Parameter Estimation of Coupled Oscillation System through Markov Chain Monte Carlo Method

楊長茂 段必輝
Parameter estimation is crucial in scientific research, especially in data analysis and modeling. Traditional optimization techniques, such as the least square method, often face challenges with increasing parameter dimensionality. The Markov Chain Monte Carlo (MCMC) method, with its Metropolis-Hastings algorithm, emerges as a powerful alternative, finding applications in areas like gravitational wave data analysis.
In this context, coupled oscillators, which offer insights into phenomena like lattice vibrations, become an interesting subject of study. We constructed an experimental system of oscillators connected through springs with unknown constants. Using the MCMC method, we aimed to estimate these unknown spring constants. Our results indicate that MCMC not only enhances computational efficiency but also provides more accurate parameter estimations compared to traditional methods.