Pith. sign in

REVIEW 2 major objections 7 minor 48 references

A distributed-memory implementation of the quasi-four-component relativistic DFT method extends self-consistent four-component simulations to periodic supercells with 3,383 atoms and 216,628 basis functions, with near-ideal scaling from 336

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

A parallel quasi-four-component relativistic DFT implementation reaches periodic systems of 3,383 atoms with near-ideal scaling between 336 and 672 CPU cores.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A real distributed-memory Q4C implementation with credible scaling to ~3,400 atoms, but the distributed matrix assembly needs an explicit small-system correctness check before I'd trust the physics. the 2 major comments →

arxiv 2608.00773 v1 pith:2BHQFXKC submitted 2026-08-01 cond-mat.mtrl-sci

A Large-scale Parallel Implementation of Quasi-Four-Component Relativistic Density Functional Theory with Numeric Atom-centered Orbitals

Wentao Zhang , Rundong Zhao , Volker Blum This is my paper
classification cond-mat.mtrl-sci PACS 71.15.-m31.15.-p
keywords quasi-four-component relativistic DFTfour-component Dirac-Kohn-Shamnumeric atom-centered orbitalsspin-orbit couplingdomain decompositionparallel scalingperiodic solidsdefect supercells
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Fully relativistic four-component DFT normally stays confined to small molecules because of steep memory and cost growth. This paper claims that the quasi-four-component (Q4C) method, where the small-component wavefunction is tied to the large component through free-atom radial functions, can be parallelized for periodic solids at previously inaccessible scale. The enabler is a locally-indexed real-space domain decomposition: integration grids are split among MPI tasks, each task evaluates only the partial integrals from its own grid batches, and the results are transformed to spinor form and redistributed directly into a block-cyclic layout, so the full Hamiltonian is never stored on one task. Tested up to 3,383 atoms per unit cell (216,628 basis functions) with near-ideal scaling between 336 and 672 CPU cores, the implementation yields fully self-consistent, spin-orbit-coupled band structures for heavy-element defect supercells. If correct, it lets production-scale defect calculations in lead-halide perovskites and similar materials be done at a genuinely four-component relativistic level.

Core claim

On its own terms, the paper shows that quasi-four-component (Q4C) relativistic DFT can be implemented with order-N real-space integration in distributed memory. Spinors are expanded in atom-centered basis functions whose small component is tied to the large component by a free-atom kinetic-balance operator, so only positive-energy states enter. Scalar integrals are evaluated on partitioned grid batches; each MPI task stores only its locally nonzero block, transforms it to spinor form with Clebsch-Gordan coefficients, and sums contributions with Bloch phases into a block-cyclic layout, never assembling the full matrix. Demonstration: band structures of Bi-doped lead-iodide perovskite supercel

What carries the argument

The central object is the quasi-four-component (Q4C) spinor basis: each four-component spinor is a linear combination of atom-centered spinors whose small component is generated from the large component by a fixed free-atom operator (atomic balance), so the basis contains separate radial functions for j=l+1/2 and j=l-1/2 channels from the start. The carrying mechanism is locally-indexed real-space domain decomposition: overlapping atom-centered integration grids are partitioned into per-task batches; each MPI task evaluates scalar-format integrals only over basis functions with nonzero support in its batches, transforms them to spinor form using Clebsch-Gordan coefficient matrices, and then

Load-bearing premise

The implementation assumes that the bookkeeping which maps local grid-batch contributions to their places in the global matrix captures every contribution exactly once; since the full matrix is never assembled for a cross-check, a dropped or double-counted batch would silently corrupt the band structure.

