For physicists tackling the complexities of quantum mechanics, simulating even modestly sized systems can be computationally prohibitive. Now, a powerful open-source library called XDiag has received a significant upgrade, streamlining these calculations. The release of version 0.4.0 introduces crucial support for sparse matrices – a game-changer for handling the vast, mostly-empty datasets inherent in many-body quantum systems – alongside performance enhancements in its Julia interface. This update not only accelerates existing research but also unlocks the potential to explore larger, more realistic models, bringing us closer to understanding the behavior of complex quantum materials.
XDiag Library: Overview and Features
XDiag is a powerful library designed for performing Exact Diagonalizations – a crucial technique for studying quantum many-body systems. Currently at version v0.4.0, released November 4, 2025—which introduces sparse matrix capabilities—XDiag distinguishes itself through optimized combinatorical algorithms for efficient Hilbert space navigation and iterative linear algebra solvers. The library is structured around two core packages: a high-performance C++ library named ‘xdiag’ and a user-friendly Julia wrapper, ‘XDiag.jl’, offering convenient access to its functionality. Notably, XDiag supports both shared and distributed memory parallelization, enabling calculations on larger systems. Researchers utilizing XDiag are encouraged to cite the software paper available on arXiv (arXiv:2505.02901) and a related publication in Phys. Rev. E (Wietek, 2018) if employing sublattice coding or distributed memory features.
Release Information and Citation Details
Currently at version 0.4.0 as of November 4, 2025, XDiag is a library designed for performing Exact Diagonalizations of quantum many-body systems, comprised of a core C++ library and a convenient Julia wrapper, XDiag.jl. The latest release introduces sparse matrix capabilities, building on the enhancements of version 0.3.3 which focused on parallelization within Julia and compatibility with Julia 1.12. To ensure proper attribution and support ongoing development, researchers utilizing XDiag or its implemented algorithms in published work are strongly encouraged to cite the software paper, currently available as an arXiv preprint (arXiv:2505.02901). Furthermore, those employing the sublattice coding techniques or distributed memory parallelization features should also cite Wietek & Lauchli (2018) published in Phys. Rev. E.
Source: https://awietek.github.io/xdiag/
