bp.computeAlpha {EDISON}R Documentation

Computes the acceptance ratio of two changepoint configurations.

Description

This function computes the acceptance ratio of two changepoint configurations with networks in a changepoint birth or death move.

Usage

bp.computeAlpha(birth, lNew, kminus, Ekl, Estar, Ekr, yL, PxL, yR, PxR, y2, Px2,
  D, delta2, q, smax, v0, gamma0, prior_ratio = 1)

Arguments

birth

1 for a changepoint birth move, -1 for a changepoint death move.

lNew

Number of edges in the new segment.

kminus

Minimal number of changepoints between the two compared models (equal to s for a birth move, s-1 for a death move.

Ekl

Changepoint on the left of proposed changepoint.

Estar

Changepoint being inserted or deleted.

Ekr

Changepoint on the right of proposed changepoint.

yL

Response data (left).

PxL

Projection matrix (left).

yR

Response data (right).

PxR

Projection matrix (right).

y2

Response data (both).

Px2

Projection matrix (both).

D

Hyperparameters for the number of edges in each segment.

delta2

Hyperparameters for the empirical covariance (signal-to-noise ratio).

q

Total number of nodes in the network.

smax

Maximum number of changepoints.

v0

Hyperparameter.

gamma0

Hyperparameter.

prior_ratio

Ratio of network structure priors.

Author(s)

Sophie Lebre

References

For more information about the model, see:

Dondelinger et al. (2012), "Non-homogeneous dynamic Bayesian networks with Bayesian regularization for inferring gene regulatory networks with gradually time-varying structure", Machine Learning.

See Also

cp.birth, cp.death


[Package EDISON version 1.1.1 Index]