wmpy.optimization.scipyoptimizer.ScipyOptimizer¶

class ScipyOptimizer(epsilon=0.01)[source]¶

Bases: object

Default constructor.

Parameters:: epsilon (float) – small float constant used to enforce strict inequality constraints

Methods

`compute_inner_box`	Returns the largest axis-aligned hyperrectangle fully enclosed in the polytope by solving the convex optimization problem on 2N variables described here:
`compute_outer_box`	Returns the smallest axis-aligned hyperrectangle fully enclosing the polytope by making 2N calls to an LP solver.
`optimize`	Solves the constrained optimization problem where:

compute_inner_box(polytope)[source]¶

Returns the largest axis-aligned hyperrectangle fully enclosed in the polytope by solving the convex optimization problem on 2N variables described here:

https://scicomp.stackexchange.com/a/26465

The result is stored for future uses.

Parameters:: polytope (Polytope) – the Polytope instance
Return type:: tuple[ndarray, ndarray]
Returns:: Two numpy arrays corresponding to the extremes of the box.

compute_outer_box(polytope)[source]¶

Returns the smallest axis-aligned hyperrectangle fully enclosing the polytope by making 2N calls to an LP solver.

The result is stored for future uses.

Parameters:: polytope (Polytope) – the Polytope instance
Return type:: tuple[ndarray, ndarray]
Returns:: Two numpy arrays corresponding to the extremes of the box.

optimize(polytope, polynomial, maximize=True)[source]¶

Solves the constrained optimization problem where:

polynomial is the objective function
polytope is the convex domain

Either polynomial is maximized (default) or minimized.

Parameters:

polytope (Polytope) – convex integration bounds
polynomial (Polynomial) – the objective function
maximize (bool) – Boolean flag (optional, def: True)

Return type:

ndarray

Returns:

A numpy array corresponding to the arg{max/min}.