Peer-Reviewed Journal Details
Mandatory Fields
Siskind J.;Pearlmutter B.
Optimization Methods and Software
Divide-and-conquer checkpointing for arbitrary programs with no user annotation
1 ()
Optional Fields
46G05 58C20 65D25 65F50 68N18 68N20 binomial checkpointing compiler theory lambda calculus programming language theory Reverse-mode automatic differentiation treeverse
© 2018, © 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. Classical reverse-mode automatic differentiation (AD) imposes only a small constant-factor overhead in operation count over the original computation, but has storage requirements that grow, in the worst case, in proportion to the time consumed by the original computation. This storage blowup can be ameliorated by checkpointing, a process that reorders application of classical reverse-mode AD over an execution interval to tradeoff space vs. time. Application of checkpointing in a divide-and-conquer fashion to strategically chosen nested execution intervals can break classical reverse-mode AD into stages which can reduce the worst-case growth in storage from linear to sublinear. Doing this has been fully automated only for computations of particularly simple form, with checkpoints spanning execution intervals resulting from a limited set of program constructs. Here we show how the technique can be automated for arbitrary computations. The essential innovation is to apply the technique at the level of the language implementation itself, thus allowing checkpoints to span any execution interval.
Grant Details