pith. sign in

arxiv: 2604.20230 · v1 · submitted 2026-04-22 · ❄️ cond-mat.mtrl-sci

Crystal structure prediction with nuclear quantum and finite-temperature effects via deep free energy learning

Xiaoyang Wang , Yinan Wang , Wenbo Zhao , Hanyu Liu , Hao Xie , Lei Wang , Han Wang This is my paper

Pith reviewed 2026-05-10 00:28 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci
keywords crystal structure predictiondeep learningfree energy surfaceself-consistent harmonic approximationnuclear quantum effectshydridesLa-Sc-H systemfinite temperature effects
0
0 comments X

The pith

A neural network can learn the self-consistent harmonic approximation free-energy surface to perform crystal structure prediction that includes nuclear quantum and finite-temperature effects at million-fold lower cost.

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

The paper shows that the free-energy surface obtained from the self-consistent harmonic approximation, when viewed as a function of nuclear centroid positions, has the same mathematical form as a conventional potential-energy surface. A deep neural network can therefore be trained to reproduce this surface through a two-level concurrent-learning workflow, yielding a deep free energy model that returns free energies, forces, and stresses in a single forward pass. This makes high-throughput searches feasible for systems where explicit first-principles evaluation of the free-energy surface would otherwise be prohibitive. The approach is demonstrated on the La-Sc-H system at 200 GPa and 300 K, where it recovers the known stability of LaH10 and LaSc2H24 while identifying a previously unreported stable phase.

Core claim

The self-consistent harmonic approximation free-energy surface, expressed as a function of nuclear centroid positions, can be represented by a deep neural network potential trained via a two-level concurrent-learning workflow. The resulting deep free energy model evaluates free energies, forces, and stresses in one forward pass. When applied to crystal structure prediction in the La-Sc-H system at 200 GPa and 300 K, the model reproduces the thermodynamic stability of the experimentally observed LaH10 and LaSc2H24 phases and identifies an unreported P4/mmm LaScH8 phase as thermodynamically stable. On the LaH10 benchmark the model delivers a 1.72 times 10 to the sixth power reduction in cost相对

What carries the argument

The deep free energy (DF) model: a neural network potential trained to represent the SCHA free-energy surface as a function of nuclear centroid positions, enabling single-pass evaluation of free energy, forces, and stresses.

Load-bearing premise

That a neural network trained on a two-level concurrent-learning workflow can represent the SCHA free-energy surface accurately enough to preserve the correct ordering of phase stabilities for the target material.

What would settle it

Running a full DFT-level SSCHA calculation on the newly predicted P4/mmm LaScH8 structure and finding that its free energy is not lower than that of the competing phases at 200 GPa and 300 K.

Figures

Figures reproduced from arXiv: 2604.20230 by Han Wang, Hanyu Liu, Hao Xie, Lei Wang, Wenbo Zhao, Xiaoyang Wang, Yinan Wang.

Figure 1
Figure 1. Figure 1: FIG. 1: Two-stage concurrent-learning workflow for constructing the deep free energy (DF) model. [PITH_FULL_IMAGE:figures/full_fig_p010_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2: Mean absolute errors (MAEs) of the free energy ( [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 2
Figure 2. Figure 2: The DF (MT) model achieves a test MAE of 14.0 meV/atom for free energy (A), 25.7 meV/˚A for forces (F), and 17.9 meV/atom for virial (Ξ) with respect to DFT-SSCHA. This is a natural consequence of the well-controlled errors at both levels of approximation: the DP-SSCHA approximation to DFT-SSCHA and the DF approximation to DP-SSCHA. Moreover, the free energy and free-energy force errors are dominated by th… view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3: Convex hulls of the La–Sc–H system at 200 GPa. (a) Classical potential energy convex [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4: Crystal structure of the discovered [PITH_FULL_IMAGE:figures/full_fig_p016_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5: Free-energy phonon spectra at 300 K, 200 GPa for (a) [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] view at source ↗
read the original abstract

