v0.0.22

Released December 3, 2021

Breaking changes

  • Random integer generation is handled using new style random generators. Now each .Random() call will generate a new seed rather than using the NumPy “global” seed used with np.random.randint().

  • Add a seed=None keyword argument to FieldArray.Random() and Poly.Random(). A reproducible script can be constructed like this:

    rng = np.random.default_rng(123456789)
    x = GF.Random(10, seed=rng)
    y = GF.Random(10, seed=rng)
    poly = galois.Poly.Random(5, seed=rng, field=GF)
    

Changes

  • Official support for Python 3.9.

  • Major performance improvements to “large” finite fields (those with dtype=np.object_).

  • Minor performance improvements to all finite fields.

  • Add the Number Theoretic Transform (NTT) in ntt() and intt().

  • Add the trace of finite field elements in FieldArray.field_trace().

  • Add the norm of finite field elements in FieldArray.field_norm().

  • Support len() on Poly objects, which returns the length of the coefficient array (polynomial order + 1).

  • Support x.dot(y) syntax for the expression np.dot(x, y).

  • Minimum NumPy version bumped to 1.18.4 for new style random usage.

  • Various bug fixes.

Contributors


Last update: May 18, 2022