Skip to content

Project: implement ZZ mod n in the engine for composite n

Daniel R. Grayson edited this page Oct 22, 2021 · 6 revisions
  • Potential advisor/consultant(s): Mike Stillman
  • Goal: Implement ZZ/n for composite n in the engine
  • Current status: volunteers welcome!
  • Macaulay2 skill level: beginner with some Macaulay2 experience, knowledge of C++ is necessary
  • Mathematical experience: advanced undergraduate
  • Reason(s) to participate: learn Macaulay2 internals, provide a faster implementation of ZZ/n to yourself and to other researchers

Project Description

Currently ZZ/n "works" only for prime n.
Implementing ZZ/n for composite n in the engine would allow fast operations (e.g., linear algebra) over ZZ/n for arbitrary n.

Workaround

Currently one can create a quotient of a polynomial ring with no variables: ZZ[]/n. This "works", albeit slowly.

Clone this wiki locally