File Formats#

Instead of building a model using one of the APIs, all model information can be read from a text file. It is also possible to write model information into a text file. Moreover, other information, such as solution values, attributes, or parameters, can be read and written via text files as well. The Gurobi Optimizer works with a variety of text file formats which are discussed in this section.

Note that all of the Gurobi file I/O routines can work with compressed versions of these files. Specifically, the Gurobi Optimizer can read or write files with the following extensions: .zip, .gz, .bz2, .7z or .xz (assuming that the associated compression tool, e.g., 7zip for .7z, is installed on your machine and a corresponding entry is part of your PATH environment variable).

Model file formats

These text file formats can be used to read or write model information:

File format

Description

MPS

This is the most widely used format for storing math programming models. It is recommended to use this format (or REW) for optimization.

REW

This format is identical to the MPS format but with anonymized variable and constraint names.

DUA

This format is identical to the MPS format but it holds the dual formulation of a pure LP model.

LP

This format is a human readable version of MPS file format. It, however, does not preserve column order when read, and typically does not preserve the exact numerical values of the coefficients.

RLP

This format is identical to the LP format but with anonymized variable and constraint names.

DLP

This format is identical to the LP format but it holds the dual formulation of a pure LP model.

ILP

This format is identical to the LP format but it is used for Irreducible Inconsistent Subsystem (IIS) models.

OPB

This format is used to store pseudo-boolean satisfaction and pseudo-boolean optimization models.

Solution file formats

These text file formats can be used to read or write solution information:

File format

Description

SOL

This format holds solution information, i.e., solution values for all variables.

JSON solution

This format holds solution information, i.e., solution values for all variables and additional information about the results of the optimization.

MST

This format holds the solution information necessary for a full MIP start.

BAS

This format holds simplex basis information. Importing this data into a continuous models allows the simplex algorithm to start from the given simplex basis.

Solution information can only be written after the model has been optimized and a solution is available. On the other hand, all solution file formats can be used to import solution information to start optimization with a known feasible solution. However, preferably an MST file is used when providing MIP start information for a MIP model. It is also possible to hand over partial solution information, i.e., MST or SOL files that contain only a subset of variables with values.

Other File Formats

File format

Description

HNT

This format is used to hold MIP hints. Importing this data into a MIP model guides the MIP search towards a guess at a high-quality feasible solution.

ORD

This format is used to hold MIP variable branching priorities. Importing this data into a MIP model affects the search strategy.

ATTR

This format stores a collection of attributes of a model, including (multiple) MIP starts, solution, basis information, partitions, variable hints and branching priorities.

PRM

This format holds parameter values. Importing this data into a model changes the values of the referenced parameters.