penalty_BirgeMassart_shape1: Penalty function type Birgé-Massart 1

View source: R/model_selection.R

penalty_BirgeMassart_shape1R Documentation

Penalty function type Birgé-Massart 1

Description

penalty_BirgeMassart_shape1 is the penalty shape defined by : pen_shape = (sqrt(K) + sqrt(2 * K * L_K))^2 with sum(exp(- K * L_K)) < infty : L_K = B + 1/K * log(model_complexity).

Usage

penalty_BirgeMassart_shape1(K, p, model_complexity, B = 0.1)

Arguments

K

the number of shifts

p

the dimension of the data

model_complexity

the complexity of the set of models with dimension K

B

a non-negative constant. Default is 0.1 (as suggested in Cleynen & Lebarbier 2015)

Details

See Birgé Massart (2001). Must be applied to least-square criterion. This penalty should be calibrated using the slope heuristic.

Value

value of the penalty

See Also

penalty_BaraudGiraudHuet_likelihood, penalty_BirgeMassart_shape2


PhylogeneticEM documentation built on Aug. 31, 2022, 9:16 a.m.