Supported environments

The 3.2 series requires Python >=3.12,<4 without restricting future Python 3 minor releases, and NumPy >=2.2,<2.5, subject to its own Python requirements. The Python and NumPy minimums follow the three-year and two-year support windows, respectively, recommended by SPEC 0. CI covers NumPy 2.x on Python 3.12–3.14, both with and without the optional Numba extra (Numba >=0.64,<1). Python 3.8–3.11 users should retain the 3.1 series.

Real evaluation uses a C compiler (gcc or cc) by default. Install multivar_horner[numba] for accelerated recipe evaluation when a compiler is unavailable. Complex evaluation always uses the recipe backend. Compiled evaluator caches are platform-specific, so the same installation can be used safely on ARM64 and x86_64 systems without reusing incompatible binaries.

Quick Guide:

pip install multivar_horner

For efficiency this package is compiling the instructions required for polynomial evaluation to C by default. If you don’t have a C compiler (gcc or cc) installed you also need to install numba for using an alternative method:

pip install multivar_horner[numba]

Simple example:

import numpy as np
from multivar_horner import HornerMultivarPolynomial

# input parameters defining the polynomial
#   p(x) = 5.0 + 1.0 x_1^3 x_2^1 + 2.0 x_1^2 x_3^1 + 3.0 x_1^1 x_2^1 x_3^1
coefficients = np.array([[5.0], [1.0], [2.0], [3.0]], dtype=np.float64)
exponents = np.array([[0, 0, 0], [3, 1, 0], [2, 0, 1], [1, 1, 1]], dtype=np.uint32)

# [#ops=7] p(x) = x_1 (x_1 (x_1 (1.0 x_2) + 2.0 x_3) + 3.0 x_2 x_3) + 5.0
horner_polynomial = HornerMultivarPolynomial(coefficients, exponents)
x = np.array([-2.0, 3.0, 1.0], dtype=np.float64)
p_x = horner_polynomial(x)

Also see:

• Documentation

• GitHub

• PyPI

• paper (JOSS)

• paper (arXiv)