A parallel-in-time approach for accelerating direct-adjoint studies

JOURNAL OF COMPUTATIONAL PHYSICS

卷:429

Article

Skene, C. S.^1,2  Eggl, M. F.^1,3  Schmid, P. J.¹ 

[1] Imperial Coll London, Dept Math, London SW7 2AZ, England

[2] Univ Leeds, Dept Appl Math, Leeds LS2 9JT, W Yorkshire, England

[3] Princeton Univ, Dept Mech & Aerosp Engn, Princeton, NJ 08544 USA

关键词: Parallel-in-time;    Direct-adjoint looping;    Optimization;    Exponential integration;   

DOI  :  10.1016/j.jcp.2020.110033

来源: Elsevier

【 摘 要 】

Parallel-in-time methods are developed to accelerate the direct-adjoint looping procedure. Particularly, we utilize the Paraexp algorithm, previously developed to integrate equations forward in time, to accelerate the direct-adjoint looping that arises from gradient-based optimization. We consider both linear and non-linear governing equations and exploit the linear, time-varying nature of the adjoint equations. Gains in efficiency are seen across all cases, showing that a Paraexp based parallel-in-time approach is feasible for the acceleration of direct-adjoint studies. This signifies a possible approach to further increase the run-time performance for optimization studies that either cannot be parallelized in space or are at their limit of efficiency gains for a parallel-in-space approach. Code demonstrating the algorithms considered in this paper is available. (C) 2020 Elsevier Inc. All rights reserved.

【 授权许可】

Free