What would settle it

Run the same small periodic system (for example the 40-atom CsPbBr3 cell at one k-point) through the distributed path and through a single-task, full-matrix assembly, and compare the Hamiltonian and overlap matrix elements or the resulting eigenvalues to tight numerical tolerance. Any discrepancy beyond integration tolerance would show that the locally-indexed decomposition is incomplete; agreement would validate the phase and index bookkeeping.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Defect-scale supercells of heavy-element semiconductors can be treated at a fully relativistic, self-consistent level, including spin-orbit coupling from the start, rather than only as a post-processing correction.
  • The two-step scalar-relativistic plus non-self-consistent SOC route is shown to be incomplete for such systems: without separate p1/2 radial functions it can misplace valence bands and defect states relative to Q4C.
  • Real-space integration and density update scale roughly linearly with system size, while the dense eigenproblem dominates; the near-ideal 336-to-672-core speedup suggests the implementation is not yet at its scaling limit.
  • The Q4C calculation needs four times as many scalar basis functions as scalar relativity and raises eigenproblem time by roughly 18-24x, identifying the eigensolver as the critical bottleneck for even larger cells.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the order-N integration claim holds, the practical ceiling for fully relativistic periodic DFT is set by the dense eigensolver, so iterative or localized eigensolvers could push the demonstrated few-thousand-atom limit substantially higher.
  • The same locally-indexed decomposition and block-cyclic assembly pattern could in principle be lifted into other local-basis relativistic codes, since it is formulated in terms of scalar partial integrals and Clebsch-Gordan transformations rather than code-specific data structures.
  • The Q4C band differences for the doped perovskites are specific enough to be tested: optical or transport measurements sensitive to the Bi-derived defect level, or a reference four-component calculation on a smaller cell, would indicate whether the extra gap reduction and level shifts are quantitatively right.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 7 minor

Summary. The paper describes a distributed-memory parallel implementation of quasi-four-component (Q4C) relativistic DFT in FHI-aims using numeric atom-centered orbitals. The authors use a locally-indexed real-space domain decomposition for the Hamiltonian, overlap, and density-update integrations, transform scalar matrix elements to the spinor basis, redistribute them into a block-cyclic layout, and solve the generalized eigenvalue problem with ELPA via ELSI. They benchmark the code on CsPbBr3 supercells and on Bi-doped (PEA)2(Pb1-xBix)I4 supercells up to 3,383 atoms (216,628 scalar basis functions), reporting near-ideal strong scaling between 336 and 672 CPU cores and reduced per-task memory with more nodes. They also present SR, SR+SOC, and Q4C band structures for the 1,504- and 3,383-atom perovskite supercells. The central claim is that the implementation extends fully relativistic DFT to several thousand atoms per unit cell while retaining near-linear real-space integration and efficient distributed eigensolution.

Significance. If the implementation is correct, this is a substantial advance for relativistic DFT: it moves four-component-type calculations from the ~100-atom scale to the few-thousand-atom scale for periodic systems, using a well-described algorithmic framework and publicly deposited data. The scaling evidence in Tables I and II and Fig. 6 supports the parallel-efficiency claims for the tested range, and the paper is transparent about the underlying approximations (Dirac-Coulomb Hamiltonian, scalar-relativistic xc functional). The authors also provide machine-readable input structures and timing/memory data, which is a strength. However, the physical validity of the large-scale results rests on the completeness and correctness of the distributed matrix assembly, and the paper does not yet supply a direct numerical verification against an independent full-matrix construction or the previously validated Q4C implementation. That missing check is the main uncertainty.

