pyepo.model.mpax

Optimization Model based on MPAX

Submodules

• pyepo.model.mpax.knapsack
• pyepo.model.mpax.mpaxmodel
• pyepo.model.mpax.shortestpath

Classes

optMpaxModel	This is an abstract class for an MPAX-based optimization model
shortestPathModel	This class is an optimization model for the shortest path problem
knapsackModel	This class is an optimization model for the relaxed knapsack problem

Package Contents

class pyepo.model.mpax.optMpaxModel(A=None, b=None, G=None, h=None, l=None, u=None, use_sparse_matrix=True, minimize=True)

Bases: pyepo.model.opt.optModel

This is an abstract class for an MPAX-based optimization model

A

The matrix of equality constraints.

Type:

jnp.ndarray, BCOO or BCSR

b

The right hand side of equality constraints.

Type:

jnp.ndarray

G

The matrix for inequality constraints.

Type:

jnp.ndarray, BCOO or BCSR

h

The right hand side of inequality constraints.

Type:

jnp.ndarray

l

The lower bound of the variables.

Type:

jnp.ndarray

u

The upper bound of the variables.

Type:

jnp.ndarray

use_sparse_matrix

Whether to use sparse matrix format, by default True.

Type:

bool

minimize

Whether to minimize objective, by default True.

Type:

bool

l

u

use_sparse_matrix = True

modelSense = 1

device = None

_has_jax_gpu

__repr__()

_rebuild_jit()

Rebuild JIT-compiled solve functions with current constraints.

property num_cost

number of costs to be predicted

setObj(c)

A method to set the objective function

Parameters:

c (np.ndarray / list) – cost of objective function

solve()

A method to solve the model

Returns:

optimal solution (list) and objective value (jnp.float32)

Return type:

tuple

static _jitted_solve(c, A, b, G, h, l, u, use_sparse_matrix)

A static method for JIT compile

copy()

A method to copy the model

Returns:

new copied model

Return type:

optModel

addConstr(coefs, rhs)

A method to add a new constraint

Parameters:

• coefs (np.ndarray / list) – coefficients of new constraint

• rhs (jnp.float32) – right-hand side of new constraint

Returns:

new model with the added constraint

Return type:

optModel

_getModel()

Placeholder method for MPAX. MPAX does not require an explicit model creation.

class pyepo.model.mpax.shortestPathModel(grid)

Bases: pyepo.model.mpax.mpaxmodel.optMpaxModel

This class is an optimization model for the shortest path problem

grid

Size of grid network

Type:

tuple of int

arcs

List of arcs

Type:

list

grid

arcs = []

_constructMatrix()

Constructs the incidence matrix A, supply/demand vector b, and upper bound u for the shortest path problem.

Returns:

Incidence matrix for flow conservation b (jnp.ndarray): Supply/demand vector u (jnp.ndarray): Upper bound for flow variables

Return type:

A (jnp.ndarray)

class pyepo.model.mpax.knapsackModel(weights, capacity)

Bases: pyepo.model.mpax.mpaxmodel.optMpaxModel

This class is an optimization model for the relaxed knapsack problem

_model

MPAX model

weights

Weights of items

Type:

np.ndarray / list

capacity

Total capacity

Type:

np.ndarray / list

items

List of item index

Type:

list

weights

capacity

items

_constructMatrix()

Constructs the inequality constraint matrix G, right-hand side h, and upper bound u for the knapsack problem.

Returns:

Weights of items h (jnp.ndarray): Total capacity u (jnp.ndarray): Upper bound for item selection

Return type:

G (jnp.ndarray)