uniform_succ_given_alpha: Second step of SUCCOTASH with uniform mixture.

In dcgerard/succotashr: Surrogate and Confounder Correction Occuring Together with Adaptive Shrinkage

Description

Second step of SUCCOTASH with uniform mixture.

Usage

uniform_succ_given_alpha(Y, alpha, sig_diag, num_em_runs = 2, a_seq = NULL,
  b_seq = NULL, lambda = NULL, em_itermax = 200, em_tol = 10^-6,
  pi_init = NULL, Z_init = NULL, em_z_start_sd = 1,
  pi_init_type = c("random", "uniform", "zero_conc"),
  lambda_type = c("zero_conc", "ones"), print_progress = TRUE,
  print_ziter = FALSE, true_Z = NULL, var_scale = TRUE,
  optmethod = c("coord", "em"), likelihood = c("normal", "t"), df = NULL,
  z_init_type = c("null_mle", "random"), var_scale_init_type = c("null_mle",
  "one", "random"))

Arguments

Y	A p by 1 matrix of numerics. The data.
alpha	A p by k matrix of numerics. The confounder coefficients.
sig_diag	A vector of the variances of Y.
num_em_runs	An integer. The number of em iterations to perform, starting at random locations.
a_seq	A vector of negative numerics in increasing order. The negative end points in an [a, 0] grid.
b_seq	A vector of positive numerics in increasing order. The positive end points in a [0, b] grid.
lambda	A vector of numerics greater than or equal to 1, of length length(a) + length(b) + 1.
em_itermax	A positive integer. The maximum number of iterations to perform on the em step.
em_tol	A positive numeric. The stopping criterion for the EM algorithm.
pi_init	A vector of non-negative numerics that sum of 1 of length length(a) + length(b) + 1. The initial values of the mixture probs.
Z_init	A vector of length k of numerics. Starting values of Z.
em_z_start_sd	A positive numeric. Z is initialized by iid normals with this standard deviation and mean 0.
pi_init_type	Either "random", "uniform", or "zero_conc". How should we choose the initial mixture probabilities if pi_init is NULL? "random" will draw draw pi uniformly from the simplex. "uniform" will give each value equal mass. "zero_conc" will give more mass to 0 than any other probability.
lambda_type	How should we regularize? 'unif' gives no regularization. 'zero_conf' gives regularization at zero alone.
print_progress	A logical. Should we plot the progress?
print_ziter	A logical. Should we print the progress of the Newton iterations for updating Z?
true_Z	The true Z values. Used for testing.
var_scale	A logical. Should we update the scaling on the variances (TRUE) or not (FALSE).
optmethod	Either coordinate ascent ("coord") or an EM algorithm ("em"). Coordinate ascent is currently only implemented in the uniform mixtures case, for which it is the default.
likelihood	Which likelihood should we use? Normal ("normal") or t ("t")?
df	The degrees of freedom of the the t-likelihood if likleihood = "t".
z_init_type	How should we initiate the confounders? At the all-null MLE ("null_mle") or from iid standard normals ("random")?
var_scale_init_type	If var_scale = TRUE, how should we initiate the variance inflaiton parameter? From the all-null MLE ("null_mle"), at no inflation ("one"), or from a chi-squared distribution with one degree of freedom ("random")?

See Also

succotash_llike_unif succotash_unif_fixed

dcgerard/succotashr documentation built on May 15, 2019, 1:25 a.m.