Robust Problem

class lropt.RobustProblem(objective, constraints, eval_exp=None, train_flag=True, cons_data=None, verify_x_parameters: bool = True)

Create a Robust Optimization Problem with uncertain variables

evaluate() → float

Evaluates the out-of-sample value of the current solution and cvxpy/context parameters, with respect to the input data-set of uncertain parameters. The dataset is taken from u.eval_data_sol for each uncertain parameter u.

order_args(z_batch: list[Tensor], x_batch: list[Tensor], u_batch: list[Tensor] | Tensor)

This function orders z_batch (decisions), x_batch (context), and u_batch (uncertainty) according to the order in vars_params.

remove_uncertainty(override=False, solver=None)

This function canonizes a problem and saves it to self.problem_no_unc

Args:

override

If True, will override current problem_no_unc. If False and problem_no_unc exists, does nothing.

Returns:

None

solve(solver: str = None, warm_start: bool = True, verbose: bool = False, gp: bool = False, qcp: bool = False, requires_grad: bool = False, enforce_dpp: bool = False, ignore_dpp: bool = False, canon_backend: str | None = None, **kwargs)

This function solves the robust problem, and dualizes it first if it has not been dualized

Returns: the solution to the original problem

uncertain_parameters()

Find uncertain (u) parameters

unpack(solution) → None

Updates the problem state given a Solution.

Updates problem.status, problem.value and value of primal and dual variables. If solution.status is in cvxpy.settins.ERROR, this method is a no-op.

Arguments

solutioncvxpy.Solution

A Solution object.

Raises

ValueError

If the solution object has an invalid status

unpack_results_unc(solution, chain, inverse_data, solvername) → None

Updates the problem state given the solver results.

Updates problem.status, problem.value and value of primal and dual variables.

Arguments

solutionobject

The solution returned by applying the chain to the problem and invoking the solver on the resulting data.

chainSolvingChain

A solving chain that was used to solve the problem.

inverse_datalist

The inverse data returned by applying the chain to the problem.

Raises

cvxpy.error.SolverError

If the solver failed

verify_x_parameters()

This function verifies that x and u are in the correct diemsnions.

x_parameters()

Find context (x) parameters