PolyHE

Simple & easy Homomorphic Encryption

• Supports Python's operators & pickle
• Supports arbitrary typed, shaped & sized arrays

Example

import polyhe
ctx = polyhe.new()

# Numbers
p1 = 1
p2 = -1
c1 = ctx.encrypt(p1)
c2 = ctx.encrypt(p2)
c3 = c1 * c2 + p1
p3 = ctx.decrypt(c3)
# 0

# Lists
p1 = [0.3, 0.4]
p2 = [0.4, 0.3]
c1 = ctx.encrypt(p1)
c2 = ctx.encrypt(p2)
c3 = -c1 + c2
p3 = ctx.decrypt(c3)
# [0.1, -0.1]

# NumPy
import numpy as np
p1 = np.full((2, 3, 4), 0.4, np.float64)
p2 = np.full((2, 3, 4), 0.3, np.float64)
c1 = ctx.encrypt(p1)
c2 = ctx.encrypt(p2)
c3 = c1 - c2
p3 = ctx.decrypt(c3)
# np.full((2, 3, 4), 0.1, np.float64)

Install

Production

pip install polyhe

Development

git clone https://github.com/hpca-uji/polyhe.git
cd polyhe
pip install -e .

Documentation

Constants

• Backend:

  Encryption backend

  • OPENFHE (source)
  • TENSEAL (source)
  • UARCHFHE

Structures

• Options(...)

  Encryption options

  • slots: int = 13

    2 ** slots elements per backend ciphertext

  • scale: int = 40

    2 ** scale typical operational scale

  • security: int = 128

    Security level

    Typical values: 128, 192 and 256.

Functions

• new(backend, options)

  Create a communicator.

  • backend: Backend = Backend.OPENFHE
  • options: Options = Options()

Classes

• core.Context(options)

  Encryption context implementation

  • options: Options = Options()

  ---

  • encrypt(data) -> Ciphertext

    Encrypt data to ciphertext

  • decrypt(obj: Ciphertext)

    Decrypt cypertext to data

• core.Ciphertext(...)

  Ciphertext has -, + and * operator support, with either another ciphertext or plain data.

  Note: Support for operation may vary between backends

• {backend}.Context(options)

  Concrete encryption context implementation for the given backend

Notes

PolyHE prioritizes simplicity and ease of use over advanced features. It is designed for prototyping and experimentation purposes rather than production-grade deployments. If you require fine-grained control over encryption parameters, performance tuning, or advanced schemes, we recommend using the underlying backend libraries directly. Alternatively, Pyfhel (source) is a mature Python library that provides more comprehensive access to homomorphic encryption features.

Acknowledgments

The library has been partially supported by:

• Project PID2023-146569NB-C22 "Inteligencia sostenible en el Borde-UJI" funded by the Spanish Ministry of Science, Innovation and Universities.
• Project C121/23 Convenio "CIBERseguridad post-Cuántica para el Aprendizaje FEderado en procesadores de bajo consumo y aceleradores (CIBER-CAFE)" funded by the Spanish National Cybersecurity Institute (INCIBE).