estimate_mixprop: Estimate mixture proportions of a mixture g given noisy (error-prone) data from that mixture.

Description

Estimate mixture proportions of a mixture g given noisy (error-prone) data from that mixture.

Usage

estimate_mixprop(
  data,
  g,
  prior,
  optmethod = c("mixSQP", "mixEM", "mixVBEM", "cxxMixSquarem", "mixIP", "w_mixEM"),
  control,
  weights = NULL
)

Value

list, including the final loglikelihood, the null loglikelihood, an n by k likelihood matrix with (j,k)th element equal to \(f_k(x_j)\), the fit and results of optmethod

Arguments

data: list to be passed to log_comp_dens_conv; details depend on model
g: an object representing a mixture distribution (eg normalmix for mixture of normals; unimix for mixture of uniforms). The component parameters of g (eg the means and variances) specify the components whose mixture proportions are to be estimated. The mixture proportions of g are the parameters to be estimated; the values passed in may be used to initialize the optimization (depending on the optmethod used)
prior: numeric vector indicating parameters of "Dirichlet prior" on mixture proportions
optmethod: name of function to use to do optimization
control: list of control parameters to be passed to optmethod, typically affecting things like convergence tolerance
weights: vector of weights (for use with w_mixEM; in beta)

Details

This is used by the ash function. Most users won't need to call this directly, but is exported for use by some other related packages.