The Python Simulations of Chemistry Framework: 10 years of an open-source quantum chemistry project

Qiming Sun, Matthew R Hermes, Xiaojie Wu, Huanchen Zhai, Xing Zhang, Abdelrahman M. Ahmed, Juan José Aucar, Oliver J. Backhouse, Samragni Banerjee, Peng Bao, Nikolay A. Bogdanov, Kyle Bystrom, Frédéric Chapoton, Ning-Yuan Chen, Ivan Yu. Chernyshov, Helen S. Clifford, Sander Cohen-Janes, Zhi-Hao Cui, Nike Dattani, Linus Bjarne Dittmer, Sebastian Ehlert, Janus Juul Eriksen, Francesco A. Evangelista, Simon A. Ewing, Ardavan Farahvash, Kevin Focke, Yang Gao, Kevin E. Gasperich, Nathan Gillispie, Jonas Greiner, Matthew R. Hennefarth, Jan Hermann, Christopher Hillenbrand, Joonatan Huhtasalo, Basil Ibrahim, Bhavnesh Jangid, Alireza Nejati Javaremi, Andrew J. Jenkins, Yu Jin, Daniel S. King, Derk Pieter Kooi, Henrik R. Larsson, Bryan Tak Gwong Lau, Seunghoon Lee, Susi Lehtola, Chenghan Li, Hao Li, Jiachen Li, Rui Li, Shuhang Li, Aleksandr O. Lykhin, Nastasia Mauger, Pablo del Mazo-Sevillano, Jonathan Moussa, Kousuke Nakano, Verena A. Neufeld, Linqing Peng, Hung Q. Pham, Peter Pinski, Pavel Pokhilko, Zhichen Pu, Yubing Qian, Stephen Jon Quiton, Wanja T. Schulze, Thais R. Scott, Aniruddha Seal, James E. T. Smith, Kori E. Smyser, Terrence Stahl, Chong Sun, Kevin J. Sung, Egor Trushin, Shiv Upadhyay, Ethan A. Vo, Thijs Vogels, Shirong Wang, Tai Wang, Xiao Wang, Xubo Wang, Yuanheng Wang, Mark Williamson, Junjie Yang, Hong-Zhou Ye, Chia-Nan Yeh, Haiyang Yu, Jincheng Yu, Victor Wen-zhe Yu, Chaoqun Zhang, Dayou Zhang, Zijun Zhao, Zehao Zhou, Andrew J. Zhu, Tianyu Zhu, Timothy C. Berkelbach, Laura Gagliardi, Sandeep Sharma, Alexander Sokolov, Garnet Kin-Lic Chan · 2026

Read The Python Simulations of Chemistry Framework: 10 years of an open-source quantum chemistry project by Qiming Sun, Matthew R Hermes, Xiaojie Wu, Huanchen Zhai, Xing Zhang, Abdelrahman M. Ahmed, Juan José Aucar, Oliver J. Backhouse, Samragni Banerjee, Peng Bao, Nikolay A. Bogdanov, Kyle Bystrom, Frédéric Chapoton, Ning-Yuan Chen, Ivan Yu. Chernyshov, Helen S. Clifford, Sander Cohen-Janes, Zhi-Hao Cui, Nike Dattani, Linus Bjarne Dittmer, Sebastian Ehlert, Janus Juul Eriksen, Francesco A. Evangelista, Simon A. Ewing, Ardavan Farahvash, Kevin Focke, Yang Gao, Kevin E. Gasperich, Nathan Gillispie, Jonas Greiner, Matthew R. Hennefarth, Jan Hermann, Christopher Hillenbrand, Joonatan Huhtasalo, Basil Ibrahim, Bhavnesh Jangid, Alireza Nejati Javaremi, Andrew J. Jenkins, Yu Jin, Daniel S. King, Derk Pieter Kooi, Henrik R. Larsson, Bryan Tak Gwong Lau, Seunghoon Lee, Susi Lehtola, Chenghan Li, Hao Li, Jiachen Li, Rui Li, Shuhang Li, Aleksandr O. Lykhin, Nastasia Mauger, Pablo del Mazo-Sevillano, Jonathan Moussa, Kousuke Nakano, Verena A. Neufeld, Linqing Peng, Hung Q. Pham, Peter Pinski, Pavel Pokhilko, Zhichen Pu, Yubing Qian, Stephen Jon Quiton, Wanja T. Schulze, Thais R. Scott, Aniruddha Seal, James E. T. Smith, Kori E. Smyser, Terrence Stahl, Chong Sun, Kevin J. Sung, Egor Trushin, Shiv Upadhyay, Ethan A. Vo, Thijs Vogels, Shirong Wang, Tai Wang, Xiao Wang, Xubo Wang, Yuanheng Wang, Mark Williamson, Junjie Yang, Hong-Zhou Ye, Chia-Nan Yeh, Haiyang Yu, Jincheng Yu, Victor Wen-zhe Yu, Chaoqun Zhang, Dayou Zhang, Zijun Zhao, Zehao Zhou, Andrew J. Zhu, Tianyu Zhu, Timothy C. Berkelbach, Laura Gagliardi, Sandeep Sharma, Alexander Sokolov, Garnet Kin-Lic Chan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-14155.txt.

