pyepo.utils
===========

.. py:module:: pyepo.utils

.. autoapi-nested-parse::

   Package-wide utility functions



Functions
---------

.. autoapisummary::

   pyepo.utils.objective_offset
   pyepo.utils.require_linear_objective
   pyepo.utils.costToNumpy


Module Contents
---------------

.. py:function:: objective_offset(model: pyepo.model.opt.optModel) -> float

   Bare objective constant of a compiled DSL problem.

   :param model: optimization model

   :returns: objective constant, 0.0 for models without one
   :rtype: float


.. py:function:: require_linear_objective(optmodel: pyepo.model.opt.optModel) -> None

   Reject compiled DSL models carrying a quadratic objective term.

   :param optmodel: optimization model


.. py:function:: costToNumpy(c: numpy.ndarray | torch.Tensor | list, dtype: Any = np.float32) -> numpy.ndarray

   Normalize a cost vector to a numpy array, detaching torch tensors as needed.

   :param c: cost vector
   :param dtype: target dtype when ``c`` is not already a tensor; torch
                 tensors are converted via ``.detach().cpu().numpy()`` and keep their
                 existing dtype.

   :returns: numpy cost vector
   :rtype: np.ndarray