look_for_mode_x computes optimization algorithms to find the mode of the posterior
Usage
look_for_mode_x(
A = A,
x0,
tol0,
tol1,
k0,
a = 0.5,
y,
d,
n,
strategy = "ls-quasi-newton",
Qx,
verbose,
cores
)Arguments
- A
Matrix which links latent field with linear predictor.
- x0
Initial optimization value.
- tol0
Tolerance for |x_new - x_old| and |f_new - f_old|.
- tol1
Tolerance for the gradient such that |grad| < tol1 * max(1, |f|)
- k0
Number of iterations.
- a
Step length in the algorithm.
- y
Response variable. Number of columns correspond to the number of categories.
- d
Number of categories.
- n
Number of individuals.
- strategy
Strategy to use to optimize.
- Qx
Prior precision matrix for the fixed effects.
- verbose
By default is FALSE. If TRUE, the computation process is shown in the scream.
- cores
Number of cores for parallel computation. The package parallel is used.
Value
x_hat Matrix with the x of the iterations.
Hk Hessian in eta. This Hessian is a combination of the real Hessian (when it is positive definite) and the expected Hessian (when the real Hessian is not positive definite).
gk Gradient in eta.
Lk Cholesky decomposition matrix.
eta Linear predictor.
z New pseudo observation conditioned to eta.
Author
Joaquín Martínez-Minaya jomarminaya@gmail.com