pyepo.metric.unambregret

Unambiguous regret loss

Attributes

logger

Functions

unambRegret(→ float)	Normalized unambiguous regret (worst-case-tie SPO loss).
calUnambRegret(→ float)	A function to calculate normalized unambiguous regret for a batch

Module Contents

pyepo.metric.unambregret.logger

pyepo.metric.unambregret.unambRegret(predmodel: torch.nn.Module, optmodel: pyepo.model.opt.optModel, dataloader: torch.utils.data.DataLoader, tolerance: float = 1e-05) → float

Normalized unambiguous regret (worst-case-tie SPO loss).

When the predicted cost vector \(\hat{\mathbf{c}}\) yields multiple optimal solutions, regret reports the realized one while unambRegret reports the worst case over all optima: \(l_{\mathrm{URegret}}(\hat{\mathbf{c}}, \mathbf{c}) = \max_{\mathbf{w} \in W^*(\hat{\mathbf{c}})} \mathbf{c}^\top \mathbf{w} - z^*(\mathbf{c})\). More theoretically rigorous than regret, but in practice the two are nearly identical and unambRegret is rarely required.

Parameters:

• predmodel – a regression neural network for cost prediction

• optmodel – a PyEPO optimization model

• dataloader – PyTorch DataLoader over an optDataset

• tolerance – precision used when rounding predicted costs to find ties

Returns:

normalized unambiguous regret

Return type:

float

pyepo.metric.unambregret.calUnambRegret(optmodel: pyepo.model.opt.optModel, pred_cost: numpy.ndarray, true_cost: numpy.ndarray, true_obj: float, tolerance: float = 1e-05, max_iter: int = 10) → float

A function to calculate normalized unambiguous regret for a batch

Parameters:

• optmodel – optimization model

• pred_cost – predicted costs

• true_cost – true costs

• true_obj – true optimal objective value

• tolerance – tolerance for precision

• max_iter – maximum number of recursive retries

Returns:

unambiguous regret loss

Return type:

float