What's Changed

• Fix issues with overflows by @nbgl in https://github.com/data61/python-paillier/pull/73
• Default keysize increased to 3072 by @hardbyte in https://github.com/data61/python-paillier/pull/66
• Using gmpy2's mulmod to accelerate the mulmod operation by @tanjuntao in https://github.com/data61/python-paillier/pull/99

Updates to examples

• federated learning example by @wilko77 in https://github.com/data61/python-paillier/pull/64
• make learning rate independent of dataset size by @wilko77 in https://github.com/data61/python-paillier/pull/69

New Contributors

• @tanjuntao made their first contribution in https://github.com/data61/python-paillier/pull/99

Full Changelog: https://github.com/data61/python-paillier/compare/1.4.0...1.5.0

Version 1.4.0 (2018-04-19)

Complete pure Python fallback implementation.

Features

• invert now available without gmpy2, implemented using the extended Euclidean algorithm (extended_euclidean_algorithm)
• getprimeover now available without gmpy2, along with a probabilitic primality test isprime based on the Miller-Rabin test (miller_rabin)

Many thanks to Quentin Santos for this contribution.

Version 1.3.0 (2017-02-08)

Changes to enhance performance. Using Chinese Remainder Theorem for faster decryption. Exploit property of the generator to speed up encryption.

Note both the api and the serialisation has changed.

• A private key now has a p and q attribute instead of lambda and mu. To continue being able to create a private key using the totient use the PaillierPrivateKey.from_totient static constructor.
• The serialization and constructor of a public key now only requires n.