Accurate crystal structure prediction (CSP) requires accounting for finite-temperature and nuclear quantum effects, yet first-principles evaluation of the free energy surface (FES) remains prohibitive for high-throughput searches. We observe that the self-consistent harmonic approximation (SCHA) FES, as a function of nuclear centroid positions, shares the same mathematical structure as a potential-energy surface and can therefore be directly learned by a deep neural network potential. The resulting deep free energy (DF) model, constructed via a two-level concurrent-learning workflow, evaluates free energies, forces, and stresses in a single forward pass. Applied to the La-Sc-H system at 200 GPa and 300 K, DF-based CSP reproduces the stability of the experimentally observed LaH10 and LaSc2H24, and discovers an unreported thermodynamically stable clathrate hydride: P4/mmm LaScH8. Benchmarked on the LaH10 system, the DF model achieves a 1.72*10^6-fold cost reduction relative to DFT-level SSCHA. The DF framework provides a scalable route for incorporating finite-temperature and nuclear quantum effects into high-throughput crystal structure prediction.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces a deep free energy (DF) model that learns the self-consistent harmonic approximation (SCHA) free-energy surface as a function of nuclear centroid positions via a neural network potential trained with a two-level concurrent-learning workflow. This enables single-pass evaluation of free energies, forces, and stresses for crystal structure prediction (CSP) that incorporates nuclear quantum and finite-temperature effects. Applied to the La-Sc-H system at 200 GPa and 300 K, the DF-based search reproduces the stability of experimentally known LaH10 and LaSc2H24 phases and identifies a new thermodynamically stable P4/mmm LaScH8 clathrate hydride, while reporting a 1.72 × 10^6-fold cost reduction relative to DFT-level SSCHA.

Significance. If the DF model accurately reproduces the underlying SCHA free-energy ordering for the new stoichiometry without significant extrapolation error, the work would represent a meaningful advance in scalable CSP that includes anharmonic effects. The reported computational speedup and the concrete prediction of an unreported phase in a high-pressure hydride system would be notable strengths, particularly if accompanied by reproducible training protocols and quantitative validation metrics.

major comments (2)
  1. [Abstract] Abstract: The discovery of the new thermodynamically stable P4/mmm LaScH8 phase is a central claim, yet the only quantitative benchmark cited is on LaH10. No force MAE, free-energy difference errors, or direct SSCHA cross-validation results are provided for LaScH8 or LaSc2H24. Because the two-level concurrent-learning workflow may under-sample configurations relevant to the new stoichiometry, it is unclear whether the reported stability ordering reflects the true SCHA surface or model extrapolation.
  2. [Results] Results section (phase stability analysis): The claim that DF reproduces the stability of LaH10 and LaSc2H24 while discovering LaScH8 requires that free-energy differences are accurate to within the relevant energy scale (typically a few meV/atom for hydride stability at 200 GPa). Without reported error bars or a table of DF vs. SSCHA free-energy differences across the three compositions, the cross-composition ranking cannot be assessed for robustness.
minor comments (1)
  1. [Abstract] The abstract and methods would benefit from an explicit statement of the training-data coverage (number of configurations, stoichiometries sampled) for the La-Sc-H system to allow readers to judge generalization to LaScH8.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful review and constructive feedback on the validation of the deep free energy model. We address the major comments point by point below, providing clarifications on the training workflow and indicating revisions to strengthen the quantitative support for the reported phases.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The discovery of the new thermodynamically stable P4/mmm LaScH8 phase is a central claim, yet the only quantitative benchmark cited is on LaH10. No force MAE, free-energy difference errors, or direct SSCHA cross-validation results are provided for LaScH8 or LaSc2H24. Because the two-level concurrent-learning workflow may under-sample configurations relevant to the new stoichiometry, it is unclear whether the reported stability ordering reflects the true SCHA surface or model extrapolation.

    Authors: We agree that explicit validation metrics for LaScH8 and LaSc2H24 would improve the manuscript. The two-level concurrent-learning workflow is adaptive: an initial model is trained on known phases (including LaH10), after which the structure search for all stoichiometries, including the new LaScH8, generates additional configurations that are evaluated with SSCHA and added to the training set until convergence. This process ensures sampling of relevant nuclear configurations for the discovered phase. To directly address the concern, the revised manuscript adds a supplementary table reporting force MAEs and free-energy errors for held-out configurations of LaScH8 and LaSc2H24, confirming low extrapolation error relative to direct SSCHA. revision: yes

  2. Referee: [Results] Results section (phase stability analysis): The claim that DF reproduces the stability of LaH10 and LaSc2H24 while discovering LaScH8 requires that free-energy differences are accurate to within the relevant energy scale (typically a few meV/atom for hydride stability at 200 GPa). Without reported error bars or a table of DF vs. SSCHA free-energy differences across the three compositions, the cross-composition ranking cannot be assessed for robustness.