major comments (2)
  1. [Sec. III A, Eqs. (33)-(36) and Algorithm 1] The correctness of the distributed assembly is asserted but not verified. The global spinor Hamiltonian/overlap matrices are assembled by summing per-task partial integrals and k-phases (Eqs. 33-36), with the statement that the full matrix is never gathered. A dropped or double-counted batch contribution, or a phase/index error, would leave all timings and scalability results unchanged while making the band structures wrong. The paper does not compare against the previous validated Q4C implementation (Ref. 3) even for a small molecule or small supercell, nor against a serial full-matrix construction, and the NOMAD deposit does not include raw matrices. Please add a numerical validation: reproduce total energies and/or band structures for a small system with both the new parallel code and the previous implementation, or compare selected assembled H/S matrix elements against a reference fu
  2. [Sec. IV A, Table II and surrounding text] The sentence 'Decreasing the node counts from 16 to 8 effectively reduces the maximum peak memory by nearly 55–63%' contradicts Table II. Going from 16 to 8 nodes increases per-task peak memory from 1693 to 2688 MB (1504-atom system) and from 3096 to 5585 MB (3383-atom system). The actual per-task memory reductions when going from 8 to 16 nodes are about 37% and 45%, not 55–63%. Please correct the direction and the percentages, or rewrite the sentence to describe the memory-per-task reduction as the node count is increased.
minor comments (7)
  1. [Sec. IV B, Fig. 7] The text appears to swap panels (b) and (d) for the 1504-PEPI case: it refers to 'the SR band structure from Fig. 7 (d)' and 'The quasi-four-component (Q4C) from Fig.7 (b)', while the caption labels (b) as SR and (d) as Q4C. Please correct the references.
  2. [Sec. IV A, memory paragraph] The statement that D_S accounts for 'roughly 15–20%' of peak per-task memory is slightly off for the 1504-atom cases (about 21%). Please check the rounding.
  3. [Eq. (36)] There is a typo in the phase factor: it should be T_{M(µ')}, not 'T_{M)(µ')'.
  4. [Fig. 6] The axes of Fig. 6(b) are not described in the text. Please define the abscissa (e.g., number of atoms or basis functions) and state whether the plot uses logarithmic axes, so the claimed O(N)/near-ideal scaling can be read quantitatively.
  5. [Sec. V] Typo: 'Several directions are are still needed' should be 'Several directions are still needed'.
  6. [Sec. IV B] The strong-scaling claim rests on only one doubling (336 to 672 cores). This is acceptable for the stated range, but a third data point or a weak-scaling test would strengthen the generality of the 'nearly ideal' conclusion.
  7. [Sec. II A / Abstract] The paper uses the phrase 'fully relativistic DFT' while explicitly neglecting orbital current terms and using scalar-relativistic xc functionals. These qualifications are stated in Sec. II A; consider repeating a brief qualifier in the abstract or conclusions to avoid overstatement.

Circularity Check

0 steps flagged

No significant circularity: the paper's central claims are direct benchmarks and demonstrable band structures, not outputs forced by fitted inputs or self-citation.

full rationale

The paper's central assertions—per-SCF wall times, scaling factors, peak memory, and band structures for the 1504- and 3383-atom supercells—are direct measurements of an implementation, not quantities derived from fitted parameters. No parameter is fitted to the scaling result and no physical prediction is extracted from a fit; the timings are presented as measured outcomes. The derivation of the Q4C Hamiltonian (Eqs. 16-19) follows from the free-atom Dirac radial solutions and the kinetic-balance relation for the basis, which are stated assumptions rather than restatements of the paper's benchmark conclusions. The paper does rely on prior work by the same authors (Ref. 3 for Q4C, Refs. 18/37 for FHI-aims locally indexed grids, Refs. 33-35 for ELPA/ELSI), but this is disclosed lineage and is not load-bearing in a circular sense: the scalable distributed-memory assembly in Section III A is the new contribution and its correctness is asserted algorithmically, not imported as a pre-fitted result. The absence of an independent full-matrix cross-check is a correctness risk (a dropped or double-counted batch would invalidate the band structures while leaving timings intact), but no equation in the paper defines the benchmark output in terms of an input fitted to that output, so this does not constitute circularity under the specified standards.

Axiom & Free-Parameter Ledger

0 free parameters · 6 axioms · 0 invented entities

The central claim is a scalability and implementation claim, so it introduces no new physical particles, forces, or fitted constants. It relies on established relativistic DFT approximations, the Q4C ansatz, standard numerical quadrature, and the correctness of the distributed summation scheme.

axioms (6)
  • domain assumption Dirac-Coulomb Hamiltonian under the no-pair approximation, without orbital currents
    Used as the starting point and total-energy expression (Eqs. 1 and 3, Section II A); standard for relativistic electronic structure but an approximation to full QED.
  • domain assumption Q4C shares one coefficient vector C+ between large and small components (Eq. 6)
    Section II B; this is the central approximation that makes Q4C quasi-four-component rather than exact 4C.
  • domain assumption Atomic-balance kinetic identities, Eqs. (17a,b), with free-atom epsilon and V
    Section II C; they rewrite c sigma dot p acting on basis functions so that singular integrands cancel.
  • domain assumption Overlapping atom-centered grids with Stratmann partition-of-unity weights
    Section III A; all real-space integrals assume this quadrature is exact enough at intermediate settings.
  • standard math Locally-indexed sum over batches/tasks equals the global matrix element
    Eqs. 31-36; if any batch contribution is omitted or double-counted the Hamiltonian is wrong; load-bearing for the distributed implementation.
  • domain assumption PBE exchange-correlation functional is the DFT approximation
    Section IV A; benchmarks and band structures use PBE.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Large-scale Parallel Implementation of Quasi-Four-Component Relativistic Density Functional Theory with Numeric Atom-centered Orbitals." pith.science (2026). https://pith.science/paper/2BHQFXKC

