positiveFusedLasso <- function(Y, Z, lambda, eps=1e-1,
max.iter=50, warn=FALSE, verbose=FALSE) {
## problem dimensions
M <- length(Y)
p <- ncol(Z[[1]]) ## number of subclones/archetypes/latent features
stopifnot(is.numeric(lambda), length(lambda) == M, length(Z) == M)
if (p > nrow(Y[[1]])) {
warning("Under-identified problem: more latent features than samples")
}
Should the warning "Under-identified problem: more latent features than samples" in
positiveFusedLasso()
be an error instead?