zbMATH — the first resource for mathematics

Storing versus recomputation on multiple DAGs. (English) Zbl 1251.65025
Forth, Shaun (ed.) et al., Recent advances in algorithmic differentiation. Selected papers based on the presentations at the 6th international conference on automatic differentiation (AD2012), Fort Collins, CO, USA, July 23–27, 2012. Berlin: Springer (ISBN 978-3-642-30022-6/hbk; 978-3-642-30023-3/ebook). Lecture Notes in Computational Science and Engineering 87, 197-207 (2012).
Summary: Recomputation and storing are typically seen as tradeoffs for checkpointing schemes in the context of adjoint computations. At finer granularity during the adjoint sweep, in practice, only the store-all or recompute-all approaches are fully automated. This paper considers a heuristic approach for exploiting finer granularity recomputations to reduce the storage requirements and thereby improve the overall adjoint efficiency without the need for manual intervention.
For the entire collection see [Zbl 1247.65002].
65D25 Numerical differentiation
OpenAD/F; revolve; TAF
Full Text: DOI