EUVIMEDOpdag EUVIMED
Crossref

A scalable diagonalization framework for tensor-product bitstring selected configuration interaction

Enhua Xu iD, William Dawson iD, Himadri Pathak iD, Takahito Nakajima iD

The Journal of Chemical Physics · 2026-05-21 · 164 (19)

DOI: 10.1063/5.0332736 ↗

At a glance

Journal article in The Journal of Chemical Physics, published online on 2026-05-15. Written by Enhua Xu and 3 co-authors. Institution: RIKEN Center for Computational Science, Kobe, Japan. Cited 1 times so far (Crossref); 41 references.

AbstractBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Selected configuration interaction (SCI) methods are effective for treating strongly correlated electronic systems, yet their scalability has long been limited by implementations that replicate the configuration interaction (CI) vector across processes, leading to severe memory bottlenecks. Here, we present a fully distributed diagonalization framework tailored for extremely large selected determinant spaces, directly addressing this major scalability bottleneck of modern SCI methods. The method is grounded in a tensor-product bitstring (TPB) representation, in which determinants are organized through a TPB structure constructed from selected α- and β-bitstrings, and is referred to as tensor-product bitstring SCI (TBSCI). An efficient TBSCI eigensolver is developed based on a novel bitstring-based Hamiltonian evaluation algorithm together with a suite of MPI communication strategies designed to improve parallel efficiency. Large-scale full configuration interaction (FCI) benchmarks, employed as communication-intensive stress tests, demonstrate that the implemented TBSCI eigensolver continues to reduce the wall time for distributed diagonalization of 2.6 × 1012 determinants, reaching 54 000 nodes (more than 2.5 × 106 cores) on supercomputer Fugaku. Beyond scalability, we investigate the structural compactness of the TPB representation and show that selecting α- and β-bitstrings according to their collective weights in a reference SCI wavefunction yields TPB-based wavefunctions approaching the FCI limit while using only a small fraction of determinants. These results establish TBSCI as a scalable SCI methodology and provide evidence for the intrinsic compactness of the TPB representation.

Authors and institutions

  • RIKEN Center for Computational Science, Kobe, Japan.
  • RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences, Wako, Japan.

ORCID profiles: Enhua Xu, William Dawson, Himadri Pathak, Takahito Nakajima

Cite

APA

Xu, E., Dawson, W., Pathak, H., & Nakajima, T. (2026). A scalable diagonalization framework for tensor-product bitstring selected configuration interaction. The Journal of Chemical Physics, 164(19). https://doi.org/10.1063/5.0332736

Vancouver

Xu E, Dawson W, Pathak H, Nakajima T. A scalable diagonalization framework for tensor-product bitstring selected configuration interaction. The Journal of Chemical Physics. 2026;164(19). doi:10.1063/5.0332736

BibTeX
@article{Xu2026,
  author = {Xu, Enhua and Dawson, William and Pathak, Himadri and Nakajima, Takahito},
  title = {A scalable diagonalization framework for tensor-product bitstring selected configuration interaction},
  journal = {The Journal of Chemical Physics},
  year = {2026},
  volume = {164},
  number = {19},
  doi = {10.1063/5.0332736}
}
Start for free →