Crate grb[][src]

Expand description

This crate provides Rust bindings for Gurobi Optimizer. It currently requires Gurobi 9.0 or higher.


  • Before using this crate, you should install Gurobi and obtain a license.

  • Make sure that the environment variable GUROBI_HOME is set to the installation path of Gurobi (like C:\gurobi911\win64 or /opt/gurobi911/linux64). If you are using the Conda package from the Gurobi channel, the build script will fall back to GUROBI_HOME=${CONDA_PREFIX}, so you should not set GUROBI_HOME.

Quick Start

The example below sets up and solves a Mixed Integer Program (MIP). Additional examples covering the more specific aspects of this crate’s API can be found here.

The documention for Model contains most of the details for defining, solving and querying models.

use grb::prelude::*;

let mut model = Model::new("model1")?;

// add decision variables with no bounds
let x1 = add_ctsvar!(model, name: "x1", bounds: ..)?;
let x2 = add_intvar!(model, name: "x2", bounds: ..)?;

// add linear constraints
let c0 = model.add_constr("c0", c!(x1 + 2*x2 >= -14))?;
let c1 = model.add_constr("c1", c!(-4 * x1 - x2 <= -33))?;
let c2 = model.add_constr("c2", c!(2* x1 <= 20 - x2))?;

// model is lazily updated by default
assert_eq!(model.get_obj_attr(attr::VarName, &x1).unwrap_err(), grb::Error::ModelObjectPending);
assert_eq!(model.get_attr(attr::IsMIP)?, 0);

// set the objective function, which updates the model objects (variables and constraints).
// One could also call `model.update()`
model.set_objective(8*x1 + x2, Minimize)?;
assert_eq!(model.get_obj_attr(attr::VarName, &x1)?, "x1");
assert_eq!(model.get_attr(attr::IsMIP)?, 1);

// write model to the file.

// optimize the model
assert_eq!(model.status()?, Status::Optimal);

// Querying a model attribute
assert_eq!(model.get_attr(attr::ObjVal)? , 59.0);

// Querying a model object attributes
assert_eq!(model.get_obj_attr(attr::Slack, &c0)?, -34.5);
let x1_name = model.get_obj_attr(attr::VarName, &x1)?;

// Querying an attribute for multiple model objects
let val = model.get_obj_attr_batch(attr::X, vec![x1, x2])?;
assert_eq!(val, [6.5, 7.0]);

// Querying variables by name
assert_eq!(model.get_var_by_name(&x1_name)?, Some(x1));


Due to the nature of C APIs, almost every Gurobi routine can return an error. Unless otherwise stated, if a method or function returns a Result, the error will be Error::FromAPI.


pub use attribute::attr;
pub use expr::Expr;
pub use parameter::param;


Gurobi Attributes for models, constraints and variables.

Interface to Gurobi’s callback API

This module contains the structs passed to the Model::add_constr(s) and Model::add_range(s) methods.

Algebraic expressions involving variables used to construct constraints and a helper trait for pretty-printing.

Gurobi parameters for Env and Model objects. See the manual for a list of parameters and their uses.

Most commonly used items from this crate bundled for convenient import.


Equivalent to calling add_var!(model, Binary, ...)

Equivalent to calling add_var!(model, Continuous, ...)

Equivalent to calling add_var!(model, Integer, ...)

Convienence wrapper around Model::add_var; adds a new variable to a Model object. The macro keyword arguments are optional.

A proc-macro for creating constraint objects.


A handle to an AsyncModel which is currently solving.

A wrapper around Model that supports async optimisation in the background.

A linear constraint added to a Model

Gurobi environment object (see the Gurobi manual) A Gurobi environment which hasn’t been started yet. Some Gurobi parameters, such as Record need to be set before the environment has been started.

A Gurobi Environment object.

Gurobi Model object.

A quadratic constraint added to a Model

An SOS constraint added to a Model

A Gurobi variable.


Sense for new linear/quadratic constraint

The error type for operations in Gurobi Rust API

Sense of objective function, aka direction of optimisation.

Type of cost function at feasibility relaxation

Status of a model

Gurobi variable types (see manual)


A large constant used by Gurobi to represent numeric infinity.


This trait encompasses all Gurobi model objects: Var, Constr, QConstr and SOS. Each ModelObject is associated with a particular model, and can only be used with that model. Each ModelObject also has a unique, fixed 32-bit ID. Gurobi itself uses an i32 to index objects (only positive indices are used), so the 32-bit limitation is already there. Note that IDs are only guaranteed to be unique if the concrete types of the ModelObject are the same and the objects belong to the same model. For example, if v is a Var and c is a Constr, then v and c may have the same ID. Additionally, if s is also a Var, but doesn’t belong to the same Model as v, s and v may have the same ID.


Returns the version number of Gurobi

Type Definitions

A specialized std::result::Result for library errors