The following is an alphabetically sorted list of free and open-source sparse grid software.
DisCoTec
DisCoTec is a framework for the massively parallel solution of higher-dimensional PDEs with the Sparse Grid Combination Technique. While it originated as a module in SG⁺⁺ for the parallel simulation of hot fusion plasmas, its optimization for high-performance computing (HPC) systems required a complete redesign. It consequently grew into its own project. In 2024, it was the world’s first sparse grid code to be able to compute a high-dimensional, time-dependent PDE on multiple Tier-0 supercomputers at the same time!
Link: https://github.com/SGpp/DisCoTec
SG⁺⁺
SG⁺⁺ is a universal sparse grid toolbox that focuses on spatial adaptivity. Modules are available for function interpolation, optimization, data mining, machine learning, uncertainty quantification, and partial differential equations. SG⁺⁺ is written in C++ and can be used in C++, Python, Java, and MATLAB. The toolbox is maintained by the group of Dirk Pflüger at the University of Stuttgart, Germany.
Link: sgpp.sparsegrids.org
sparseSpACE
This python project implements different variants of the spatially adaptive Combination Technique. It was first targeted to solve high dimensional numerical integration with the spatially adaptive Combination Technique but it now supports the implementation of arbitrary grid operations. It supports already numerical integration, interpolation, Uncertainty Quantification, Sparse Grid Density Estimation (with classification and clustering), regression, and PDE calculations.
Link: https://github.com/obersteiner/sparseSpACE
Sparse Grids Matlab Kit
The Sparse Grids Matlab Kit is a collection of high-level MATLAB functions for interpolation and quadrature with the sparse grid combination technique. The general focus is on teaching and uncertainty quantification. The visualization functions of the MATLAB code make it well-suited for rapid prototyping. It provides a user friendly MATLAB interface for the CT, and also dimensional adaptivity for nested and non-nested sequences of component grid collocation points. The kit is mainly developed by Lorenzo Tamellini from CNR-IMATI, Italy.
Link: sites.google.com/view/sparse-grids-kit
spinterp
spinterp is a MATLAB library for interpolation on sparse grids using Smoylak’s construction. In addition, the library supports optimization and quadrature. Different one-dimensional grid point rules are implemented. spinterp was developed from 2006 to 2008 by Andreas Klimke from the University of Stuttgart, Germany.
Link: people.math.sc.edu/Burkardt/m_src/spinterp/spinterp.html
Tasmanian
The Toolkit for Adaptive Stochastic Modeling and Non-Intrusive ApproximatioN (Tasmanian). Tasmanian implements are variety of different grid types, e.g., spatially adaptive grids with local or global polynomials, wavelet grids, and Fourier grids. Tasmanian is developed by the group of Clayton Webster and Miroslav Stoyanov from the Oak Ridge National Laboratory (ORNL), Tennessee, USA.
Link: tasmanian.ornl.gov
Other Software
If some other sparse grid software should be listed here, please contact us.