pyepo.func.jax.abcmodule¶

Abstract optimization module

Attributes¶

logger

Classes¶

optModule

An abstract module for differentiable optimization losses in end-to-end

Module Contents¶

pyepo.func.jax.abcmodule.logger¶

class pyepo.func.jax.abcmodule.optModule(optmodel: pyepo.model.opt.optModel, processes: int = 1, solve_ratio: float = 1.0, reduction: pyepo.func.runtime.Reduction = 'mean', dataset: pyepo.data.dataset.optDataset | None = None, require_solpool: bool = False, seed: int | None = None)¶

Bases: abc.ABC

An abstract module for differentiable optimization losses in end-to-end predict-then-optimize. It provides shared init validation, loss reduction, and the solution pool for all loss modules.

optmodel: pyepo.model.opt.optModel¶

processes: int¶

pool: pathos.multiprocessing.ProcessingPool | None¶

solve_ratio: float¶

reduction: pyepo.func.runtime.Reduction¶

solpool = None¶

abstractmethod forward(*args)¶: Forward pass