Skip to main content

A Python package for group action.

Project description

QUESTION

When you have to design a standard cell libray, you need to choose a number of combinatorial cells. You can simply ask a friend and start from an existing list or build your own from scratch. But how? More specifically, how many n-input Boolean functions are there?

This latter question connects to the mathematical concept of group action that is a pivotal part of Algebra. If you like Rubyk's cube, you should not feel too uncomfortable.

It can be reformulated in this context as follows. What are the orbits in the action of the symmetric group $S_n$ on the set of n-input Boolean functions $X_n=B^{B^n}$?

DESCRIPTION

The name of this package is group_action. The version of this package is 0.1.10. It contains a library module named library and an application named orbits.

It computes the orbits in the action of the symmetric group $S_n$ on the set of n-input Boolean functions $X_n=B^{B^n}$.

The result is basically a list of signatures that is presented in 2 formats and 2 levels of details. Here, a signature is a non negative integer representing a Boolean function.

For instance, what 2-input Boolean function does 12 represent? I use Big Endian format for binary words. 12 = 0101 This gives the following truth table

x0 x1 f(x0, x1)
0  0  0
1  0  1
0  1  0
1  1  1

and the following expression f(x0, x1) = x0 & ~x1 | x0 & x1 = x0

Note: x1 does not figure in this expression.

So 12 represents a 1-input Boolean function. It gives a standard cell called buffer.

To move forward, is 12 the only signature representing a buffer cell? Actually this answer is no! If you permut x0 and x1, you get the following truth table.

x0 x1 f(x0, x1)
0  0  0
1  0  0
0  1  1
1  1  1

It corresponds to f(x0, x1) = ~x0 & x1 | x0 & x1 = x1, another instance of the buffer cell.

This is what the group action concept is all about, and behyond :-)

HOW DOES IT WORK?

Each permutation of $S_n$ is applied to each n-input Boolean function f of $X_n$, computing a new n-input Boolean function g. These computations are gathered into chunks and run in parallel on a number of cores.

Each pair $\lbrace f, g \rbrace$ forms an edge of a graph that is latter analyzed. The orbits we are looking for are the connected parts of the obtained graph.

From this analysis, one can get the number of orbits, a list of representatives, and the detailed contents of each orbit. Boolean functions are represented by their signatures as non negative integers.

The results are printed out to the screen or stored into a json file named data.json. Binary data is considered as Big Endian throughout the code.

INSTALL

Run pip install group_action.

The application named orbits is installed automatically under $HOME/.local/bin under Ubuntu 22.04.

Make sure your path is updated with $HOME/.local/bin. Check the following link for more information.

You are ready to go :-)

USAGE

orbits [-h] [--version] [--n N] [--c C] [--r] [--v] [--j]

Brut force computation of orbits of n-input 1-output Boolean functions under the action of the symmetric group Sn.

options:
  -h, --help  show this help message and exit
  --version   show program's version number and exit
  --n N       Number of inputs
  --c C       Number of cores
  --v         Output every element of each orbit
  --j         Output data.json file

EXAMPLE

After installation, run orbits --n 3 --c 12 in order to run on 12 cores and to get the number of orbits and a representative of each orbit as an integer signature for 3-input, 1-output Boolean functions.

TEST

n=0      2 orbits
n=1      4 orbits
n=2     12 orbits
n=3     80 orbits
n=4  3 984 orbits

KNOWN BUGS AND LIMITATIONS

  1. The $n=5$ step requires a lot of memory. Let me know if you go through :-)
  2. The group action is concrete and set up to $G=S_n$ and $X=B^{B^n}$ in this version.
  3. The packaging is managed through PyPI, not yet synchronized to GitHub.

FEEDBACK

Any comment and/or improvement whether on optimization, packaging, documentation, or on any other appropriate topic is welcome :-)

CONTACT

You can reach me antoine AT sirianni DOT ai. I'll do my best to provide you with support.

DOCUMENTATION

Check OR Conf 2024 paper titled "Open Source Standard Cell Library Design" by Antoine Sirianni once published. Type pydoc -w group_action.library to generate the HTML documentation of the library module.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

group_action-0.1.10.tar.gz (7.9 kB view details)

Uploaded Source

Built Distribution

group_action-0.1.10-py3-none-any.whl (9.3 kB view details)

Uploaded Python 3

File details

Details for the file group_action-0.1.10.tar.gz.

File metadata

  • Download URL: group_action-0.1.10.tar.gz
  • Upload date:
  • Size: 7.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.1.1 CPython/3.10.12

File hashes

Hashes for group_action-0.1.10.tar.gz
Algorithm Hash digest
SHA256 56481f2937703b75d34b67fe7945ceaca47acaede1e8617d8c3f9d740343be66
MD5 0eedf61e497422d85ccf6492cc4db79a
BLAKE2b-256 834d78db24303a0207447df0796c2a81df195edbbaf78bcd5ab373c7b97e97eb

See more details on using hashes here.

File details

Details for the file group_action-0.1.10-py3-none-any.whl.

File metadata

File hashes

Hashes for group_action-0.1.10-py3-none-any.whl
Algorithm Hash digest
SHA256 e7bdf76a8dd89f99d1ef533080f68136355e70d398a5ba553e05369e3f788797
MD5 c0eb3736dd0a3f119e56212f6141fc1d
BLAKE2b-256 bc879eff3d66cd59bdda0415ba60605f2b9489cea75b1047cc2f531b9f926a3b

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page