Parameter estimation in SDEs via the Fokker-Planck equation: Likelihood function and adjoint based gradient computation

JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS

卷:465

Article

Kaltebbacher, Barbara¹  Pedretscher, Barbara² 

[1] Alpen Adria Univ Klagenfurt, Dept Math, Univ Str 65-67, A-9020 Klagenfurt, Austria

[2] KAI Kompetenzzentrum Automobil & Ind Elekt GmbH, Europastr 8, A-9524 Villach, Austria

关键词: Parameter identification;    Stochastic differential equation;    State space model;    Likelihood function;    Adjoint method;   

DOI  :  10.1016/j.jmaa.2018.05.048

来源: Elsevier

【 摘 要 】

In this paper we consider the problem of identifying parameters in stochastic differential equations. For this purpose, we transform the originally stochastic and nonlinear state equation to a deterministic linear partial differential equation for the transition probability density. We provide an appropriate likelihood cost function for parameter fitting, and derive an adjoint based approach for the computation of its gradient. (C) 2018 Elsevier Inc. All rights reserved.

【 授权许可】

Free