agedi.models.conditionings.time
===============================

.. py:module:: agedi.models.conditionings.time


Classes
-------

.. autoapisummary::

   agedi.models.conditionings.time.TimeConditioning


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

.. py:class:: TimeConditioning(input_dim: int = 1, output_dim: int = 2, **kwargs)

   Bases: :py:obj:`agedi.models.conditionings.base.Conditioning`


   Condition the model on the time t.

   :param t: Time tensor of shape (Nodes, 1).
   :type t: torch.Tensor


   .. py:attribute:: omega


   .. py:method:: get_hparams() -> Dict

      Return hyperparameters for this time conditioning module.



   .. py:method:: get_conditioning(t: torch.Tensor) -> torch.Tensor

      Get the conditioning tensor for the time t.

      ::math::
          egin{align*}
          \mathbf{c} = egin{bmatrix} \sin(\omega t) \ \cos(\omega t) \end{bmatrix}
          \end{align*}

      :param t: Time tensor of shape (Nodes, 1).
      :type t: torch.Tensor

      :returns: Conditioning tensor of shape (Nodes, 2).
      :rtype: torch.Tensor



   .. py:method:: get_empty_conditioning(n: int) -> torch.Tensor

      Get an empty conditioning tensor.

      :returns: Empty conditioning tensor of shape (1, 2).
      :rtype: torch.Tensor



   .. py:method:: forward(batch: AtomsGraph, empty: bool = False) -> AtomsGraph

      Forward method to get the conditioning from the input

      This ignores training and empty flags.

      :param batch: The input batch
      :type batch: AtomsGraph
      :param empty: If True, return an empty conditioning tensor
      :type empty: bool

      :returns: The batch with the conditioning added to the representation
      :rtype: AtomsGraph