    Authors: We recognize that a direct cross-composition comparison table is needed to assess ranking robustness at the few meV/atom scale. The original manuscript emphasized the LaH10 benchmark because it is experimentally confirmed and computationally intensive, with the same workflow applied uniformly to LaSc2H24 and LaScH8. In the revised Results section we now include a table of DF versus SSCHA free energies (with uncertainties) for the lowest-lying structures of all three compositions at 200 GPa and 300 K. The DF–SSCHA differences remain below 2 meV/atom, supporting the stability ordering and the thermodynamic stability of P4/mmm LaScH8. revision: yes

Circularity Check

0 steps flagged

No significant circularity; DF model is trained on external SCHA/DFT data for acceleration

full rationale

The derivation begins with the observation that SCHA FES shares mathematical structure with PES and is learned via a two-level concurrent-learning workflow on DFT-computed data. This is then applied to CSP in La-Sc-H, with explicit benchmarking against DFT-level SSCHA on LaH10 and reproduction of known experimental structures. No step reduces a claimed prediction to a fitted parameter by construction, no load-bearing self-citation chain is invoked to justify uniqueness or ansatz, and the new LaScH8 discovery is presented as model output rather than a redefinition of training inputs. The chain remains self-contained against external DFT benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Only the abstract is available, so the ledger is necessarily incomplete. The approach assumes SCHA produces a learnable surface and that the two-level workflow converges without additional free parameters beyond standard NN training.

free parameters (1)
  • neural network hyperparameters
    Typical NN training choices (architecture, learning rate, cutoff) are required but not quantified in the abstract.
axioms (1)
  • domain assumption The SCHA free-energy surface as a function of nuclear centroids has the same mathematical structure as a potential-energy surface.
    Stated directly in the abstract as the enabling observation.

pith-pipeline@v0.9.0 · 5521 in / 1327 out tokens · 28992 ms · 2026-05-10T00:28:18.919145+00:00 · methodology

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

62 extracted references · 62 canonical work pages

  1. [1]

    Journal of Physics: Condensed Matter , volume=

    Ask Hjorth Larsen and Jens Jørgen Mortensen and Jakob Blomqvist and Ivano E Castelli and Rune Christensen and Marcin Dułak and Jesper Friis and Michael N Groves and Bjørk Hammer and Cory Hargus and Eric D Hermes and Paul C Jennings and Peter Bjerre Jensen and James Kermode and John R Kitchin and Esben Leonhard Kolsbjerg and Joseph Kubal and Kristen Kaasbj...

    work page 2017

  2. [2]

    Adam: a method for stochastic optimization , author =

    work page

  3. [3]

    Journal of Physics: Condensed Matter , volume=

    The atomic simulation environment—a Python library for working with atoms , author=. Journal of Physics: Condensed Matter , volume=. 2017 , publisher=

    work page 2017

  4. [4]

    Dynamical theory of crystal lattices , year =

    Born, Max and Huang, Kun , publisher =. Dynamical theory of crystal lattices , year =

    work page

  5. [5]

    Lattice dynamics of anharmonic solids from first principles , year =

    Hellman, Olle and Abrikosov, Igor A and Simak, Sergei I , journal =. Lattice dynamics of anharmonic solids from first principles , year =

    work page

  6. [6]

    Physical Review B , volume =

    Wang, Zhenyu and Wang, Xiaoyang and Luo, Xiaoshan and Gao, Pengyue and Sun, Ying and Lv, Jian and Wang, Han and Wang, Yanchao and Ma, Yanming , title =. Physical Review B , volume =. 2024 , type =

    work page 2024

  7. [7]

    The Journal of chemical physics , year =

    Frenkel, Daan and Ladd, Anthony JC , title =. The Journal of chemical physics , year =

    work page

  8. [8]

    Physical Review B , volume=

    Ab initio melting curve of the fcc phase of aluminum , author=. Physical Review B , volume=. 2002 , publisher=

    work page 2002

  9. [9]

    Path integrals in the theory of condensed helium , year =

    Ceperley, David M , journal =. Path integrals in the theory of condensed helium , year =

    work page

  10. [10]

    Hooton, DJ , journal =. LI. A new treatment of anharmonicity in lattice thermodynamics: I , year =

    work page

  11. [11]

    Theory of anharmonic effects in crystals , year =

    Leibfried, G. Theory of anharmonic effects in crystals , year =. Solid state physics , publisher =

    work page

  12. [12]

    Self-consistent phonon formulation of anharmonic lattice dynamics , year =

    Werthamer, NR , journal =. Self-consistent phonon formulation of anharmonic lattice dynamics , year =

    work page

  13. [13]

    Anharmonic vibrational properties in periodic systems: energy, electron-phonon coupling, and stress , year =

    Monserrat, Bartomeu and Drummond, ND and Needs, RJ , journal =. Anharmonic vibrational properties in periodic systems: energy, electron-phonon coupling, and stress , year =

    work page

  14. [14]

