Lombardi Anna Maria
Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, 00143 Roma, Italy.
Sci Rep. 2015 Feb 12;5:8417. doi: 10.1038/srep08417.
This paper proposes a new algorithm to estimate the maximum likelihood parameters of an Epidemic Type Aftershock Sequences (ETAS) model. It is based on Simulated Annealing, a versatile method that solves problems of global optimization and ensures convergence to a global optimum. The procedure is tested on both simulated and real catalogs. The main conclusion is that the method performs poorly as the size of the catalog decreases because the effect of the correlation of the ETAS parameters is more significant. These results give new insights into the ETAS model and the efficiency of the maximum-likelihood method within this context.
本文提出了一种新算法,用于估计余震序列(ETAS)模型的最大似然参数。该算法基于模拟退火,这是一种通用方法,可解决全局优化问题并确保收敛到全局最优解。该程序在模拟目录和真实目录上均进行了测试。主要结论是,随着目录规模减小,该方法的性能较差,因为ETAS参数相关性的影响更为显著。这些结果为ETAS模型以及在此背景下最大似然方法的效率提供了新的见解。