dmultn(X,Y,B1, weight)
glm.private(Y, start=NULL, weights, X, family)
matrix_threshold(X, lambda, penalty)
lsq_thresholding(b, lambda)
svt(b, lambda)
objfun(alpha, x, y, d, p)
objfun.grad(alpha, x, y, d, p)
objfun.hessian(alpha, x, y, d, p)
MGLM.loss(Y, X, beta, dist, weight, regBeta = FALSE, Beta)