@misc{pith2026260800773,
  author       = {Pith},
  title        = {Pith review of: A Large-scale Parallel Implementation of Quasi-Four-Component Relativistic Density Functional Theory with Numeric Atom-centered Orbitals},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2BHQFXKC}},
  note         = {Machine review of arXiv:2608.00773}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We present a large-scale parallel implementation of fully relativistic density functional theory (DFT) for both molecules and periodic solids, using the quasi-four-component (Q4C) method and numeric atom-centered orbital basis sets. Our approach employs a domain decomposition method on nonuniform real-space integration grids, which enables order-N integration of the Q4C Hamiltonian matrix elements using efficient, distributed-memory and compute-parallel real-space operations. Next, we build the Hamiltonian and overlap matrices in a two-dimensional block-cyclic distribution layout. The resulting generalized eigenvalue problems are solved with the massively parallel ELPA eigenvalue solver library. We benchmark memory usage, parallel efficiency, and scalability across multiple MPI tasks and compute nodes. This algorithm extends the reach of fully relativistic DFT simulations for periodic solids, tested up to 3,383 atoms per unit cell (216,628 basis functions) and likely still well below the true reach of the implementation. As a demonstration, we calculate the fully relativistic band structure for a 3,383 atom-per-unit-cell doped hybrid organic-inorganic perovskite, (PEA)2(Pb1-xBix)I4 (PEA=phenethylammonium), showing nearly ideal scalability between 336 and 672 physical CPU cores.

Figures

Figures reproduced from arXiv: 2608.00773 by Rundong Zhao, Volker Blum, Wentao Zhang.

Figure 1
Figure 1. Figure 1: FIG. 1. Real-space grid partitioning for a benzene molecule. Colored points represent the local set of [PITH_FULL_IMAGE:figures/full_fig_p012_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Real-space grid partitioning for an 8 [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Modified “locally indexed” dense storage and transformation scheme for a non-periodic molecule [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. Modified “locally indexed” dense storage and transformation scheme for the periodic system with [PITH_FULL_IMAGE:figures/full_fig_p017_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. High-level workflow of the large-scale parallel distributed Q4C implementation with periodic [PITH_FULL_IMAGE:figures/full_fig_p021_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6. (a) Structure of a 1080-atom CsPbBr [PITH_FULL_IMAGE:figures/full_fig_p023_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7. Crystal structures and electronic band structures of two PEPI supercells computed with the PBE [PITH_FULL_IMAGE:figures/full_fig_p027_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

48 extracted references · 29 canonical work pages

  1. [1]

    \ ( year 2026 ) NoStop

    https://pubchem.ncbi.nlm.nih.gov/compound/241 title PubChem Compound Summary for CID 241, Benzene, Retrieved June 3, 2026 from \\ https://pubchem.ncbi.nlm.nih.gov/compound/benzene. \ ( year 2026 ) NoStop

    work page 2026

  2. [2]

    Baskin \ and\ author L

    author author Y. Baskin \ and\ author L. Meyer ,\ 10.1103/PhysRev.100.544 journal journal Phys. Rev. \ volume 100 ,\ pages 544 ( year 1955 ) NoStop

    work page doi:10.1103/physrev.100.544 1955

  3. [3]

    author author C. C. \ Stoumpos , author C. D. \ Malliakas , author J. A. \ Peters , author Z. Liu , author M. Sebastian , author J. Im , author T. C. \ Chasapis , author A. C. \ Wibowo , author D. Y. \ Chung , author A. J. \ Freeman , author B. W. \ Wessels , \ and\ author M. G. \ Kanatzidis ,\ 10.1021/cg400645t journal journal Crystal Growth & Design \ v...

    work page doi:10.1021/cg400645t 2013

  4. [4]

    Lu , author G

    author author H. Lu , author G. Koknat , author Y. Yao , author J. Hao , author X. Qin , author C. Xiao , author R. Song , author F. Merz , author M. Rampp , author S. Kokott , author C. Carbogno , author T. Li , author G. Teeter , author M. Scheffler , author J. J. \ Berry , author D. B. \ Mitzi , author J. L. \ Blackburn , author V. Blum , \ and\ author...

