Dex: Array Programming with Typed Indices

Abstract

Array programming is harder than it should be. Major pain points are managing bulk operations on high-rank arrays, and the associated shape and indexing errors. We describe Dex, a functional array processing language in the Haskell/ML family. Dex introduces a lightweight looping construct and a type system that captures common patterns of array shapes. We hope the language ideas we present here can influence the design of existing array programming systems.

PDF NeurIPSW OpenReview Semantic Scholar

Cite

Text

Maclaurin et al. "Dex: Array Programming with Typed Indices." NeurIPS 2019 Workshops: Program_Transformations, 2019.

Markdown

[Maclaurin et al. "Dex: Array Programming with Typed Indices." NeurIPS 2019 Workshops: Program_Transformations, 2019.](https://mlanthology.org/neuripsw/2019/maclaurin2019neuripsw-dex/)

BibTeX

@inproceedings{maclaurin2019neuripsw-dex,
  title     = {{Dex: Array Programming with Typed Indices}},
  author    = {Maclaurin, Dougal and Radul, Alexey and Johnson, Matthew J. and Vytiniotis, and Dimitrios},
  booktitle = {NeurIPS 2019 Workshops: Program_Transformations},
  year      = {2019},
  url       = {https://mlanthology.org/neuripsw/2019/maclaurin2019neuripsw-dex/}
}