pyepo.model.copt.vrp

Capacitated vehicle routing problem

Attributes

CallbackBase

Classes

vrpABModel	Abstract COPT-backed model for the capacitated vehicle routing problem.
vrpRCIModel	CVRP formulation with 2-degree constraints and lazy rounded-capacity cuts.
vrpMTZModel	CVRP formulation on a directed graph with MTZ-style capacity constraints
vrpMTZModelRel	LP relaxation of vrpMTZModel.

Module Contents

pyepo.model.copt.vrp.CallbackBase

class pyepo.model.copt.vrp.vrpABModel(num_nodes: int, demands: list[float] | numpy.ndarray, capacity: float, num_vehicle: int, *args, **kwargs)

Bases: pyepo.model.bases.vrpABBase, pyepo.model.copt.coptmodel.optCoptModel

Abstract COPT-backed model for the capacitated vehicle routing problem.

A single-customer route is excluded so all edge variables stay strictly binary; if a single-stop route is actually needed, duplicate the depot.

class pyepo.model.copt.vrp.vrpRCIModel(num_nodes: int, demands: list[float] | numpy.ndarray, capacity: float, num_vehicle: int, *args, **kwargs)

Bases: vrpABModel

CVRP formulation with 2-degree constraints and lazy rounded-capacity cuts.

Uses one undirected Var per edge (x[j,i] aliases x[i,j]). Subtour elimination and rounded capacity inequalities are added lazily during branch-and-cut via a COPT callback.

setObj(c: numpy.ndarray | torch.Tensor | list) → None

A method to set the objective function

Parameters:

c – cost vector aligned with self.edges

solve() → tuple[numpy.ndarray, float]

A method to solve the model

Returns:

edge-selection vector (uint8) and objective value (float)

Return type:

tuple

class pyepo.model.copt.vrp.vrpMTZModel(num_nodes: int, demands: list[float] | numpy.ndarray, capacity: float, num_vehicle: int, *args, **kwargs)

Bases: vrpABModel

CVRP formulation on a directed graph with MTZ-style capacity constraints (no lazy cuts). Cost vector is per undirected edge: cost c[k] is assigned to both x[i,j] and x[j,i].

setObj(c: numpy.ndarray | torch.Tensor | list) → None

A method to set the objective function

Parameters:

c – cost vector aligned with self.edges (one cost per undirected edge)

solve() → tuple[numpy.ndarray, float]

A method to solve the model

Returns:

edge-selection vector (uint8) and objective value (float)

Return type:

tuple

relax() → vrpMTZModelRel

A method to get linear relaxation model

class pyepo.model.copt.vrp.vrpMTZModelRel(num_nodes: int, demands: list[float] | numpy.ndarray, capacity: float, num_vehicle: int, *args, **kwargs)

Bases: vrpMTZModel

LP relaxation of vrpMTZModel.

solve() → tuple[numpy.ndarray, float]

A method to solve the model — returns fractional edge selections

relax() → NoReturn

A forbidden method to relax MIP model

getTour(sol: numpy.ndarray | torch.Tensor | list) → list[list[int]]

A forbidden method to get a tour from solution