spinterp

Sparse Grid Interpolation Toolbox for Python

Sparse Grid Interpolation Toolbox for Python.

Grid types

Grid type	Basis	Max depth
Clenshaw-Curtis	Piecewise linear	8
Chebyshev	Polynomial (DCT)	10
Gauss-Patterson	Nested Gaussian	6
Maximum	Piecewise linear	8
NoBoundary	Piecewise linear	8

Routines

Python binding	Description
spvals	Compute hierarchical surpluses
spinterp	Evaluate interpolant (and gradient) at points
spquad	Integrate interpolant over domain
spgrid	Return sparse grid node coordinates
spgetseq	Generate multi-index level sequences

Quick example

import spinterp

z = spinterp.spvals(lambda x, y, t: x**2 + y**2 - 2 * t, d=3)
f = spinterp.spinterp(z, 0.5, 0.2, 0.2)
q = spinterp.spquad(z)

Installation

pip install spinterp

Requires Python 3.10+ and NumPy.

Build from source

python bin/build.py install   # build and install via uv
python bin/build.py wheel     # produce a wheel in dist/
python bin/build.py clean     # remove all build artifacts

Requires gfortran, Meson, and f2py (numpy).

Documentation

• Theory — hierarchical basis, Smolyak construction, algorithms
• Quickstart — runnable examples
• API Reference — function signatures and arguments

License

GNU General Public License v3 (GPLv3) — see LICENSE.