND_BOOL,
)

Use relaxed implied bounds from presolve. Boolean, default False.

USE_ORIGINAL_HFACTOR_LOGIC `class-attribute` `instance-attribute` ¶

USE_ORIGINAL_HFACTOR_LOGIC = (
    "use_original_HFactor_logic",
    OPTION_KIND_BOOL,
)

Use original HFactor logic for sparse vs hyper-sparse TRANs. Boolean, default True.

USE_WARM_START `class-attribute` `instance-attribute` ¶

USE_WARM_START = ('use_warm_start', OPTION_KIND_BOOL)

Use any warm start that is available. Boolean, default True.

WRITE_BASIS_FILE `class-attribute` `instance-attribute` ¶

WRITE_BASIS_FILE = ('write_basis_file', OPTION_KIND_STRING)

Write basis file. String, default "".

WRITE_HESSIAN_IMAGE `class-attribute` `instance-attribute` ¶

WRITE_HESSIAN_IMAGE = (
    "write_hessian_image",
    OPTION_KIND_BOOL,
)

Write an image of the Hessian to a file. Boolean, default False.

WRITE_IIS_MODEL_FILE `class-attribute` `instance-attribute` ¶

WRITE_IIS_MODEL_FILE = (
    "write_iis_model_file",
    OPTION_KIND_STRING,
)

Write IIS model file. String, default "".

WRITE_MATRIX_IMAGE `class-attribute` `instance-attribute` ¶

WRITE_MATRIX_IMAGE = (
    "write_matrix_image",
    OPTION_KIND_BOOL,
)

Write an image of the constraint matrix to a file. Boolean, default False.

WRITE_MODEL_FILE `class-attribute` `instance-attribute` ¶

WRITE_MODEL_FILE = ('write_model_file', OPTION_KIND_STRING)

Write model file. String, default "".

WRITE_MODEL_TO_FILE `class-attribute` `instance-attribute` ¶

WRITE_MODEL_TO_FILE = (
    "write_model_to_file",
    OPTION_KIND_BOOL,
)

Write the model to a file. Boolean, default False.

WRITE_PRESOLVED_MODEL_FILE `class-attribute` `instance-attribute` ¶

WRITE_PRESOLVED_MODEL_FILE = (
    "write_presolved_model_file",
    OPTION_KIND_STRING,
)

Write presolved model file. String, default "".

WRITE_PRESOLVED_MODEL_TO_FILE `class-attribute` `instance-attribute` ¶

WRITE_PRESOLVED_MODEL_TO_FILE = (
    "write_presolved_model_to_file",
    OPTION_KIND_BOOL,
)

Write the presolved model to a file. Boolean, default False.

WRITE_SOLUTION_STYLE `class-attribute` `instance-attribute` ¶

WRITE_SOLUTION_STYLE = (
    "write_solution_style",
    OPTION_KIND_INT,
)

Style of solution file (raw = computer-readable, pretty = human-readable): -1 => HiGHS old raw (deprecated); 0 => HiGHS raw; 1 => HiGHS pretty; 2 => Glpsol raw; 3 => Glpsol pretty; 4 => HiGHS sparse raw. Integer, default 0.

WRITE_SOLUTION_TO_FILE `class-attribute` `instance-attribute` ¶

WRITE_SOLUTION_TO_FILE = (
    "write_solution_to_file",
    OPTION_KIND_BOOL,
)

Write the primal and dual solution to a file. Boolean, default False.

option_name `property` ¶

option_name: str

Return the option name string expected by the HiGHS C API.

value_kind `property` ¶

value_kind: str

Return the value kind label for this option.

Library information¶

cyhighs.highs_version ¶

highs_version()

Return the version string of the linked HiGHS library.

Returns:

Type	Description
	The HiGHS version, for example `"1.15.1"`.

cyhighs.highs_infinity ¶

highs_infinity()

Return the value HiGHS uses to represent infinity.

Any bound whose magnitude is greater than or equal to this value is treated by HiGHS as unbounded.

Returns:

Type	Description
	The infinity threshold, typically `1e30`.