Introduction

Sun and co-authors review ten years of PySCF, the Python Simulations of Chemistry Framework. PySCF has grown into a widely used open-source platform for electronic structure theory and new quantum chemistry methods. The article covers progress since the 2020 overview. New modules and methods are described, along with infrastructure changes under the hood. Performance benchmarks show how the code runs today. The huge author list reflects the broad community that builds the project.

Catalog record

CREATORQiming Sun, Matthew R Hermes, Xiaojie Wu, Huanchen Zhai, Xing Zhang, Abdelrahman M. Ahmed, Juan José Aucar, Oliver J. Backhouse, Samragni Banerjee, Peng Bao, Nikolay A. Bogdanov, Kyle Bystrom, Frédéric Chapoton, Ning-Yuan Chen, Ivan Yu. Chernyshov, Helen S. Clifford, Sander Cohen-Janes, Zhi-Hao Cui, Nike Dattani, Linus Bjarne Dittmer, Sebastian Ehlert, Janus Juul Eriksen, Francesco A. Evangelista, Simon A. Ewing, Ardavan Farahvash, Kevin Focke, Yang Gao, Kevin E. Gasperich, Nathan Gillispie, Jonas Greiner, Matthew R. Hennefarth, Jan Hermann, Christopher Hillenbrand, Joonatan Huhtasalo, Basil Ibrahim, Bhavnesh Jangid, Alireza Nejati Javaremi, Andrew J. Jenkins, Yu Jin, Daniel S. King, Derk Pieter Kooi, Henrik R. Larsson, Bryan Tak Gwong Lau, Seunghoon Lee, Susi Lehtola, Chenghan Li, Hao Li, Jiachen Li, Rui Li, Shuhang Li, Aleksandr O. Lykhin, Nastasia Mauger, Pablo del Mazo-Sevillano, Jonathan Moussa, Kousuke Nakano, Verena A. Neufeld, Linqing Peng, Hung Q. Pham, Peter Pinski, Pavel Pokhilko, Zhichen Pu, Yubing Qian, Stephen Jon Quiton, Wanja T. Schulze, Thais R. Scott, Aniruddha Seal, James E. T. Smith, Kori E. Smyser, Terrence Stahl, Chong Sun, Kevin J. Sung, Egor Trushin, Shiv Upadhyay, Ethan A. Vo, Thijs Vogels, Shirong Wang, Tai Wang, Xiao Wang, Xubo Wang, Yuanheng Wang, Mark Williamson, Junjie Yang, Hong-Zhou Ye, Chia-Nan Yeh, Haiyang Yu, Jincheng Yu, Victor Wen-zhe Yu, Chaoqun Zhang, Dayou Zhang, Zijun Zhao, Zehao Zhou, Andrew J. Zhu, Tianyu Zhu, Timothy C. Berkelbach, Laura Gagliardi, Sandeep Sharma, Alexander Sokolov, Garnet Kin-Lic Chan
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE60530
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.14155

