LB_RVB2: Evidence lower bound for RVB2

In hrnasif/rvb: Reparameterized Variational Bayes

Description

Evidence lower bound for RVB2

Usage

LB_RVB2(
  mu,
  C,
  y,
  X,
  Z,
  Zt,
  Zy,
  etahat,
  vbeta0,
  Sinv,
  model,
  m,
  n,
  p,
  r,
  d,
  N = 1000
)

Arguments

mu	Vector. Mean in variational approximation
C	Lower triangular matrix. Cholesky decomposition of variance
y	List. Responses per cluster
X	List. Covariates per cluster for fixed effects
Z	List. Covariates per cluster for random effects
Zt	List. Output from transformZi
Zy	Numeric or vector. Z^T y
etahat	List. Contains per cluster MLEs under RVB1 regularized model
vbeta0	Numeric. Prior variance for fixed effects
Sinv	Matrix. Inverse of prior covariance matrix for random effects
model	Character. Either "poisson" or "binomial"
m	Integer. Number of trials if "binomial"
n	Integer. Number of clusters
p	Integer. Number of fixed effects
r	Integer. Number of random effects
d	Integer. Total number of variables
N	Integer. Number of iterations for ELBO average

Value

Value of the evidence lower bound based on RVB2 optimization

hrnasif/rvb documentation built on Dec. 20, 2021, 4:49 p.m.