API Reference¶

Solver interfaces¶

cyhighs.solve_linear_problem ¶

cyhighs.solve_linear_problem_sparse ¶

cyhighs.linprog ¶

The solution record¶

cyhighs.LinearProblemSolution ¶

is_optimal property ¶

cyhighs.OptimizeResult ¶

Enumerations¶

cyhighs.ObjectiveSense ¶

MAXIMIZE class-attribute instance-attribute ¶

MINIMIZE class-attribute instance-attribute ¶

cyhighs.VariableType ¶

CONTINUOUS class-attribute instance-attribute ¶

IMPLICIT_INTEGER class-attribute instance-attribute ¶

INTEGER class-attribute instance-attribute ¶

SEMI_CONTINUOUS class-attribute instance-attribute ¶

SEMI_INTEGER class-attribute instance-attribute ¶

cyhighs.ModelStatus ¶

INFEASIBLE class-attribute instance-attribute ¶

INTERRUPT class-attribute instance-attribute ¶

ITERATION_LIMIT class-attribute instance-attribute ¶

LOAD_ERROR class-attribute instance-attribute ¶

MODEL_EMPTY class-attribute instance-attribute ¶

MODEL_ERROR class-attribute instance-attribute ¶

NOTSET class-attribute instance-attribute ¶

OBJECTIVE_BOUND class-attribute instance-attribute ¶

OBJECTIVE_TARGET class-attribute instance-attribute ¶

OPTIMAL class-attribute instance-attribute ¶

POSTSOLVE_ERROR class-attribute instance-attribute ¶

PRESOLVE_ERROR class-attribute instance-attribute ¶

SOLUTION_LIMIT class-attribute instance-attribute ¶

SOLVE_ERROR class-attribute instance-attribute ¶

TIME_LIMIT class-attribute instance-attribute ¶

UNBOUNDED class-attribute instance-attribute ¶

UNBOUNDED_OR_INFEASIBLE class-attribute instance-attribute ¶

UNKNOWN class-attribute instance-attribute ¶

cyhighs.PresolveRule ¶

AGGREGATOR class-attribute instance-attribute ¶

COL_STUFFING class-attribute instance-attribute ¶

DEPENDENT_EQUATIONS class-attribute instance-attribute ¶

DEPENDENT_FREE_COLS class-attribute instance-attribute ¶

DOMINATED_COL class-attribute instance-attribute ¶

DOUBLETON_EQUATION class-attribute instance-attribute ¶

DUAL_FIXING class-attribute instance-attribute ¶

EMPTY_COL class-attribute instance-attribute ¶

EMPTY_ROW class-attribute instance-attribute ¶

ENUMERATION class-attribute instance-attribute ¶

FIXED_COL class-attribute instance-attribute ¶

FORCING_COL class-attribute instance-attribute ¶

FORCING_ROW class-attribute instance-attribute ¶

FREE_COL_SUBSTITUTION class-attribute instance-attribute ¶

INITIAL_SWEEP class-attribute instance-attribute ¶

PARALLEL_ROWS_AND_COLS class-attribute instance-attribute ¶

PROBING class-attribute instance-attribute ¶

REDUNDANT_ROW class-attribute instance-attribute ¶

SINGLETON_ROW class-attribute instance-attribute ¶

SPARSIFY class-attribute instance-attribute ¶

mask classmethod ¶

Options¶

cyhighs.HighsOption ¶

ALLOWED_COST_SCALE_FACTOR class-attribute instance-attribute ¶

ALLOWED_MATRIX_SCALE_FACTOR class-attribute instance-attribute ¶

ALLOW_UNBOUNDED_OR_INFEASIBLE class-attribute instance-attribute ¶

BLEND_MULTI_OBJECTIVES class-attribute instance-attribute ¶

CENTRING_RATIO_TOLERANCE class-attribute instance-attribute ¶

COST_SCALE_FACTOR class-attribute instance-attribute ¶

DUAL_FEASIBILITY_TOLERANCE class-attribute instance-attribute ¶

DUAL_RESIDUAL_TOLERANCE class-attribute instance-attribute ¶

DUAL_SIMPLEX_COST_PERTURBATION_MULTIPLIER class-attribute instance-attribute ¶

DUAL_SIMPLEX_PIVOT_GROWTH_TOLERANCE class-attribute instance-attribute ¶

DUAL_STEEPEST_EDGE_WEIGHT_ERROR_TOLERANCE class-attribute instance-attribute ¶

DUAL_STEEPEST_EDGE_WEIGHT_LOG_ERROR_THRESHOLD class-attribute instance-attribute ¶

