sigex.mlefit: Fit model to the data using ML estimation

View source: R/sigex.mlefit.r

sigex.mlefitR Documentation

Fit model to the data using ML estimation

Description

Fit model to the data using ML estimation

Usage

sigex.mlefit(
  data.ts,
  param,
  constraint,
  mdl,
  method,
  thresh = Inf,
  hess = TRUE,
  whittle = FALSE,
  debug = FALSE,
  maxiter = 100
)

Arguments

data.ts

A T x N matrix ts object; any missing values must be encoded with 1i in that entry

param

model parameters entered into a list object with an intuitive structure. This is an initial specification to start the nonlinear optimization routines

constraint

Matrix of the form [Q , C], with C (constraint.mat) the matrix of constraints and Q (constraint.vec) the vector of constraint constants, such that C psi = Q. Use NULL if there are no constraints

mdl

The specified sigex model, a list object

method

"bfgs" for BFGS, "sann" for simulated annealing, "cg" for conjugate gradient

thresh

Pre-parameters theta satisfy |theta|< thresh; set thresh = Inf if no thresholding is desired

hess

A Boolean flag; if true, for BFGS it runs another round of BFGS to get Hessian matrix

whittle

A Boolean flag; if true, uses Whittle likelihood instead of default Gaussian likelihood

debug

A Boolean flag; if true, sets the DEBUGGING mode, which prints psi.last and psi.now to the global field. Also lik will be printed. psi.now gives the parameter that crashed the likelihood (if it crashed), psi.last gives the last good parameter that did not crash the lik.

maxiter

Maximum number of iterations allowed in optimization routines

Value

list with mle and par.est mle: an object of type outputted by optim par.est: type param, with the estimated parameters filled in


jlivsey/sigex documentation built on May 25, 2024, 4:17 a.m.