Parallel tempering for dynamic generalized linear models |
| |
Authors: | Guangbao Guo Wei Shao Lu Lin Xuehu Zhu |
| |
Institution: | 1. Department of Statistics, Shandong University of Technology, Zibo, China;2. School of Mathematics, Shandong University, Jinan, China;3. School of Management, Qufu Normal University, Rizhao, China;4. School of Mathematics, Shandong University, Jinan, China;5. School of Mathematics and Statistics, Xi’an Jiaotong University, Xi’an, China |
| |
Abstract: | ABSTRACTMarkov chain Monte Carlo (MCMC) methods can be used for statistical inference. The methods are time-consuming due to time-vary. To resolve these problems, parallel tempering (PT), as a parallel MCMC method, is tried, for dynamic generalized linear models (DGLMs), as well as the several optimal properties of our proposed method. In PT, two or more samples are drawn at the same time, and samples can exchange information with each other. We also present some simulations of the DGLMs in the case and provide two applications of Poisson-type DGLMs in financial research. |
| |
Keywords: | Dynamic generalized linear models Parallel tempering Time series Total variation norm |
|
|