    work page doi:10.1103/prxenergy.2.023010 2023

  5. [5]

    Dirac ,\ @noop journal journal Proc

    author author P. Dirac ,\ @noop journal journal Proc. Roy. Soc. London. Series A \ volume 117 ,\ pages 610 ( year 1928 ) NoStop

    work page 1928

  6. [6]

    author author W. P. \ Huhn \ and\ author V. Blum ,\ 10.1103/PhysRevMaterials.1.033803 journal journal Phys. Rev. Mater. \ volume 1 ,\ pages 033803 ( year 2017 ) NoStop

    doi:10.1103/physrevmaterials.1.033803 2017

  7. [7]

    Zhao , author V

    author author R. Zhao , author V. W.-z. \ Yu , author K. Zhang , author Y. Xiao , author Y. Zhang , \ and\ author V. Blum ,\ 10.1103/PhysRevB.103.245144 journal journal Phys. Rev. B \ volume 103 ,\ pages 245144 ( year 2021 ) NoStop

    doi:10.1103/physrevb.103.245144 2021

  8. [8]

    author author D. D. \ Koelling \ and\ author B. N. \ Harmon ,\ 10.1088/0022-3719/10/16/019 journal journal Journal of Physics C: Solid State Physics \ volume 10 ,\ pages 3107 ( year 1977 ) NoStop

    doi:10.1088/0022-3719/10/16/019 1977

  9. [9]

    Keller , author V

    author author L. Keller , author V. Blum , author P. Rinke , \ and\ author D. Golze ,\ 10.1063/5.0018231 journal journal The Journal of Chemical Physics \ volume 153 ,\ pages 114110 ( year 2020 ) NoStop

    doi:10.1063/5.0018231 2020

  10. [10]

    author author J. Autschbach ,\ 10.1098/rsta.2012.0489 journal journal Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences \ volume 372 ,\ pages 20120489 ( year 2014 ) ,\ http://arxiv.org/abs/https://royalsocietypublishing.org/rsta/article-pdf/doi/10.1098/rsta.2012.0489/399878/rsta.2012.0489.pdf https://royals...

    work page arXiv 2012

  11. [11]

    Liu , author F

    author author W. Liu , author F. Wang , \ and\ author L. Li ,\ 10.1142/S0219633603000471 journal journal Journal of Theoretical and Computational Chemistry \ volume 02 ,\ pages 257 ( year 2003 ) NoStop

    work page doi:10.1142/s0219633603000471 2003

  12. [12]

    Zhang , author B

    author author Y. Zhang , author B. Suo , author Z. Wang , author N. Zhang , author Z. Li , author Y. Lei , author W. Zou , author J. Gao , author D. Peng , author Z. Pu , author Y. Xiao , author Q. Sun , author F. Wang , author Y. Ma , author X. Wang , author Y. Guo , \ and\ author W. Liu ,\ 10.1063/1.5143173 journal journal The Journal of Chemical Physic...

    work page doi:10.1063/1.5143173 2020

  13. [13]

    author author D. A. \ Rehn , author J. M. \ Wills , author T. E. \ Battelle , \ and\ author A. E. \ Mattsson ,\ 10.1103/PhysRevB.101.085114 journal journal Physical Review B \ volume 101 ,\ pages 085114 ( year 2020 ) NoStop

    work page doi:10.1103/physrevb.101.085114 2020

  14. [14]

    Blaha , author K

    author author P. Blaha , author K. Schwarz , author F. Tran , author R. Laskowski , author G. K. H. \ Madsen , \ and\ author L. D. \ Marks ,\ 10.1063/1.5143061 journal journal The Journal of Chemical Physics \ volume 152 ,\ pages 074101 ( year 2020 ) NoStop

    doi:10.1063/1.5143061 2020

  15. [15]

    Saue \ and\ author T

    author author T. Saue \ and\ author T. Helgaker ,\ https://doi.org/10.1002/jcc.10066 journal journal Journal of Computational Chemistry \ volume 23 ,\ pages 814 ( year 2002 ) NoStop

    work page doi:10.1002/jcc.10066 2002

  16. [16]

