ldhmm.gamma_init: Initializing tansition probability paramter

In ldhmm: Hidden Markov Model for Financial Time-Series Based on Lambda Distribution

Description

This utility has multiple purposes. It can generate a simple transition probability matrix, using p1 and p2, if prob is left as NULL. The generated gamma is raw and not normalized. If prob is provided as a vector, the utility converts it into a matrix as gamma. Furthermore, if prob is provided as a vector or matrix, the utility applies min.gamma, and normalize the sum of t.p.m. rows to 1. This is mainly an internal function used by MLE, not be concerned by external users.

Usage

ldhmm.gamma_init(m, p1 = 0.04, p2 = 0.01, prob = NULL,
  min.gamma = 0)

Arguments

m	numeric, number of states
p1	numeric, the first-neighbor transition probability, default is 0.04.
p2	numeric, the second-neighbor transition probability, default is 0.01.
prob	numeric or matrix, a fully specified transition probability by user, default is NULL. If this is specified, p1, p2 would be ignored.
min.gamma	numeric, a minimum transition probability added to gamma to avoid singularity, default is 0. This is only used when prob is not NULL.

Value

a matrix as gamma

Author(s)

Stephen H. Lihn

Examples

  gamma0 <- ldhmm.gamma_init(m=3)
  prob=c(0.9, 0.1, 0.1, 
         0.1, 0.9, 0.0,
         0.1, 0.1, 0.8)
  gamma1 <- ldhmm.gamma_init(m=3, prob=prob)
  gamma2 <- ldhmm.gamma_init(m=2, prob=gamma1, min.gamma=1e-6)

ldhmm documentation built on Jan. 11, 2020, 9:16 a.m.