armijo.lr Performs Armijo backtracking line search along a pre-specified search direction
armijo.lr(y0, theta, theta.tmp, alpha.mat = 0, w, z, lambda1, var.type,
b = 0.5, thresh = 1e-05, nu = 0.1, zeta = 0.1, th = 0.1,
step = 1)nxp observations matrix
nxp matrix direction of update for matrix of interactions
nxp matrix, current matrix of interactions
nxp matrix, matrix of main effects
weights of the quadratic approximation
matrix around which the quadratic approximation is done
positive number, regularization parameter for nuclear norm penalty
vector of length p indicating column types for y (gaussian, binary, poisson)
positive number in (0,1) factor by which the step size is reduced
positive number, convergence criterion
positive number, backtracking line search parameter, default 0.01
positive number, backtracking line search parameter, default 0.1
positive number, backtracking line search parameter, default 0.1
number in (0,1), initial backtracking line search step, default 1
A list with the following elements
(nb individuals) x (nb variables) matrix containing the individual effects
a vector containing the value of the objective function at every iteration
the step size