phylox.generators.mcmc.base

Functions

acceptance_probability(network, ...[, ...])

Computes the acceptance probability of a move.
phylox.generators.mcmc.base.acceptance_probability(network, result_network, move, move_type_probabilities, number_of_leaves=None, current_reticulation_number=None, symmetries=False)

Computes the acceptance probability of a move.

Parameters:

  • network – the network before the move.

  • result_network – the network after the move.

  • move – the move.

  • move_type_probabilities – the move type probabilities.

  • number_of_leaves – the number of leaves in the network.

  • current_reticulation_number – the current number of reticulations in the network.

  • symmetries – whether to correct for symmetries.

Returns:

the acceptance probability of the move.

phylox.generators.mcmc.base.sample_mcmc_networks(starting_network, move_type_probabilities, restriction_map=None, correct_symmetries=True, burn_in=1000, number_of_samples=1, add_root_if_necessary=False, seed=None)

Samples phylogenetic networks using a Markov-Chain Monte Carlo method.

Parameters:

  • starting_network – the phylox.DiNetwork used as the starting point of the Markov chain.

  • move_type_probabilities – a dictionary mapping MoveTypes to probabilities.

  • restriction_map – a boolean function that takes a phylox.DiNetwork as input.

  • correct_symmetries – whether to correct for symmetries in the acceptance probability, set to True for uniform distribution.

  • burn_in – the number of steps (including rejected proposals) between each sample.

  • number_of_samples – the number of networks to sample.

  • add_root_if_necessary – whether to add a root edge to each root if it has out-degree > 1.

  • seed – the seed for the random number generator.

Returns:

a list of phylox.DiNetwork objects.

Example:

>>> from phylox import DiNetwork
>>> from phylox.rearrangement.movetype import MoveType
>>> from phylox.generators.mcmc import sample_mcmc_networks
>>> starting_network = DiNetwork(
...     edges = ((0,1), (0,2), (1,2), (1,3), (2,4)),
...     labels = ((3, "A"), (4, "B")),
... )
>>> move_type_probabilities = {
...     MoveType.TAIL: 0.2,
...     MoveType.VPLU: 0.4,
...     MoveType.VMIN: 0.4,
... }
>>> restriction_map = (lambda nw: nw.reticulation_number < 2)
>>> sampled_networks = sample_mcmc_networks(
...     starting_network,
...     move_type_probabilities,
...     restriction_map=restriction_map,
...     correct_symmetries=False,
...     burn_in=100,
...     number_of_samples=50,
...     add_root_if_necessary=False,
...     seed=1,
... )
>>> all([network.reticulation_number<2 for network in sampled_networks])
True
>>> all([len(network.leaves)==2 for network in sampled_networks])
True