    Belpassi , author M

    author author L. Belpassi , author M. De Santis , author H. M. \ Quiney , author F. Tarantelli , \ and\ author L. Storchi ,\ 10.1063/5.0002831 journal journal The Journal of Chemical Physics \ volume 152 ,\ pages 164118 ( year 2020 ) NoStop

    work page doi:10.1063/5.0002831 2020

  17. [17]

    Kadek , author M

    author author M. Kadek , author M. Repisky , \ and\ author K. Ruud ,\ 10.1103/PhysRevB.99.205103 journal journal Phys. Rev. B \ volume 99 ,\ pages 205103 ( year 2019 ) NoStop

    doi:10.1103/physrevb.99.205103 2019

  18. [18]

    author author D. B. \ Williams-Young , author A. Petrone , author S. Sun , author T. F. \ Stetina , author P. Lestrange , author C. E. \ Hoyer , author D. R. \ Nascimento , author L. Koulias , author A. Wildman , author J. Kasper , author J. J. \ Goings , author F. Ding , author A. E. \ DePrince III , author E. F. \ Valeev , \ and\ author X. Li ,\ @noop j...

    work page 2019

  19. [19]

    Visscher \ and\ author E

    author author L. Visscher \ and\ author E. van Lenthe ,\ https://doi.org/10.1016/S0009-2614(99)00458-3 journal journal Chemical Physics Letters \ volume 306 ,\ pages 357 ( year 1999 ) NoStop

    work page doi:10.1016/s0009-2614(99)00458-3 1999

  20. [20]

    author author W. A. \ de Jong , author R. J. \ Harrison , \ and\ author D. A. \ Dixon ,\ 10.1063/1.1329891 journal journal The Journal of Chemical Physics \ volume 114 ,\ pages 48 ( year 2001 ) NoStop

    doi:10.1063/1.1329891 2001

  21. [21]

    author author A. D. \ Becke ,\ 10.1063/1.454033 journal journal The Journal of Chemical Physics \ volume 88 ,\ pages 2547 ( year 1988 ) NoStop

    doi:10.1063/1.454033 1988

  22. [22]

    Blum , author R

    author author V. Blum , author R. Gehrke , author F. Hanke , author P. Havu , author V. Havu , author X. Ren , author K. Reuter , \ and\ author M. Scheffler ,\ https://doi.org/10.1016/j.cpc.2009.06.022 journal journal Computer Physics Communications \ volume 180 ,\ pages 2175 ( year 2009 ) NoStop

    doi:10.1016/j.cpc.2009.06.022 2009

  23. [23]

    Havu , author V

    author author V. Havu , author V. Blum , author P. Havu , \ and\ author M. Scheffler ,\ https://doi.org/10.1016/j.jcp.2009.08.008 journal journal Journal of Computational Physics \ volume 228 ,\ pages 8367 ( year 2009 ) NoStop

    doi:10.1016/j.jcp.2009.08.008 2009

  24. [24]

    uneis , author J. G\

    author author J. W. \ Abbott , author C. M. \ Acosta , author A. Akkoush , author A. Ambrosetti , author V. Atalla , author A. Bagrets , author J. Behler , author D. Berger , author H. Bertschi , author B. Bieniek , author J. Bj\"ork , author V. Blum , author S. Bohloul , author C. L. \ Box , author N. Boyer , author D. S. \ Brambila , author G. A. \ Bram...

    2026

  25. [25]

    author author A. K. \ Rajagopal \ and\ author J. Callaway ,\ 10.1103/PhysRevB.7.1912 journal journal Phys. Rev. B \ volume 7 ,\ pages 1912 ( year 1973 ) NoStop

    doi:10.1103/physrevb.7.1912 1912

  26. [26]

    author author A. K. \ Rajagopal ,\ 10.1088/0022-3719/11/24/002 journal journal Journal of Physics C: Solid State Physics \ volume 11 ,\ pages L943 ( year 1978 ) NoStop

    work page doi:10.1088/0022-3719/11/24/002 1978

  27. [27]

    author author K. G. \ Dyall \ and\ author K. F gri Jr. ,\ @noop title Introduction to Relativistic Quantum Chemistry \ ( publisher Oxford University Press ,\ year 2007 ) NoStop

    work page 2007

  28. [28]

