English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

The identification of primal variables and adjoint variables is usually donevia indices in operator overloading algorithmic differentiation tools. Oneapproach is a linear management scheme, which is easy to implement and supportsmemory optimization for copy statements. An alternative approach performs areuse of indices, which requires more implementation effort but results in muchsmaller adjoint vectors. Therefore, the vector mode of algorithmicdifferentiation scales better with the reuse management scheme. In this paper,we present a novel approach that reuses the indices and allows the copyoptimization, thus combining the advantages of the two aforementioned schemes.The new approach is compared to the known approaches on a simple synthetic testcase and a real-world example using the computational fluid dynamics solverSU2.

Assign optimization for algorithmic differentiation reuse index management strategies