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The Python Simulations of Chemistry Framework: 10 years of an open-source quantum chemistry project Qiming Sun, Matthew R Hermes, Xiaojie Wu, Huanchen Zhai, Xing Zhang, Abdelrahman M. Ahmed, Juan José Aucar, Oliver J. Backhouse, Samragni Banerjee, Peng Bao, Nikolay A. Bogdanov, Kyle Bystrom, Frédéric Chapoton, Ning-Yuan Chen, Ivan Yu. Chernyshov, Helen S. Clifford, Sander Cohen-Janes, Zhi-Hao Cui, Nike Dattani, Linus Bjarne Dittmer, Sebastian Ehlert, Janus Juul Eriksen, Francesco A. Evangelista, Simon A. Ewing, Ardavan Farahvash, Kevin Focke, Yang Gao, Kevin E. Gasperich, Nathan Gillispie, Jonas Greiner, Matthew R. Hennefarth, Jan Hermann, Christopher Hillenbrand, Joonatan Huhtasalo, Basil Ibrahim, Bhavnesh Jangid, Alireza Nejati Javaremi, Andrew J. Jenkins, Yu Jin, Daniel S. King, Derk Pieter Kooi, Henrik R. Larsson, Bryan Tak Gwong Lau, Seunghoon Lee, Susi Lehtola, Chenghan Li, Hao Li, Jiachen Li, Rui Li, Shuhang Li, Aleksandr O. Lykhin, Nastasia Mauger, Pablo del Mazo-Sevillano, Jonathan Moussa, Kousuke Nakano, Verena A. Neufeld, Linqing Peng, Hung Q. Pham, Peter Pinski, Pavel Pokhilko, Zhichen Pu, Yubing Qian, Stephen Jon Quiton, Wanja T. Schulze, Thais R. Scott, Aniruddha Seal, James E. T. Smith, Kori E. Smyser, Terrence Stahl, Chong Sun, Kevin J. Sung, Egor Trushin, Shiv Upadhyay, Ethan A. Vo, Thijs Vogels, Shirong Wang, Tai Wang, Xiao Wang, Xubo Wang, Yuanheng Wang, Mark Williamson, Junjie Yang, Hong-Zhou Ye, Chia-Nan Yeh, Haiyang Yu, Jincheng Yu, Victor Wen-zhe Yu, Chaoqun Zhang, Dayou Zhang, Zijun Zhao, Zehao Zhou, Andrew J. Zhu, Tianyu Zhu, Timothy C. Berkelbach, Laura Gagliardi, Sandeep Sharma, Alexander Sokolov, Garnet Kin-Lic Chan Abstract Over the past decade, the Python-based Simulations of Chemistry Framework (PySCF) has developed into a widely used open-source platform for electronic-structure theory and quantum chemical method development. This article reviews the major advances since the previous overview in 2020, covering new modules and methodology, infrastructure changes, and performance benchmarks. I Introduction The Python based Simulations of Chemistry Framework ( PySCF ) Sun et al. 2018 ; Sun et al. 2020 is a widely used quantum chemistry library. Initially founded as an internal project within the Chan group, the first stable public release (version 1.0) was in 2015, and the version at the time of writing (version 2.12) marks ten years of this open-source project. Currently, the PySCF repositories encompass more than 500,000 lines of code, with more than 1,000 dependent projects on GitHub, and the package receives more than 1,000,000 downloads a year, making it among the most widely used development frameworks in quantum chemistry. The original intention behind PySCF was to create a framework that supported not only the traditional quantum chemistry paradigm where the user supplies an input file and reads the calculation output, but also the development of new quantum chemistry capabilities. To this end, inspired by the ubiquitous NumPy Harris et al. 2020 and SciPy Virtanen et al. 2020 libraries used in numerical computation in Python, PySCF is structured as a set of loosely coupled modules with small, largely pure, functions that implement a rich set of APIs. The implementation is mainly in Python with performance hotspots rewritten in C. In addition, in part due to the initial applications that motivated the development of PySCF (as recounted in Ref. 2 ), the code is designed to support molecular, materials, and model Hamiltonians on a (relatively) equal footing, at both the mean-field and many-body level. The flexibility of this design has contributed to PySCF being adopted in areas outside of the chemistry community: materials science, condensed matter physics, machine learning, and quantum information science. The usage of PySCF has thus moved far beyond the interests and expertise of the original developers. Fortunately, to