    Autschbach ,\ 10.1063/1.3702628 journal journal The Journal of Chemical Physics \ volume 136 ,\ pages 150902 ( year 2012 ) NoStop

    author author J. Autschbach ,\ 10.1063/1.3702628 journal journal The Journal of Chemical Physics \ volume 136 ,\ pages 150902 ( year 2012 ) NoStop

    work page doi:10.1063/1.3702628 2012

  29. [29]

    Almoukhalalati , author S

    author author A. Almoukhalalati , author S. Knecht , author H. J. A. \ Jensen , author K. G. \ Dyall , \ and\ author T. Saue ,\ 10.1063/1.4959452 journal journal The Journal of Chemical Physics \ volume 145 ,\ pages 074104 ( year 2016 ) NoStop

    work page doi:10.1063/1.4959452 2016

  30. [30]

    Liu \ and\ author D

    author author W. Liu \ and\ author D. Peng ,\ 10.1063/1.3159445 journal journal The Journal of Chemical Physics \ volume 131 ,\ pages 031104 ( year 2009 ) NoStop

    doi:10.1063/1.3159445 2009

  31. [31]

    Kutzelnigg \ and\ author W

    author author W. Kutzelnigg \ and\ author W. Liu ,\ 10.1063/1.2137315 journal journal The Journal of Chemical Physics \ volume 123 ,\ pages 241102 ( year 2005 ) NoStop

    doi:10.1063/1.2137315 2005

  32. [32]

    author author K. G. \ Dyall ,\ 10.1063/1.473860 journal journal The Journal of Chemical Physics \ volume 106 ,\ pages 9618 ( year 1997 ) NoStop

    work page doi:10.1063/1.473860 1997

  33. [33]

    Liu \ and\ author D

    author author W. Liu \ and\ author D. Peng ,\ 10.1063/1.2222365 journal journal The Journal of Chemical Physics \ volume 125 ,\ pages 044102 ( year 2006 ) NoStop

    work page doi:10.1063/1.2222365 2006

  34. [34]

    Visscher , author P

    author author L. Visscher , author P. J. C. \ Aerts , author O. Visser , \ and\ author W. C. \ Nieuwpoort ,\ https://doi.org/10.1002/qua.560400816 journal journal International Journal of Quantum Chemistry \ volume 40 ,\ pages 131 ( year 1991 ) NoStop

    work page doi:10.1002/qua.560400816 1991

  35. [35]

    Čertík , author J

    author author O. Čertík , author J. E. \ Pask , \ and\ author J. Vackář ,\ https://doi.org/10.1016/j.cpc.2013.02.014 journal journal Computer Physics Communications \ volume 184 ,\ pages 1777 ( year 2013 ) NoStop

    work page doi:10.1016/j.cpc.2013.02.014 2013

  36. [36]

    Marek , author V

    author author A. Marek , author V. Blum , author R. Johanni , author V. Havu , author B. Lang , author T. Auckenthaler , author A. Heinecke , author H.-J. \ Bungartz , \ and\ author H. Lederer ,\ 10.1088/0953-8984/26/21/213201 journal journal Journal of Physics: Condensed Matter \ volume 26 ,\ pages 213201 ( year 2014 ) NoStop

    doi:10.1088/0953-8984/26/21/213201 2014

  37. [37]

    author author V. W. \ zhe Yu , author F. Corsetti , author A. García , author W. P. \ Huhn , author M. Jacquelin , author W. Jia , author B. Lange , author L. Lin , author J. Lu , author W. Mi , author A. Seifitokaldani , author Álvaro Vázquez-Mayagoitia , author C. Yang , author H. Yang , \ and\ author V. Blum ,\ https://doi.org/10.1016/j.cpc.2017.09.007...

    doi:10.1016/j.cpc.2017.09.007 2017

  38. [38]

    author author V. W. \ zhe Yu , author C. Campos , author W. Dawson , author A. García , author V. Havu , author B. Hourahine , author W. P. \ Huhn , author M. Jacquelin , author W. Jia , author M. Keçeli , author R. Laasner , author Y. Li , author L. Lin , author J. Lu , author J. Moussa , author J. E. \ Roman , author Álvaro Vázquez-Mayagoitia , author C...

    work page arXiv 2020

  39. [39]

