We study the meta-learning of numerical algorithms for scientific computing . It combines the mathematically driven, handcrafted design of general algorithm structure with a data-driven adaptation to specific classes of tasks . This represents a departure from the classical approaches in numericalanalysis, which typically do not feature such learning-based adaptations . We demonstrate that in certain cases we can obtain superiorperformance to classical RK methods. This can be attributed to certainproperties of the ODE families being identified and exploited by the approach.Overall, this work demonstrates an effective, learning . approach to thedesign of algorithms for the numerical solution of differential equations, anapproach that can be readily extended to other numerical tasks. It can be easily extended to . other numerical . tasks. This work shows an effective and efficient approach, which can be

Author(s) : Yue Guo, Felix Dietrich, Tom Bertalan, Danimir T. Doncevic, Manuel Dahmen, Ioannis G. Kevrekidis, Qianxiao Li

Links : PDF - Abstract

Code :
Coursera

Keywords : learning - numerical - tasks - approach - driven -

Leave a Reply

Your email address will not be published. Required fields are marked *