https://github.com/snowkylin/npn

A boolean matcher that computes the NPN canonical representative for a given boolean function.
https://github.com/snowkylin/npn

boolean-matcher cpp eda logic-synthesis npn pypi-package python

Last synced: about 1 month ago
JSON representation

A boolean matcher that computes the NPN canonical representative for a given boolean function.

• Host: GitHub
• URL: https://github.com/snowkylin/npn
• Owner: snowkylin
• Created: 2023-07-01T13:33:22.000Z (over 1 year ago)
• Default Branch: master
• Last Pushed: 2024-03-24T02:38:19.000Z (9 months ago)
• Last Synced: 2024-04-24T08:16:31.823Z (8 months ago)
• Topics: boolean-matcher, cpp, eda, logic-synthesis, npn, pypi-package, python
• Language: C++
• Homepage:
• Size: 20.5 KB
• Stars: 0
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# NPN: Compute canonical representatives for boolean functions

This package is a boolean matcher. It simply computes the canonical representative, which is unique for each NPN equivalence class, for a given boolean function represented by truth table.

The backend of this package is implemented by C++ for efficiency.

Currently, this package supports boolean functions with number of inputs <= 8.

## Install

```bash

pip install npn

```

## Usage

```python

import npn

# the truth table of f(x_2, x_1, x_0) as [f(0, 0, 0), f(0, 0, 1), f(0, 1, 0), f(0, 1, 1), ...]

tt = [True, True, True, False, True, True, True, True]

c = npn.npn_canonical_representative(tt)    # [True, True, True, True, True, True, True, False]

c_int = npn.tt_to_int(c)                    # 254 (11111110)

c, phase, perm, output_inv = npn.npn_canonical_representative(tt, return_details=True, refinement=True)

```

- Set `return_details` parameter to `True` to return the NPN transformation information (`phase, perm, output_inv`).

For example, here `phase = [False, False, True]`, `perm = [0, 1, 2]` and `output_inv = False` 

means `(x_0, x_1, x_2)` should be mapped to `(x_0, x_1, x_2)`, then the third variable (`x_2`) should be inverted

and the final result should not be inverted.

- Set `refinement` parameter to `True` to accelerate computation while breaking the lexicographical order of truth tables.

See "Building a Better Boolean Matcher and Symmetry Detector" for details.

## Build

Compile the shared library `libnpn.dll` and `libnpn.so` via CMake on Windows and Linux environment, place them in the `\npn` folder, then

```bash

python setup.py sdist

```

To upload to PyPI, run

```bash

twine upload .\dist\npn-X.X.tar.gz

```

## Test

```bash

pip install pytest  # if you haven't install it

python -m pytest

```

## References

1. Chai, D., and A. Kuehlmann. “Building a Better Boolean Matcher and Symmetry Detector.” In Proceedings of the Design Automation & Test in Europe Conference, 1:1–6, 2006. https://doi.org/10.1109/DATE.2006.243959.

2. Debnath, D., and T. Sasao. “Efficient Computation of Canonical Form for Boolean Matching in Large Libraries.” In ASP-DAC 2004: Asia and South Pacific Design Automation Conference 2004 (IEEE Cat. No.04EX753), 591–96, 2004. https://doi.org/10.1109/ASPDAC.2004.1337660.

3. Hinsberger, U., and R. Kolla. “Boolean Matching for Large Libraries.” In Proceedings 1998 Design and Automation Conference. 35th DAC. (Cat. No.98CH36175), 206–11, 1998. https://doi.org/10.1145/277044.277100.