aeppl.dists.DiscreteMarkovChainFactory.__init__#

DiscreteMarkovChainFactory.__init__(inputs, outputs, inline=False, lop_overrides='default', grad_overrides='default', rop_overrides='default', connection_pattern=None, name=None, **kwargs)#
Parameters
  • inputs (List[Variable]) – The inputs to the graph.

  • outputs (List[Variable]) – The outputs to the graph.

  • inline (bool) –

    Defaults to False

    True : Cause the Op’s original graph being used during compilation, the Op will not be visible in the compiled graph but rather its internal graph.

    False : will use a pre-compiled function inside.

  • grad_overrides (str) –

    Defaults to 'default'. This argument is mutually exclusive with lop_overrides.

    'default' : Do not override, use default grad() result

    OpFromGraph: Override with another OpFromGraph, should accept inputs as the same order and types of inputs and output_grads arguments as one would specify in :meth:`Op.grad`() method.

    callable: Should take two args: inputs and output_grads. Each argument is expected to be a list of :class:`Variable `. Must return list of :class:`Variable `.

  • lop_overrides (str) –

    Defaults to 'default'.

    This argument is mutually exclusive with grad_overrides.

    These options are similar to the grad_overrides above, but for the Op.L_op method.

    'default': Do not override, use the default Op.L_op result

    OpFromGraph: Override with another OpFromGraph, should accept inputs as the same order and types of inputs, outputs and output_grads arguments as one would specify in Op.grad method.

    callable: Should take three args: inputs, outputs and output_grads. Each argument is expected to be a list of Variable. Must return list of Variable.

    NullType instance: Treat as non-differentiable DisconnectedType instance: Treat as disconnected gradient, numerically gives zero

    list: Each OpFromGraph/callable must return a single Variable. Each list element corresponds to gradient of a specific input, length of list must be equal to number of inputs.

  • rop_overrides (str) –

    One of {'default', OpFromGraph, callable, Variable}.

    Defaults to 'default'.

    'default': Do not override, use the default Op.R_op result

    OpFromGraph: Override with another OpFromGraph, should accept inputs as the same order and types of inputs and eval_points arguments as one would specify in Op.R_op method.

    callable: Should take two args: inputs and eval_points. Each argument is expected to be a list of Variable. Must return list of Variable.

    NullType instance: Treat as non-differentiable DisconnectedType instance: Treat as zero since DisconnectedType is not yet supported in Op.R_op.

    list: Each OpFromGraph/callable must return a single Variable. Each list element corresponds to a specific output of Op.R_op, length of list must be equal to number of outputs. connection_pattern If not None, this will be used as the connection_pattern for this Op.

  • name (Optional[str]) – A name for debugging purposes.

  • kwargs – Check aesara.function for more arguments, only works when not inline.

  • connection_pattern (Optional[List[List[bool]]]) –