orthogonalizeAndQSparsifyOld.Rd
This implements a quantile based sparsification operation
orthogonalizeAndQSparsifyOld(
v,
sparsenessQuantile = 0.5,
positivity = "either",
orthogonalize = TRUE,
softThresholding = FALSE,
unitNorm = FALSE,
sparsenessAlg = NA
)
input matrix
quantile to control sparseness - higher is sparser
restrict to positive or negative solution (beta) weights. choices are positive, negative or either as expressed as a string.
run gram-schmidt if TRUE.
use soft thresholding
set each vector to unit norm
NA is default otherwise basic, spmp or orthorank
matrix
mat <- replicate(100, rnorm(20))
mat <- orthogonalizeAndQSparsify(mat)