a large degree, the original design choices have held through this expansion. As we look to the next decade of this project, we see user demands for new kinds of calculations (e.g. high throughput for dataset generation); simulations of ever larger systems; and the emergence of new computing hardware, and programming libraries and paradigms, and with it, a growing complexity of management. In this work, we document the main developments in PySCF since the last publication (at the time of version 1.7) Sun et al. 2020 that aim to address some of these emerging challenges, and provide some outlook on the future. II Main developments II.1 Code organization Since Ref. 2 , the codebase has greatly expanded. In addition to new methods (and new implementations of existing methods), significant effort has been made to optimize for hardware (especially GPUs) and to encompass new development paradigms, for example, auto-differentiation and just-in-time (JIT) compilation frameworks. Because of this expansion, PySCF is now separated into the core pyscf repository, which provides a compact set of modules that can be expected to be used in the widest range of applications; the pyscf-forge repository, which provides a staging ground for experimental features that may later be transitioned to the core pyscf repository; as well as companion repositories such as GPU4PySCF , PySCFAD , and MPI4PySCF . Finally, PySCF has a native mechanism to support custom extensions, and a list of extensions developed by the community is maintained on the PySCF website, including links to their independent repositories. Unlike those in pyscf and pyscf-forge , the features available as extensions are not officially supported by the developers. II.2 Periodic electronic structure infrastructure A key design principle of PySCF is that it should support both molecular and materials [periodic boundary condition (PBC)] calculations on a relatively equal footing. To this end, PySCF implements multiple mathematical formalisms to support periodic electronic structure calculations (mean-field and beyond) in Gaussian basis sets. The different formalisms are primarily based on different decompositions of the periodic electron-electron repulsion integrals via the density-fitting (DF) approximation, using either an auxiliary basis of plane waves (FFTDF) or atom-centered Gaussians (GDF). As these have different performance characteristics, users can select the optimal choice depending on their application. Since Ref. 2 , multiple enhancements have been made to the FFTDF and GDF infrastructure in PySCF to improve performance and accuracy. The improvements to the FFTDF module now make it suitable for use in a wide range of applications so long as pseudopotentials are acceptable. For example, the multigrid FFTDF density functional theory (DFT) implementation now achieves state-of-the-art performance on both CPUs and GPUs (see Table 1 ), and the GPU parallelization can saturate the compute of modern NVIDIA A100 and H100 GPUs. Using multigrid FFTDF, DFT iterations on systems with as many as 20000 basis functions now take only a few seconds on a single GPU. Similarly, using the occ-RI and interpolative separable density fitting (ISDF) techniques, Sharma, White, and Beylkin 2022 ; Yang et al. 2026 Hartree-Fock (HF) and hybrid DFT calculations can also now be carried out with good performance using FFTDF on both CPUs and GPUs (see Fig. 1 ). Table 1: Wall times (in seconds) for one self-consistent field (SCF) iteration and nuclear gradient in multigrid FFTDF restricted KS-DFT calculations for water clusters, using the PBE Perdew, Burke, and Ernzerhof 1997 ; Lehtola et al. 2018 density functional, GTH-TZV2P VandeVondele and Hutter 2007 basis set and GTH-PADE Hartwigsen, Gœdecker, and Hutter 1998 pseudopotential. A plane wave cutoff energy of 140 a.u. was used. The calculations were performed on NVIDIA A100 (80 GB) and H100 (80 GB) GPUs, as well as

Read the complete text