FACTOR_PIVOT_THRESHOLD class-attribute instance-attribute ¶

FACTOR_PIVOT_TOLERANCE class-attribute instance-attribute ¶

GLPSOL_COST_ROW_LOCATION class-attribute instance-attribute ¶

HIGHS_ANALYSIS_LEVEL class-attribute instance-attribute ¶

HIGHS_DEBUG_LEVEL class-attribute instance-attribute ¶

HIPO_BLOCK_SIZE class-attribute instance-attribute ¶

is_optimal `property` ¶

MAXIMIZE `class-attribute` `instance-attribute` ¶

MINIMIZE `class-attribute` `instance-attribute` ¶

CONTINUOUS `class-attribute` `instance-attribute` ¶

IMPLICIT_INTEGER `class-attribute` `instance-attribute` ¶

INTEGER `class-attribute` `instance-attribute` ¶

SEMI_CONTINUOUS `class-attribute` `instance-attribute` ¶

SEMI_INTEGER `class-attribute` `instance-attribute` ¶

INFEASIBLE `class-attribute` `instance-attribute` ¶

INTERRUPT `class-attribute` `instance-attribute` ¶

ITERATION_LIMIT `class-attribute` `instance-attribute` ¶

LOAD_ERROR `class-attribute` `instance-attribute` ¶

MODEL_EMPTY `class-attribute` `instance-attribute` ¶

MODEL_ERROR `class-attribute` `instance-attribute` ¶

NOTSET `class-attribute` `instance-attribute` ¶

OBJECTIVE_BOUND `class-attribute` `instance-attribute` ¶

OBJECTIVE_TARGET `class-attribute` `instance-attribute` ¶

OPTIMAL `class-attribute` `instance-attribute` ¶

POSTSOLVE_ERROR `class-attribute` `instance-attribute` ¶

PRESOLVE_ERROR `class-attribute` `instance-attribute` ¶

SOLUTION_LIMIT `class-attribute` `instance-attribute` ¶

SOLVE_ERROR `class-attribute` `instance-attribute` ¶

TIME_LIMIT `class-attribute` `instance-attribute` ¶

UNBOUNDED `class-attribute` `instance-attribute` ¶

UNBOUNDED_OR_INFEASIBLE `class-attribute` `instance-attribute` ¶

UNKNOWN `class-attribute` `instance-attribute` ¶

AGGREGATOR `class-attribute` `instance-attribute` ¶

COL_STUFFING `class-attribute` `instance-attribute` ¶

DEPENDENT_EQUATIONS `class-attribute` `instance-attribute` ¶

DEPENDENT_FREE_COLS `class-attribute` `instance-attribute` ¶

DOMINATED_COL `class-attribute` `instance-attribute` ¶

DOUBLETON_EQUATION `class-attribute` `instance-attribute` ¶

DUAL_FIXING `class-attribute` `instance-attribute` ¶

EMPTY_COL `class-attribute` `instance-attribute` ¶

EMPTY_ROW `class-attribute` `instance-attribute` ¶

ENUMERATION `class-attribute` `instance-attribute` ¶

FIXED_COL `class-attribute` `instance-attribute` ¶

FORCING_COL `class-attribute` `instance-attribute` ¶

FORCING_ROW `class-attribute` `instance-attribute` ¶

FREE_COL_SUBSTITUTION `class-attribute` `instance-attribute` ¶

INITIAL_SWEEP `class-attribute` `instance-attribute` ¶

PARALLEL_ROWS_AND_COLS `class-attribute` `instance-attribute` ¶

PROBING `class-attribute` `instance-attribute` ¶

REDUNDANT_ROW `class-attribute` `instance-attribute` ¶

SINGLETON_ROW `class-attribute` `instance-attribute` ¶

SPARSIFY `class-attribute` `instance-attribute` ¶

mask `classmethod` ¶

ALLOWED_COST_SCALE_FACTOR `class-attribute` `instance-attribute` ¶

ALLOWED_MATRIX_SCALE_FACTOR `class-attribute` `instance-attribute` ¶

ALLOW_UNBOUNDED_OR_INFEASIBLE `class-attribute` `instance-attribute` ¶

BLEND_MULTI_OBJECTIVES `class-attribute` `instance-attribute` ¶

CENTRING_RATIO_TOLERANCE `class-attribute` `instance-attribute` ¶

COST_SCALE_FACTOR `class-attribute` `instance-attribute` ¶

DUAL_FEASIBILITY_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_RESIDUAL_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_SIMPLEX_COST_PERTURBATION_MULTIPLIER `class-attribute` `instance-attribute` ¶

DUAL_SIMPLEX_PIVOT_GROWTH_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_STEEPEST_EDGE_WEIGHT_ERROR_TOLERANCE `class-attribute` `instance-attribute` ¶

DUAL_STEEPEST_EDGE_WEIGHT_LOG_ERROR_THRESHOLD `class-attribute` `instance-attribute` ¶

FACTOR_PIVOT_THRESHOLD `class-attribute` `instance-attribute` ¶

FACTOR_PIVOT_TOLERANCE `class-attribute` `instance-attribute` ¶

GLPSOL_COST_ROW_LOCATION `class-attribute` `instance-attribute` ¶

HIGHS_ANALYSIS_LEVEL `class-attribute` `instance-attribute` ¶

HIGHS_DEBUG_LEVEL `class-attribute` `instance-attribute` ¶

HIPO_BLOCK_SIZE `class-attribute` `instance-attribute` ¶

HIPO_ORDERING `class-attribute` `instance-attribute` ¶

HIPO_PARALLEL_TYPE `class-attribute` `instance-attribute` ¶

HIPO_SYSTEM `class-attribute` `instance-attribute` ¶

ICRASH `class-attribute` `instance-attribute` ¶

ICRASH_APPROX_ITER `class-attribute` `instance-attribute` ¶

ICRASH_BREAKPOINTS `class-attribute` `instance-attribute` ¶

ICRASH_DUALIZE `class-attribute` `instance-attribute` ¶

ICRASH_EXACT `class-attribute` `instance-attribute` ¶

ICRASH_ITERATIONS `class-attribute` `instance-attribute` ¶

ICRASH_STARTING_WEIGHT `class-attribute` `instance-attribute` ¶

ICRASH_STRATEGY `class-attribute` `instance-attribute` ¶

IIS_STRATEGY `class-attribute` `instance-attribute` ¶

IIS_TIME_LIMIT `class-attribute` `instance-attribute` ¶

INFINITE_BOUND `class-attribute` `instance-attribute` ¶

INFINITE_COST `class-attribute` `instance-attribute` ¶