Do the generalized polynomial chaos and Fr?benius methods retain the statistical moments of random differential equations?

详细信息    查看全文

• 作者：Benito M. Chen-Charpentier^a ; ^{bmchen@uta.edu} ; J.-C. Corté ; s^b ; ^{jccortes@imm.upv.es} ; ^{jccortes@mat.upv.es} ; J.-V. Romero^b ; ^{jvromero@imm.upv.es} ; M.-D. Roselló ; ^b ; ^{drosello@imm.upv.es}
• 关键词：Random Fr&ouml ; benius method ; Generalized polynomial chaos ; Statistical moments ; Random differential equations
• 刊名：Applied Mathematics Letters
• 出版年：2013
• 出版时间：May, 2013
• 年：2013
• 卷：26
• 期：5
• 页码：553-558
• 全文大小：380 K

文摘

The aim of this paper is to explore whether the generalized polynomial chaos (gPC) and random Fr?benius methods preserve the first three statistical moments of random differential equations. There exist exact solutions only for a few cases, so there is a need to use other techniques for validating the aforementioned methods in regards to their accuracy and convergence. Here we present a technique for indirectly study both methods. In order to highlight similarities and possible differences between both approaches, the study is performed by means of a simple but still illustrative test-example involving a random differential equation whose solution is highly oscillatory. This comparative study shows that the solutions of both methods agree very well when the gPC method is developed in terms of the optimal orthogonal polynomial basis selected according to the statistical distribution of the random input. Otherwise, we show that results provided by the gPC method deteriorate severely. A study of the convergence rates of both methods is also included.