quimb.experimental.belief_propagation.bp_common

Module Contents

Classes

RollingDiffMean

Tracker for the absolute rolling mean of diffs between values, to

BeliefPropagationCommon

Common interfaces for belief propagation algorithms.

Functions

prod(xs)

Product of all elements in xs.

initialize_hyper_messages(tn[, fill_fn, smudge_factor])

Initialize messages for belief propagation, this is equivalent to doing

combine_local_contractions(tvals, mvals, backend[, ...])

contract_hyper_messages(tn, messages[, ...])

Estimate the contraction of tn given messages, via the

compute_index_marginal(tn, ind, messages)

Compute the marginal for a single index given messages.

compute_tensor_marginal(tn, tid, messages)

Compute the marginal for the region surrounding a single tensor/factor

compute_all_index_marginals_from_messages(tn, messages)

Compute all index marginals from belief propagation messages.

maybe_get_thread_pool(thread_pool)

Get a thread pool if requested.

create_lazy_community_edge_map(tn[, site_tags, ...])

For lazy BP algorithms, create the data structures describing the
quimb.experimental.belief_propagation.bp_common.prod(xs)

Product of all elements in xs.

class quimb.experimental.belief_propagation.bp_common.RollingDiffMean(size=16)

Tracker for the absolute rolling mean of diffs between values, to assess effective convergence of BP above actual message tolerance.

update(x)
absmeandiff()
class quimb.experimental.belief_propagation.bp_common.BeliefPropagationCommon

Common interfaces for belief propagation algorithms.

Parameters:

  • max_iterations (int, optional) – The maximum number of iterations to perform.

  • tol (float, optional) – The convergence tolerance for messages.

  • progbar (bool, optional) – Whether to show a progress bar.

run(max_iterations=1000, tol=5e-06, info=None, progbar=False)
quimb.experimental.belief_propagation.bp_common.initialize_hyper_messages(tn, fill_fn=None, smudge_factor=1e-12)

Initialize messages for belief propagation, this is equivalent to doing a single round of belief propagation with uniform messages.

Parameters:

  • tn (TensorNetwork) – The tensor network to initialize messages for.

  • fill_fn (callable, optional) – A function to fill the messages with, of signature fill_fn(shape).

  • smudge_factor (float, optional) – A small number to add to the messages to avoid numerical issues.

Returns:

messages – The initial messages. For every index and tensor id pair, there will be a message to and from with keys (ix, tid) and (tid, ix).

Return type:

dict

quimb.experimental.belief_propagation.bp_common.combine_local_contractions(tvals, mvals, backend, strip_exponent=False, check_for_zero=True)
quimb.experimental.belief_propagation.bp_common.contract_hyper_messages(tn, messages, strip_exponent=False, backend=None)

Estimate the contraction of tn given messages, via the exponential of the Bethe free entropy.

quimb.experimental.belief_propagation.bp_common.compute_index_marginal(tn, ind, messages)

Compute the marginal for a single index given messages.

Parameters:

  • tn (TensorNetwork) – The tensor network to compute the marginal for.

  • ind (int) – The index to compute the marginal for.

  • messages (dict) – The messages to use, which should match tn.

Returns:

marginal – The marginal probability distribution for the index ind.

Return type:

array_like

quimb.experimental.belief_propagation.bp_common.compute_tensor_marginal(tn, tid, messages)

Compute the marginal for the region surrounding a single tensor/factor given messages.

Parameters:

  • tn (TensorNetwork) – The tensor network to compute the marginal for.

  • tid (int) – The tensor id to compute the marginal for.

  • messages (dict) – The messages to use, which should match tn.

Returns:

marginal – The marginal probability distribution for the tensor/factor tid.

Return type:

array_like

quimb.experimental.belief_propagation.bp_common.compute_all_index_marginals_from_messages(tn, messages)

Compute all index marginals from belief propagation messages.

Parameters:

  • tn (TensorNetwork) – The tensor network to compute marginals for.

  • messages (dict) – The belief propagation messages.

Returns:

marginals – The marginals for each index.

Return type:

dict

quimb.experimental.belief_propagation.bp_common.maybe_get_thread_pool(thread_pool)

Get a thread pool if requested.

quimb.experimental.belief_propagation.bp_common.create_lazy_community_edge_map(tn, site_tags=None, rank_simplify=True)

For lazy BP algorithms, create the data structures describing the effective graph of the lazily grouped ‘sites’ given by site_tags.