Struct grb::Model [−][src]
pub struct Model { /* fields omitted */ }
Expand description
Gurobi Model object.
This will be where the bulk of interactions with Gurobi occur.
Implementations
Create a new model with the given environment. The original environment is
copied by Gurobi. To modify the environment of the model, use Model::get_env_mut
.
Examples
let mut env = Env::new("")?;
env.set(param::OutputFlag, 0)?;
let mut model = Model::with_env("Model", &env)?;
assert_eq!(model.get_param(param::OutputFlag)?, 0);
// Equivalent to model.set_param(param::OutputFlag, 1)?
model.get_env_mut().set(param::OutputFlag, 1)?;
assert_eq!(env.get(param::OutputFlag).unwrap(), 0); // original env is unchanged
Create a new model with the default environment, which is lazily initialised.
Create a copy of the model. This method is fallible due to the lazy update approach and the underlying
Gurobi C API, so a Clone
implementation is not provided.
Errors
Error::FromAPI
if a Gurobi error occursError::ModelUpdateNeeded
if model objects have been added to the model since the last update.
Read a model from a file. See the manual for accepted file formats.
Create the fixed model associated with the current MIP model.
The model must be MIP and have a solution loaded. In the fixed model, each integer variable is fixed to the value that it takes in the current MIP solution.
Get shared reference to the environment associated with the model.
Get mutable reference to the environment associated with the model.
Apply all queued modification of the model and update internal lookups.
Some operations like Model::try_clone
require this method to be called.
Examples
let mut m = Model::new("model")?;
let x = add_ctsvar!(m);
assert_eq!(m.try_clone().err().unwrap(), grb::Error::ModelUpdateNeeded);
m.update();
assert!(m.try_clone().is_ok());
Optimize the model synchronously. This method will always trigger a Model::update
.
Optimize the model with a callback. The callback is any type that implements the
Callback
trait. Closures, and anything else that implements FnMut(CbCtx) -> Result<()>
implement the Callback
trait automatically. This method will always trigger a Model::update
.
See crate::callback
for details on how to use callbacks.
Panics
This function panics if Gurobi errors on clearing the callback.
Compute an Irreducible Inconsistent Subsystem (IIS) of the model. The constraints in the IIS can be identified
by checking their IISConstr
attribute
Example
fn compute_iis_constraints(m: &mut Model) -> grb::Result<Vec<Constr>> {
m.compute_iis()?;
let constrs = m.get_constrs()?; // all constraints in model
let iis_constrs = m.get_obj_attr_batch(attr::IISConstr, constrs.iter().copied())?
.into_iter()
.zip(constrs)
// IISConstr is 1 if constraint is in the IIS, 0 otherwise
.filter_map(|(is_iis, c)| if is_iis > 0 { Some(*c)} else { None })
.collect();
Ok(iis_constrs)
}
Send a request to the model to terminate the current optimization process.
Reset the model to an unsolved state.
All solution information previously computed are discarded.
Perform an automated search for parameter settings that improve performance on the model. See also references on official manual.
Prepare to retrieve the results of tune()
.
See also references on official
manual.
Insert a message into log file.
Panics
Panics when message
cannot be converted to a nul-terminated C string.
Import a model from a file. See Model::write
for details on valid file types.
Export a model to a file.
The file type is encoded in the file name suffix. Valid suffixes are .mps
, .rew
, .lp
, or .rlp
for
writing the model itself, .ilp
for writing just the IIS associated with an infeasible model,
.sol
for writing the current solution, .mst
for writing
a start vector, .hnt
for writing a hint file, .bas
for writing an LP basis, .prm
for writing modified
parameter settings, .attr
for writing model attributes, or .json
for writing solution information in
JSON format. If your system has compression utilities installed (e.g., 7z or zip for Windows, and gzip,
bzip2, or unzip for Linux or Mac OS), then the files can be compressed, so additional suffixes of .gz
,
.bz2
, or .7z
are accepted.
Add a decision variable to the model. This method allows the user to give the entire column (constraint coefficients).
The add_var!
macro and its friends are usually easier to use.
Add multiple linear constraints to the model in a single Gurobi API call.
Accepts anything that can be turned into an iterator of (name, constraint)
pairs
where name : AsRef<str>
(eg &str
or String
) and constraint
is a linear IneqExpr
.
Examples
let mut m = Model::new("model")?;
let x = add_ctsvar!(m)?;
let y = add_ctsvar!(m)?;
let constraints = vec![
(&"c1", c!(x <= 1 - y )),
(&"c2", c!(x == 0.5*y )),
];
m.add_constrs(constraints)?;
// store owned names in Vec to ensure they live long enough
let more_constraints_names : Vec<_> = (0..10).map(|i| format!("r{}", i)).collect();
// A Map iterator of (&String, IneqConstr)
let more_constraints = (0..10).map(|i| (&more_constraints_names[i], c!(x >= i*y )));
m.add_constrs(more_constraints)?;
Errors
Error::AlgebraicError
if a nonlinear constraint is given.Error::ModelObjectPending
if some variables haven’t yet been added to the model.Error::ModelObjectRemoved
if some variables have been removed from the model.Error::ModelObjectMismatch
if some variables are from a different model.Error::FromAPI
if a Gurobi API error occurs.
