pyhgf.updates.vectorised.continuous.vectorised_continuous_posterior_update_unbounded#

pyhgf.updates.vectorised.continuous.vectorised_continuous_posterior_update_unbounded(layer, volatility_child, time_step=1.0, max_posterior_precision=10000000000.0, mean_field_updates=False)[source]#

Unbounded posterior update for a pure volatility-parent layer.

Vectorised port of pyhgf.updates.posterior.continuous.continuous_node_posterior_update_unbounded() (Lambert \(W_0\) dual-quadratic expansion with a variational energy-based blend). The approximation is derived for a single volatility child per node, so the coupling must be one-to-one: kappa square and diagonal, each parent node coupled to exactly one child node. The layer must have no value child (the nodalised update ignores value children; the vectorised builder rejects such a topology instead of silently dropping the edge).

Parameters:

  • layer (LayerState) – The parent layer’s state (being updated).

  • volatility_child (VolatilityChild) – The volatility child, with a square diagonal kappa.

  • time_step (float) – Current time step.

  • max_posterior_precision (float) – Upper bound applied to the posterior precision write.

  • mean_field_updates (bool)

Returns:

Updated layer state with posterior precision and mean.

Return type:

LayerState