    Knuth , author C

    author author F. Knuth , author C. Carbogno , author V. Atalla , author V. Blum , \ and\ author M. Scheffler ,\ https://doi.org/10.1016/j.cpc.2015.01.003 journal journal Computer Physics Communications \ volume 190 ,\ pages 33 ( year 2015 ) NoStop

    work page doi:10.1016/j.cpc.2015.01.003 2015

  40. [40]

    author author W. P. \ Huhn , author B. Lange , author V. W. \ zhe Yu , author M. Yoon , \ and\ author V. Blum ,\ https://doi.org/10.1016/j.cpc.2020.107314 journal journal Computer Physics Communications \ volume 254 ,\ pages 107314 ( year 2020 ) NoStop

    work page arXiv 2020

  41. [41]

    Kokott , author F

    author author S. Kokott , author F. Merz , author Y. Yao , author C. Carbogno , author M. Rossi , author V. Havu , author M. Rampp , author M. Scheffler , \ and\ author V. Blum ,\ 10.1063/5.0208103 journal journal The Journal of Chemical Physics \ volume 161 ,\ pages 024112 ( year 2024 ) NoStop

    doi:10.1063/5.0208103 2024

  42. [42]

    Delley ,\ 10.1063/1.458452 journal journal The Journal of Chemical Physics \ volume 92 ,\ pages 508 ( year 1990 ) NoStop

    author author B. Delley ,\ 10.1063/1.458452 journal journal The Journal of Chemical Physics \ volume 92 ,\ pages 508 ( year 1990 ) NoStop

    work page doi:10.1063/1.458452 1990

  43. [43]

    Lebedev ,\ https://doi.org/10.1016/0041-5553(75)90133-0 journal journal USSR Computational Mathematics and Mathematical Physics \ volume 15 ,\ pages 44 ( year 1975 ) NoStop

    author author V. Lebedev ,\ https://doi.org/10.1016/0041-5553(75)90133-0 journal journal USSR Computational Mathematics and Mathematical Physics \ volume 15 ,\ pages 44 ( year 1975 ) NoStop

    work page doi:10.1016/0041-5553(75)90133-0 1975

  44. [44]

    Lebedev ,\ https://doi.org/10.1016/0041-5553(76)90100-2 journal journal USSR Computational Mathematics and Mathematical Physics \ volume 16 ,\ pages 10 ( year 1976 ) NoStop

    author author V. Lebedev ,\ https://doi.org/10.1016/0041-5553(76)90100-2 journal journal USSR Computational Mathematics and Mathematical Physics \ volume 16 ,\ pages 10 ( year 1976 ) NoStop

    doi:10.1016/0041-5553(76)90100-2 1976

  45. [45]

    Stratmann , author G

    author author R. Stratmann , author G. E. \ Scuseria , \ and\ author M. J. \ Frisch ,\ https://doi.org/10.1016/0009-2614(96)00600-8 journal journal Chemical Physics Letters \ volume 257 ,\ pages 213 ( year 1996 ) NoStop

    work page doi:10.1016/0009-2614(96)00600-8 1996

  46. [46]

    author author J. P. \ Perdew , author K. Burke , \ and\ author M. Ernzerhof ,\ 10.1103/PhysRevLett.77.3865 journal journal Phys. Rev. Lett. \ volume 77 ,\ pages 3865 ( year 1996 ) NoStop

    doi:10.1103/physrevlett.77.3865 1996

  47. [47]

    Sun , author T

    author author S. Sun , author T. F. \ Stetina , author T. Zhang , author H. Hu , author E. F. \ Valeev , author Q. Sun , \ and\ author X. Li ,\ 10.1021/acs.jctc.1c00137 journal journal Journal of Chemical Theory and Computation \ volume 17 ,\ pages 3388 ( year 2021 ) ,\ note pMID: 34029469 NoStop

    work page doi:10.1021/acs.jctc.1c00137 2021

  48. [48]

    author author M. A. \ Blanco , author M. Flórez , \ and\ author M. Bermejo ,\ https://doi.org/10.1016/S0166-1280(97)00185-1 journal journal Journal of Molecular Structure: THEOCHEM \ volume 419 ,\ pages 19 ( year 1997 ) NoStop

    work page doi:10.1016/s0166-1280(97)00185-1 1997

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.