Skip to contents

An R6 Class object, a collection of model for population of sbm netowrks

bmpop.Rd

An R6 Class object, a collection of model for population of sbm netowrks

An R6 Class object, a collection of model for population of sbm netowrks

Public fields

n

A list of size M with the number of nodes per network

A

List of incidence Matrix of size n \times n

M

Number of networks

mask

List of M masks, indicating NAs in the matrices. 1 for NA, 0 else

directed

A boolean indicating if the networks are directed or not

distribution

Emission distribution either : "poisson" or "bernoulli"

net_id

A vector containing the "ids" or names of the networks (if none given, they are set to their number in A list)

model_list

A list of size Q containing the best models

global_opts

A list of options for the model space exploration. See details for more information on which options are available.

fit_opts

A list of options specifically for fitting the models.

fit_sbm

Pre-fitted sbm objects

Z_init

A list of initializations for the Z memberships.

free_density

A boolean indicating if we consider free density or if all networks have the same density

free_mixture

A boolean signaling if there is free mixture

ICL_sbm

A list storing the ICL for each sbm fitted

ICL

A list of size Q storing the best ICL found for each Q

BICL

A list of size Q storing the best BICL found for each Q

vbound

A list of size Q storing the best vbound found for each Q

best_fit

A fitSimpleSBMPop object changing regularly to store the current best fit.

logfactA

A quantity used with the Poisson probability distribution

improved

A field use at each step to check if it has improved.

Methods

Public methods

Method new()

Create a new instance of the bisbmpop object

This class is generally called via the user function estimate_colBiSBM

Usage

bmpop$new(
  netlist = NULL,
  net_id = NULL,
  directed = NULL,
  distribution = "bernoulli",
  free_density = FALSE,
  free_mixture = FALSE,
  fit_sbm = NULL,
  Z_init = NULL,
  global_opts = list(),
  fit_opts = list()
)

Arguments

netlist

The list of M networks

net_id

A list of name for the networks, defaults to 1 to M if not provided

directed

A boolean indicating if the networks are directed or not

distribution

The emission distribution either "bernoulli" or "poisson"

free_density

If we account for different density between networks

free_mixture

A boolean indicating if there is free mixture

fit_sbm

The pre-fitted SBM

Z_init

A list providing a clustering of the nodes

global_opts

A list of global options used by the algorithm. See details of the user function for more information.

fit_opts

A list of fit options used by the algorithm. See details of the user function for more information.

Returns

A new 'sbmpop' object. Fit a list of SBM if fit_sbm == TRUE

Method optimize_sbm()

Usage

bmpop$optimize_sbm()

Returns

nothing; but stores the values Fit the colSBM mode using sbm as initializations

Method optimize_from_sbm()

Usage

bmpop$optimize_from_sbm(index, Q, nb_clusters)

Arguments

index

The sequence of networks number going from 1 to M

Q

The number of clusters

nb_clusters

A subindex for model list, in practice always 1

Returns

bmpop object Fit the colSBM mode using spectral decompositions as initializations

Method optimize_spectral()

Usage

bmpop$optimize_spectral(index, Q, nb_clusters)

Arguments

index

The sequence of networks number going from 1 to M

Q

The number of clusters

nb_clusters

A subindex for model list, in practice always 1

Returns

bmpop object Fit a colSBM model using a given initialization provided with Z for a specific network

Method optimize_init()

Usage

bmpop$optimize_init(index, Z, Q, nb_clusters, Cpi = NULL, Calpha = NULL)

Arguments

index

The sequence of networks number going from 1 to M

Z

The provided initialization

Q

The number of clusters

nb_clusters

A subindex for model list, in practice always 1

Cpi

A list of size M containing the support for the possibly absent blocks, defaults to NULL

Calpha

A list of size M containing the support for the possibly absent interaction parameter, defaults to NULL

Returns

bmpop object Fit the whole colSBM mode using a given initialization provided with a given Z_init

Method optimize_from_zinit()

Usage

bmpop$optimize_from_zinit(index, Q, nb_clusters)

Arguments

index

The sequence of networks number going from 1 to M

Q

The number of clusters

nb_clusters

A subindex for model list, in practice always 1

Returns

bmpop object Burn-in method that performs the initialization necessary to next begin the search and model selection

Method burn_in()

Usage

bmpop$burn_in()

Returns

nothing; but stores the values This forward pass split the clusters found previously and reallocate the nodes to the new clusters

Method forward_pass()

Usage

bmpop$forward_pass(
  Q_min = self$global_opts$Q_min,
  Q_max = self$global_opts$Q_max,
  index = seq(self$M),
  nb_clusters = 1L
)

Arguments

Q_min

The minimal number of clusters, defaults to the global options Q_min parameter

Q_max

The max number of clusters, defaults to the global options Q_max parameter

index

The sequence of networks number going from 1 to M

nb_clusters

A subindex for model list, in practice always 1

Returns

Q - 1 This backward pass merges the clusters found previously

Method backward_pass()

Usage

bmpop$backward_pass(
  Q_min = self$global_opts$Q_min,
  Q_max = self$global_opts$Q_max,
  index = seq(self$M),
  nb_clusters = 1L
)

Arguments

Q_min

The minimal number of clusters, defaults to the global options Q_min parameter

Q_max

The max number of clusters, defaults to the global options Q_max parameter

index

The sequence of networks number going from 1 to M

nb_clusters

A subindex for model list, in practice always 1

Returns

Q + 1 The optimization method

Method optimize()

This method performs the burn in and the steps of moving window with cluster splitting (forward pass) and merging (backward pass) around the mode found.

Usage

bmpop$optimize()

Returns

nothing; but stores the values Performs the model selection based on the BICL criterion

Method choose_models()

Usage

bmpop$choose_models(models, Q, index = seq(self$M), nb_clusters = 1L)

Arguments

models

The list of models in which to choose the best

Q

The value of Q for which the model selection is performed

index

The sequence of networks number going from 1 to M

nb_clusters

A subindex for model list, in practice always 1

Returns

The best models The message printed when one prints the object

Method show()

Usage

bmpop$show(type = "Fitted Collection of Simple SBM")

Arguments

type

The title above the message. The print method

Method print()

Usage

bmpop$print()

Returns

nothing; print to console

Method clone()

The objects of this class are cloneable with this method.

Usage

bmpop$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.