Add a range constraint to the model.
This operation adds a decision variable with lower/upper bound, and a linear
equality constraint which states that the value of variable must equal to expr
.
As with Model::add_constr
, the c!
macro is usually used to construct
the second argument.
Errors
Error::AlgebraicError
if the expression in the range constraint is not linear.Error::ModelObjectPending
if some variables haven’t yet been added to the model.Error::ModelObjectRemoved
if some variables have been removed from the model.Error::ModelObjectMismatch
if some variables are from a different model.Error::FromAPI
if a Gurobi API error occurs.
Examples
let mut m = Model::new("model")?;
let x = add_ctsvar!(m)?;
let y = add_ctsvar!(m)?;
m.add_range("", c!(x - y in 0..1))?;
let r = m.add_range("", c!(x*y in 0..1));
assert!(matches!(r, Err(grb::Error::AlgebraicError(_))));
Add multiple range constraints to the model in a single API call, analagous to
Model::add_constrs
.
Errors
Error::AlgebraicError
if the expression a the range constraint is not linear.Error::ModelObjectPending
if some variables haven’t yet been added to the model.Error::ModelObjectRemoved
if some variables have been removed from the model.Error::ModelObjectMismatch
if some variables are from a different model.Error::FromAPI
if a Gurobi API error occurs.
Add a quadratic constraint to the model. See the manual for which quadratic expressions are accepted by Gurobi.
Errors
Error::ModelObjectPending
if some variables haven’t yet been added to the model.Error::ModelObjectRemoved
if some variables have been removed from the model.Error::ModelObjectMismatch
if some variables are from a different model.Error::FromAPI
if a Gurobi API error occurs.
Add a single Special Order Set (SOS) constraint to the model.
Errors
Error::ModelObjectPending
if some variables haven’t yet been added to the model.Error::ModelObjectRemoved
if some variables have been removed from the model.Error::ModelObjectMismatch
if some variables are from a different model.Error::FromAPI
if a Gurobi API error occurs.
Set the objective function of the model and optimisation direction (min or max).
Because this requires setting a Var
attribute (the Obj
attribute), this method
always triggers a model update.
Errors
Error::ModelObjectPending
if some variables haven’t yet been added to the model.Error::ModelObjectRemoved
if some variables have been removed from the model.Error::ModelObjectMismatch
if some variables are from a different model.Error::FromAPI
if a Gurobi API error occurs.
Get a constraint by name. Returns either a constraint if one was found, or None
if none were found.
If multiple constraints match, the method returns an arbitary one.
Usage
let mut m = Model::new("model")?;
let x = add_binvar!(m)?;
let y = add_binvar!(m)?;
let c = m.add_constr("constraint", c!(x + y == 1))?;
assert_eq!(m.get_constr_by_name("constraint").unwrap_err(), grb::Error::ModelUpdateNeeded);
m.update()?;
assert_eq!(m.get_constr_by_name("constraint")?, Some(c));
assert_eq!(m.get_constr_by_name("foo")?, None);
Errors
Error::NulError
if thename
cannot be converted to a C-stringError::ModelUpdateNeeded
if a model update is needed.Error::ModelObjectRemoved
if the constraint has been removed from the model.Error::ModelObjectMismatch
if the constraint is from a different model.Error::FromAPI
if a Gurobi API error occurs.
Get a variable object by name. See Model::get_constr_by_name
for details
Errors
Error::NulError
if thename
cannot be converted to a C-stringError::ModelUpdateNeeded
if a model update is needed.Error::ModelObjectRemoved
if the variable has been removed from the model.Error::ModelObjectMismatch
if the variable is from a different model.Error::FromAPI
if a Gurobi API error occurs.
Query a Model attribute. Model attributes (objects with the ModelAttr
trait) can be found in the attr
module.
pub fn get_obj_attr<A, O, V>(&self, attr: A, obj: &O) -> Result<V> where
A: ObjAttrGet<O, V>,
O: ModelObject,
pub fn get_obj_attr<A, O, V>(&self, attr: A, obj: &O) -> Result<V> where
A: ObjAttrGet<O, V>,
O: ModelObject,
pub fn get_obj_attr_batch<A, I, O, V>(&self, attr: A, objs: I) -> Result<Vec<V>> where
A: ObjAttrGet<O, V>,
I: IntoIterator<Item = O>,
O: ModelObject,
pub fn get_obj_attr_batch<A, I, O, V>(&self, attr: A, objs: I) -> Result<Vec<V>> where
A: ObjAttrGet<O, V>,
I: IntoIterator<Item = O>,
O: ModelObject,
pub fn set_obj_attr<A, O, V>(&self, attr: A, obj: &O, val: V) -> Result<()> where
A: ObjAttrSet<O, V>,
O: ModelObject,
pub fn set_obj_attr<A, O, V>(&self, attr: A, obj: &O, val: V) -> Result<()> where
A: ObjAttrSet<O, V>,
O: ModelObject,
Set an attribute of a Model object (Const, Var, etc). Attributes (objects with the Attr
trait) can be found
in the attr
module.
