Tensor comprehensions in SaC

Sven-Bodo Scholz; Artjoms Šinkarovs

doi:10.1145/3412932.3412947

Tensor comprehensions in SaC

Sven-Bodo Scholz, Artjoms Šinkarovs

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

4 Citations (Scopus)

Abstract

We propose a new notation for data parallel operators on multi-dimensional arrays named tensor comprehensions. This notation combines the basic principle of array-comprehensions with syntactical shortcuts very close to those found in the so-called Tensor Notations used in Physics and Mathematics. As a result, complex operators with rich semantics can be defined concisely. The key to this conciseness lies in the ability to define shape-polymorphic operations combined with the ability to infer array shapes from the immediate context. The paper provides a definition of the proposed notation, a formal shape inference process, as well as a set of re-write rules that translates tensor comprehensions as a zero-cost syntactic sugar into standard SaC expressions.

Original language	English
Title of host publication	Proceedings of the 31st Symposium on Implementation and Application of Functional Languages
Editors	Jurrien Stutterheim, Wei Ngan Chin
Publisher	Association for Computing Machinery
ISBN (Electronic)	9781450375627
DOIs	https://doi.org/10.1145/3412932.3412947
Publication status	Published - 25 Sept 2019
Event	31st Symposium on Implementation and Application of Functional Languages 2019 - Singapore, Singapore Duration: 25 Sept 2019 → 27 Sept 2019

Conference

Conference	31st Symposium on Implementation and Application of Functional Languages 2019
Abbreviated title	IFL 2019
Country/Territory	Singapore
City	Singapore
Period	25/09/19 → 27/09/19

ASJC Scopus subject areas

Human-Computer Interaction
Computer Networks and Communications
Computer Vision and Pattern Recognition
Software

Access to Document

10.1145/3412932.3412947

Cite this

@inproceedings{41b11bd12fc342eeaa9511a40a636170,

title = "Tensor comprehensions in SaC",

abstract = "We propose a new notation for data parallel operators on multi-dimensional arrays named tensor comprehensions. This notation combines the basic principle of array-comprehensions with syntactical shortcuts very close to those found in the so-called Tensor Notations used in Physics and Mathematics. As a result, complex operators with rich semantics can be defined concisely. The key to this conciseness lies in the ability to define shape-polymorphic operations combined with the ability to infer array shapes from the immediate context. The paper provides a definition of the proposed notation, a formal shape inference process, as well as a set of re-write rules that translates tensor comprehensions as a zero-cost syntactic sugar into standard SaC expressions. ",

author = "Sven-Bodo Scholz and Artjoms {\v S}inkarovs",

note = "Funding Information: We would like to thank the reviewers for their many constructive suggestions. This work is supported by the Engineering and Physical Sciences Research Council through grants EP/L00058X/1 and EP/N028201/1. Publisher Copyright: {\textcopyright} 2019 Owner/Author.; 31st Symposium on Implementation and Application of Functional Languages 2019, IFL 2019 ; Conference date: 25-09-2019 Through 27-09-2019",

year = "2019",

month = sep,

day = "25",

doi = "10.1145/3412932.3412947",

language = "English",

editor = "Jurrien Stutterheim and Chin, {Wei Ngan}",

booktitle = "Proceedings of the 31st Symposium on Implementation and Application of Functional Languages",

publisher = "Association for Computing Machinery",

address = "United States",

}

Scholz, S-B & Šinkarovs, A 2019, Tensor comprehensions in SaC. in J Stutterheim & WN Chin (eds), Proceedings of the 31st Symposium on Implementation and Application of Functional Languages., 15, Association for Computing Machinery, 31st Symposium on Implementation and Application of Functional Languages 2019, Singapore, Singapore, 25/09/19. https://doi.org/10.1145/3412932.3412947

TY - GEN

T1 - Tensor comprehensions in SaC

AU - Scholz, Sven-Bodo

AU - Šinkarovs, Artjoms

N1 - Funding Information: We would like to thank the reviewers for their many constructive suggestions. This work is supported by the Engineering and Physical Sciences Research Council through grants EP/L00058X/1 and EP/N028201/1. Publisher Copyright: © 2019 Owner/Author.

PY - 2019/9/25

Y1 - 2019/9/25

N2 - We propose a new notation for data parallel operators on multi-dimensional arrays named tensor comprehensions. This notation combines the basic principle of array-comprehensions with syntactical shortcuts very close to those found in the so-called Tensor Notations used in Physics and Mathematics. As a result, complex operators with rich semantics can be defined concisely. The key to this conciseness lies in the ability to define shape-polymorphic operations combined with the ability to infer array shapes from the immediate context. The paper provides a definition of the proposed notation, a formal shape inference process, as well as a set of re-write rules that translates tensor comprehensions as a zero-cost syntactic sugar into standard SaC expressions.

AB - We propose a new notation for data parallel operators on multi-dimensional arrays named tensor comprehensions. This notation combines the basic principle of array-comprehensions with syntactical shortcuts very close to those found in the so-called Tensor Notations used in Physics and Mathematics. As a result, complex operators with rich semantics can be defined concisely. The key to this conciseness lies in the ability to define shape-polymorphic operations combined with the ability to infer array shapes from the immediate context. The paper provides a definition of the proposed notation, a formal shape inference process, as well as a set of re-write rules that translates tensor comprehensions as a zero-cost syntactic sugar into standard SaC expressions.

UR - http://www.scopus.com/inward/record.url?scp=85111133204&partnerID=8YFLogxK

U2 - 10.1145/3412932.3412947

DO - 10.1145/3412932.3412947

M3 - Conference contribution

BT - Proceedings of the 31st Symposium on Implementation and Application of Functional Languages

A2 - Stutterheim, Jurrien

A2 - Chin, Wei Ngan

PB - Association for Computing Machinery

T2 - 31st Symposium on Implementation and Application of Functional Languages 2019

Y2 - 25 September 2019 through 27 September 2019

ER -

Tensor comprehensions in SaC

Abstract

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this