fit_variational Function — fit_variational

This function performs the inference using the ADVI algorithm. Repeat inference if it fails and repeat to avoid local minima, taking the best run.

Usage

fit_variational_h(
  input_data,
  max_attempts = 2,
  initialization = NULL,
  INIT = TRUE,
  initial_iter = 1000,
  grad_samples = 10,
  elbo_samples = 100,
  tolerance = 0.01
)

Arguments

input_data: list: List of 7: $S: int, $N: int, $karyotype: num (0 or 1), $seg_assignment: num, $peaks:List of N of num (1:2), $NV: num, $DP: num
max_attempts: num: max number of repeated inference for ADVI
initialization: list: List of 4: $w: num (1:S, 1:3), $tau: num (1:K), $phi: num (1:K), $kappa: num
INIT: logical: boolean variable to set the initialization phase to TRUE or FALSE
initial_iter: description
grad_samples: description
elbo_samples: description
tolerance: num: tolerance in the ELBO optimization procedure

Value

best_fit