update_hyperparams: update hyperparameters in the latent class model with...

In zhenkewu/lotR: Latent class analysis with Observed Trees in R (lotR)

update hyperparameters in the latent class model with observations organized in a tree

Description

update hyperparameters in the latent class model with observations organized in a tree

Usage

update_hyperparams(
  Y,
  A,
  leaf_ids_units,
  ancestors,
  h_pau,
  levels,
  v_units,
  X,
  n,
  J,
  p,
  pL,
  Fg,
  prob,
  prob_gamma,
  mu_gamma,
  mu_alpha,
  rmat,
  sigma_gamma,
  Sigma_alpha,
  tau_1_t,
  tau_2_t,
  a_t,
  b_t,
  E_beta_sq,
  E_eta_sq,
  E_beta,
  E_eta,
  psi,
  g_psi,
  phi,
  g_phi,
  tau_1,
  tau_2,
  shared_tau,
  a,
  b,
  K,
  tau_update_levels,
  update_hyper,
  quiet
)

Arguments

Y	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
A	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
leaf_ids_units	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
ancestors	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
h_pau	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
levels	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
v_units	outputs from design_tree(), which reorders the nodes by internal and leaf nodes; the observations are also ordered from low- to high-indexed leaf nodes.
X	computed from the data by lcm_tree() at the beginning before the VI updates
n	computed from the data by lcm_tree() at the beginning before the VI updates
J	computed from the data by lcm_tree() at the beginning before the VI updates
p	computed from the data by lcm_tree() at the beginning before the VI updates
pL	computed from the data by lcm_tree() at the beginning before the VI updates
Fg	computed from the data by lcm_tree() at the beginning before the VI updates
prob	variational parameters updated by update_vi_params()
prob_gamma	variational parameters updated by update_vi_params()
mu_gamma	variational parameters updated by update_vi_params()
mu_alpha	variational parameters updated by update_vi_params()
rmat	variational parameters updated by update_vi_params()
sigma_gamma	variational parameters updated by update_vi_params()
Sigma_alpha	variational parameters updated by update_vi_params()
tau_1_t	variational parameters updated by update_vi_params()
tau_2_t	variational parameters updated by update_vi_params()
a_t	variational parameters updated by update_vi_params()
b_t	variational parameters updated by update_vi_params()
E_beta_sq, E_eta_sq, E_beta, E_eta	moments computed by update_vi_params()
psi	parameters updated by update_hyperparams()
g_psi	parameters updated by update_hyperparams()
phi	parameters updated by update_hyperparams()
g_phi	parameters updated by update_hyperparams()
tau_1	parameters updated by update_hyperparams()
tau_2	parameters updated by update_hyperparams()
shared_tau	parameters updated by update_hyperparams()
a	fixed hyperparameters not to be updated.
b	fixed hyperparameters not to be updated.
K	fixed hyperparameters not to be updated.
tau_update_levels	a numeric vector, specifies which levels of hyperparameters to update
update_hyper	Logical, TRUE or FALSE to indicate whether to update tau_1 and tau_2. This is computed at every iteration in fit_lcm_tree()
quiet	default to FALSE, which prints intermediate updates of hyperparameters

Value

a list of updated hyperparameters: tau_1,tau_2,psi,g_psi,phi,g_phi, along with a new ELBO value.

zhenkewu/lotR documentation built on April 24, 2022, 2:36 a.m.