Reducing the memory requirement in reverse mode automatic differentiation by solving TBR flow equations.

PDF Version Also Available for Download.

Description

The fast computation of gradients in reverse mode Automatic Differentiation (AD) requires the generation of adjoint versions of every statement in the original code. Due to the resulting reversal of the control flow certain intermediate values have to be made available in reverse order to compute the local partial derivatives. This can be achieved by storing these values or by recomputing them when they become required. In any case one is interested in minimizing the size of this set. Following an extensive introduction of the ''To-Be-Recorded'' (TBR) problem the authors present flow equations for propagating the TBR status of variables ... continued below

Physical Description

10 pages

Creation Information

Naumann, U. January 11, 2002.

Context

This article is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided by UNT Libraries Government Documents Department to Digital Library, a digital repository hosted by the UNT Libraries. More information about this article can be viewed below.

Who

People and organizations associated with either the creation of this article or its content.

Author

Sponsor

Publisher

Provided By

UNT Libraries Government Documents Department

Serving as both a federal and a state depository library, the UNT Libraries Government Documents Department maintains millions of items in a variety of formats. The department is a member of the FDLP Content Partnerships Program and an Affiliated Archive of the National Archives.

Contact Us

What

Descriptive information to help identify this article. Follow the links below to find similar items on the Digital Library.

Description

The fast computation of gradients in reverse mode Automatic Differentiation (AD) requires the generation of adjoint versions of every statement in the original code. Due to the resulting reversal of the control flow certain intermediate values have to be made available in reverse order to compute the local partial derivatives. This can be achieved by storing these values or by recomputing them when they become required. In any case one is interested in minimizing the size of this set. Following an extensive introduction of the ''To-Be-Recorded'' (TBR) problem the authors present flow equations for propagating the TBR status of variables in the context of reverse mode AD of structured programs.

Physical Description

10 pages

Source

  • 2002 International Conference on Computational Science (ICCS 2002), Amsterdam (NL), 04/21/2002--04/24/2002

Language

Item Type

Identifier

Unique identifying numbers for this article in the Digital Library or other systems.

  • Report No.: ANL/MCS/CP-106609
  • Grant Number: W-31-109-ENG-38
  • Office of Scientific & Technical Information Report Number: 793901
  • Archival Resource Key: ark:/67531/metadc741615

Collections

This article is part of the following collection of related materials.

Office of Scientific & Technical Information Technical Reports

What responsibilities do I have when using this article?

When

Dates and time periods associated with this article.

Creation Date

  • January 11, 2002

Added to The UNT Digital Library

  • Oct. 19, 2015, 7:39 p.m.

Description Last Updated

  • March 30, 2016, 3:24 p.m.

Usage Statistics

When was this article last used?

Yesterday: 0
Past 30 days: 2
Total Uses: 7

Interact With This Article

Here are some suggestions for what to do next.

Start Reading

PDF Version Also Available for Download.

Citations, Rights, Re-Use

Naumann, U. Reducing the memory requirement in reverse mode automatic differentiation by solving TBR flow equations., article, January 11, 2002; Illinois. (digital.library.unt.edu/ark:/67531/metadc741615/: accessed August 23, 2017), University of North Texas Libraries, Digital Library, digital.library.unt.edu; crediting UNT Libraries Government Documents Department.