estimateBLL: Estimate Ornstein-Uhlenbeck process

In gherardovarando/crossSectional: Estimate Dynamical Model from Cross Sectional Data

Description

Estimate the coefficients and the noise term of an Ornstein-Uhlenbeck process (dX(t) = BX(t) + √{C} dW(t)) using penalized maximum-likelihood.

Usage

estimateBLL(Sigma, B, C = diag(ncol(Sigma)), eps = 0.01, alpha = 0.2,
  beta = 0.5, maxIter = 1000, trace = 0, lambda = 0, r = FALSE,
  h = FALSE, pert = FALSE)

estimateCLL(Sigma, B, C = diag(ncol(Sigma)), C0 = diag(ncol(Sigma)),
  eps = 0.01, alpha = 0.2, beta = 0.5, maxIter = 1000, trace = 0,
  lambda = 0, r = FALSE, t0 = 1)

estimateBF(Sigma, B, C = diag(ncol(Sigma)), C0 = diag(ncol(Sigma)),
  eps = 0.01, alpha = 1, maxIter = 1000, trace = 0, lambda = 0.1,
  beta = 0.5, r = FALSE, h = FALSE)

Arguments

Sigma	the empirical covariance matrix
B	an initial guess for the coefficient matrix B
C	the initial guess for the noise matrix C
eps	stopping criteria
alpha	parameter backtracking
beta	parameter backtracking
maxIter	maximum number of iterations
trace	if >0 print info
lambda	penalization coefficient
r	logical the set of parameter will be updated at every iteration
h	logical if TRUE only the non-zero entries of B will be updated
C0	penalization matrix
delt	stepsize for internal stable approx

Value

the estimated B matrix (estimateBLL) or the estiamted C matrix (estiamteCLL).

gherardovarando/crossSectional documentation built on July 7, 2019, 12:44 a.m.