Automated Generation of Performance Values for Algorithmic Differentiation. Issue 1 (October 2016)

• Record Type: Journal Article
• Authors: Sagebaum, Max; Albring, Tim; Gauger, Nicolas R.
• Is Part Of: Proceedings in applied mathematics and mechanics. Volume 16:Issue 1(2016)
• Page Start: 863
• Page End: 864
• Publication Date: 2016-10
• Subjects: Applied mathematics -- Periodicals; Engineering mathematics -- Periodicals; Mathematical physics -- Periodicals; 519
• DOI: 10.1002/pamm.201610420
• Languages: English
• ISSNs: 1617-7061

Abstract

Abstract: The transition from simulation software to optimization software is often quite cumbersome and error prone. With Algorithmic Differentiation(AD) a semi automatic transition can be achieved, but most of the times the performance characteristics of the modified software are not clear in advance. Therefore the performance model from Griewank is tested against real world examples and an automated measurement for the performance characteristics of the hardware is proposed. This enables a more accurate calculation of the performance characteristics of the modified software. (© 2016 Wiley‐VCH Verlag GmbH & Co. KGaA, Weinheim)