Example
let mut model = Model::new("")?;
let x = add_ctsvar!(model)?;
let c = model.add_constr("", c!(x <= 1))?;
model.set_obj_attr(attr::VarName, &x, "x")?;
model.set_obj_attr(attr::ConstrName, &c, "c")?;
Trying to set an attribute on a model object that belongs to another model object type will fail to compile:
model.set_obj_attr2(attr::ConstrName, &x, "c")?;
pub fn set_obj_attr_batch<A, O, I, V>(
&self,
attr: A,
obj_val_pairs: I
) -> Result<()> where
A: ObjAttrSet<O, V>,
I: IntoIterator<Item = (O, V)>,
O: ModelObject,
pub fn set_obj_attr_batch<A, O, I, V>(
&self,
attr: A,
obj_val_pairs: I
) -> Result<()> where
A: ObjAttrSet<O, V>,
I: IntoIterator<Item = (O, V)>,
O: ModelObject,
Set an attribute of multiple Model objects (Const, Var, etc). Attributes (objects with the Attr
trait) can be
found in the attr
module.
Modify the model to create a feasibility relaxation.
Given a Model
whose objective function is $f(x)$, the feasibility relaxation seeks to minimise
$$
\text{min}\quad f(x) + \sum_{j} w_j \cdot p(s_j)
$$
where $s_j > 0$ is the slack variable of $j$ -th constraint or bound, $w_j$ is the $j$-th weight
and $p(s)$ is the penalty function.
The ty
argument sets the penalty function:
RelaxType variant | Penalty function |
---|---|
Quadratic | $ p(s) = {s}^2 $ |
Linear | $ p(s) = {s} $ |
Cardinality | $ p(s) = \begin{cases} 1 & \text{if } s > 0 \\ 0 & \text{otherwise} \end{cases} $ |
This method will modify the model - if this is not desired copy the model before invoking
this method with Model::try_clone()
.
Arguments
-
ty
: The type of cost function used when finding the minimum cost relaxation. -
minrelax
: How the objective should be minimised.If
false
, optimizing the returned model gives a solution that minimizes the cost of the violation. Iftrue
, optimizing the returned model finds a solution that minimizes the original objective, but only from among those solutions that minimize the cost of the violation. Note that this method must solve an optimization problem to find the minimum possible relaxation when set totrue
, which can be quite expensive. -
lb_pen
: Variables whose lower bounds are allowed to be violated, and their penalty weights. -
ub_pen
: Variables whose upper bounds are allowed to be violated, and their penalty weights. -
constr_pen
: Constraints which are allowed to be violated, and their penalty weights.
Returns
- The objective value for the relaxation performed (if
minrelax
istrue
). - Slack variables for relaxation and related linear/quadratic constraints.
pub fn set_pwl_obj(
&mut self,
var: &Var,
points: impl IntoIterator<Item = (f64, f64)>
) -> Result<()>
pub fn set_pwl_obj(
&mut self,
var: &Var,
points: impl IntoIterator<Item = (f64, f64)>
) -> Result<()>
Set a piecewise-linear objective function for the variable.
Given a sequence of points $(x_1, y_1), \dots, (x_n, y_n)$, the piecewise-linear objective function $f(x)$ is defined as follows: $$ f(x) = \begin{cases} y_1 + \dfrac{y_2 - y_1}{x_2 - x_1} \, (x - x_1) & \text{if $x \leq x_1$}, \\ \\ y_i + \dfrac{y_{i+1} - y_i}{x_{i+1}-x_i} \, (x - x_i) & \text{if $x_i \leq x \leq x_{i+1}$}, \\ \\ y_n + \dfrac{y_n - y_{n-1}}{x_n-x_{n-1}} \, (x - x_n) & \text{if $x \geq x_n$}, \end{cases} $$
The Obj
attribute of the Var
object will be set to 0. To delete the piecewise-linear function on the
variable, set the value of Obj
attribute to non-zero.
The points
argument contains the pairs $(x_i,y_i)$ and must satisfy $x_i < x_{i+1}$.
Retrieve the quadratic constraints in the model.
Errors
Returns an error if a model update is needed
Remove a variable or constraint from the model.
Retrieve a single constant matrix coefficient of the model.
Change a single constant matrix coefficient of the model.
pub fn set_coeffs(
&mut self,
coeffs: impl IntoIterator<Item = (Var, Constr, f64)>
) -> Result<()>
pub fn set_coeffs(
&mut self,
coeffs: impl IntoIterator<Item = (Var, Constr, f64)>
) -> Result<()>
Change a set of constant matrix coefficients of the model.
Trait Implementations
Performs the conversion.