An MCMC computational approach for a continuous time state-dependent regime switching diffusion process.

Area of Science:

• Computational Statistics
• Stochastic Processes
• Bayesian Inference

Background:

• State-dependent regime switching diffusion processes (HSD) pose simulation challenges for classical methods.
• Real-world phenomena are observed at discrete times, necessitating data imputation for accurate diffusion process density approximation.

Purpose of the Study:

• To develop and apply a Markov chain Monte Carlo (MCMC) Bayesian approach for estimating complex Hybrid Switching Diffusion (HSD) models.
• To address the simulation difficulties of HSD processes by integrating Bayesian inference.

Main Methods:

• Utilized a Markov chain Monte Carlo (MCMC) Bayesian framework for model estimation.
• Employed a Metropolis Hastings algorithm for updating latent imputed data at random intervals.
• Adapted a forward filtering backward smoothing algorithm for computing regime switching states.
• Applied Gibbs sampler or Metropolis Hastings algorithm for parameter estimation after prior specification.

Main Results:

• Successfully computed latent imputed observations within a Bayesian context.
• Accurately determined regime switching states using an adapted forward filtering backward smoothing algorithm.
• Estimated model parameters through conditional posterior densities using MCMC methods.

Conclusions:

• The proposed MCMC Bayesian approach is effective for estimating and simulating Hybrid Switching Diffusion (HSD) processes.
• Data imputation and Bayesian methods are crucial for handling discrete observations in diffusion modeling.
• The methodology provides a robust framework for analyzing complex stochastic systems.