Refit subject to sparsity constraints

Source: R/refit_log_contrast.R

Given output of sparse_log_contrast, solves the least squares problem with compositional constraint on the features selected by sparse_log_contrast.

refit_sparse_log_contrast(fit, Z, y, tol = 1e-05)

Arguments

fit

output of trac
Z, y

same arguments as passed to sparse_log_contrast
tol

tolerance for deciding whether a beta value is zero

Details

minimize_beta, beta0 1/(2n) || y - beta0 1_n - Z beta ||^2 subject to beta_nonselected = 0, 1_p^T beta = 0