modopt.CUTEstProblem

class modopt.CUTEstProblem(**kwargs)

Class that wraps pyCUTEst problems for modOpt.

General functionality pyCUTEst provides:

• find_problems(): Find problems that satisfy given criteria as per CUTEst classification scheme.

• problem_properties(): Get properties of a given problem.

• print_available_sif_params(): Print available optional input parameters for a given problem.

• import_problem(): Import a given problem with optional parameters. Returns a pycutest.CUTEstProblem object.

Examples

Import a CUTEst problem named ‘ROSENBR’ and solve it using modOpt

>>> import pycutest
>>> cutest_problem = pycutest.import_problem('ROSENBR')
>>> import modopt as mo
>>> problem = mo.CUTEstProblem(cutest_problem=cutest_problem)
>>> optimizer = mo.SLSQP(problem, solver_options={'maxiter':100})
>>> optimizer.solve()

Methods

_compute_all(x, force_rerun=False, check_failure=False)

Compute and return the objective, constraints, objective gradient, and constraint Jacobian for the given design variable vector.

Parameters

x: np.ndarray

Vector of design variable values.

Returns

self.ffloat

Objective value.

self.cnp.ndarray

Constraint vector.

self.gnp.ndarray

Objective gradient vector.

self.jnp.ndarray

Constraint Jacobian vector.

_compute_constraint_jacobian(x, force_rerun=False, check_failure=False)

Compute and return the constraint Jacobian ‘j’ for the given design variable vector ‘x’.

_compute_constraint_jvp(x, v)

Compute and return the Jacobian-vector product ‘jvp’ for the given design variable vector ‘x’ and vector ‘v’.

_compute_constraint_vjp(x, v)

Compute and return the vector-Jacobian product ‘vjp’ for the given design variable vector ‘x’ and vector ‘v’.

_compute_constraints(x, force_rerun=False, check_failure=False)

Compute and return the constraints ‘c’ for the given design variable vector ‘x’.

_compute_lagrangian(x, mu)

Compute and return the Lagrangian ‘l’ for the given design variable vector ‘x’ and Lagrange multipliers ‘mu’.

_compute_lagrangian_gradient(x, mu)

Compute and return the Lagrangian gradient ‘lg’ for the given design variable vector ‘x’ and Lagrange multipliers ‘mu’.

_compute_lagrangian_hessian(x, mu)

Compute and return the Lagrangian Hessian ‘lh’ for the given design variable vector ‘x’ and Lagrange multipliers ‘mu’.

_compute_lagrangian_hvp(x, mu, v)

Compute and return the Lagrangian Hessian-vector product ‘lhvp’ for the given design variable vector ‘x’, Lagrange multipliers ‘mu’, and vector ‘v’. Only works for CONSTRAINED problems. (v=mu must be specified for constrained problems)

_compute_objective(x, force_rerun=False, check_failure=False)

Compute and return the objective ‘f’ for the given design variable vector ‘x’.

_compute_objective_gradient(x, force_rerun=False, check_failure=False)

Compute and return the objective gradient ‘g’ for the given design variable vector ‘x’.

_compute_objective_hessian(x)

Compute and return the objective Hessian ‘h’ for the given design variable vector ‘x’.

_compute_objective_hvp(x, v)

Compute and return the objective Hessian-vector product ‘hvp’ for the given design variable vector ‘x’ and vector ‘v’. Only works for UNCONSTRAINED problems. (v must be None for unconstrained problems)