Skip to content

LatentEncoder

moppy.interfaces.LatentEncoder

The LatentEncoder class serves as an abstract base class for all latent encoder implementations. It defines the interface for encoding a trajectory into a latent variable and sampling from it.

Functions

encode_to_latent_variable()

Encodes a trajectory into a latent variable z, which is a vector of a fixed dimension.

Method Signature:

encode_to_latent_variable(self, trajectory: Trajectory)

Parameters:

  • trajectory (Trajectory): The trajectory to encode.

Returns:

  • Tuple: A tuple containing the sampled latent variable z represented as mu (mean) and sigma (standard deviation).

Note: This method is abstract and must be implemented by subclasses.

@abstractmethod
def encode_to_latent_variable(self, trajectory: Trajectory):
    pass

sample_latent_variable()

Samples a latent variable z from a normal distribution specified by mu and sigma.

Method Signature:

sample_latent_variable(self, mu, sigma, percentage_of_standard_deviation=None)

Parameters:

  • mu: The mean of the normal distribution.
  • sigma: The standard deviation of the normal distribution.
  • percentage_of_standard_deviation (optional): The percentage of the standard deviation to sample from.

Returns:

  • Tensor: The sampled latent variable z.

Note: This method is abstract and must be implemented by subclasses.

@abstractmethod
def sample_latent_variable(self, mu, sigma, percentage_of_standard_deviation=None):
    pass