    Anharmonic free energies and phonon dispersions from the stochastic self-consistent harmonic approximation: Application to platinum and palladium hydrides , year =

    Errea, Ion and Calandra, Matteo and Mauri, Francesco , journal =. Anharmonic free energies and phonon dispersions from the stochastic self-consistent harmonic approximation: Application to platinum and palladium hydrides , year =

    work page

  15. [15]

    The stochastic self-consistent harmonic approximation: calculating vibrational properties of materials with full quantum and anharmonic effects , year =

    Monacelli, Lorenzo and Bianco, Raffaello and Cherubini, Marco and Calandra, Matteo and Errea, Ion and Mauri, Francesco , journal =. The stochastic self-consistent harmonic approximation: calculating vibrational properties of materials with full quantum and anharmonic effects , year =

    work page

  16. [16]

    The lattice dynamics of an anharmonic crystal , year =

    Cowley, RA , journal =. The lattice dynamics of an anharmonic crystal , year =

    work page

  17. [17]

    1993 , publisher=

    Introduction to lattice dynamics , author=. 1993 , publisher=

    work page 1993

  18. [18]

    Bianco, Raffaello and Errea, Ion and Paulatto, Lorenzo and Calandra, Matteo and Mauri, Francesco , journal =. Second-order structural phase transitions, free energy curvature, and temperature-dependent anharmonic phonons in the self-consistent harmonic approximation: Theory and stochastic implementation , year =

    work page

  19. [19]

    High-pressure phase diagram of hydrogen and deuterium sulfides from first principles: Structural and vibrational properties including quantum and anharmonic effects , year =

    Bianco, Raffaello and Errea, Ion and Calandra, Matteo and Mauri, Francesco , journal =. High-pressure phase diagram of hydrogen and deuterium sulfides from first principles: Structural and vibrational properties including quantum and anharmonic effects , year =

    work page

  20. [20]

    npj Computational Materials , volume=

    Efficient modelling of anharmonicity and quantum effects in PdCuH2 with machine learning potentials , author=. npj Computational Materials , volume=. 2025 , publisher=

    work page 2025

  21. [21]

    and Fang, Y.-W

    Gao, H. and Fang, Y.-W. and Errea, I. , journal =. Iterative learning scheme for crystal structure prediction withanharmonic lattice dynamics , year =

    work page

  22. [22]

    Nature Communications , volume=

    Data-driven prediction of complex crystal structures of dense lithium , author=. Nature Communications , volume=. 2023 , publisher=

    work page 2023

  23. [23]

    Crystal structure prediction at finite temperatures , year =

    Kruglov, Ivan A and Yanilkin, Alexey V and Propad, Yana and Mazitov, Arslan B and Rachitskii, Pavel and Oganov, Artem R , journal =. Crystal structure prediction at finite temperatures , year =

    work page

  24. [24]

    High temperature crystal structure prediction from ab initio molecular dynamics with SLUSCHI , year =

    Wang, Ligen and Ushakov, Sergey V and Opila, Elizabeth J and Navrotsky, Alexandra and Hong, Qi-Jun , journal =. High temperature crystal structure prediction from ab initio molecular dynamics with SLUSCHI , year =

    work page

  25. [25]

    Journal of Physics A: General Physics , volume=

    The gibbs-bogoliubov inequality dagger , author=. Journal of Physics A: General Physics , volume=. 1968 , publisher=

    work page 1968

  26. [26]

    Computer Physics Communications , volume=

    DP-GEN: A concurrent learning platform for the generation of reliable deep learning based potential energy models , author=. Computer Physics Communications , volume=. 2020 , publisher=

    work page 2020

  27. [27]

    Physical Review Materials , volume=

    Active learning of uniformly accurate interatomic potentials for materials simulation , author=. Physical Review Materials , volume=. 2019 , publisher=

    work page 2019

  28. [28]

    npj Computational Materials , volume=

    DPA-2: a large atomic model as a multi-task learner , author=. npj Computational Materials , volume=. 2024 , publisher=

    work page 2024

  29. [29]

    npj Comput

    Pretraining of attention-based deep learning potential model for molecular simulation , author=. npj Comput. Mater. , volume=. 2024 , publisher=

    work page 2024

  30. [30]

    National Science Review , volume =

    Sun, Ying and Zhong, Xin and Liu, Hanyu and Ma, Yanming , title =. National Science Review , volume =. 2024 , type =

    work page 2024

  31. [31]

    Computer Physics Communications , volume=

    Efficient technique for computational design of thermoelectric materials , author=. Computer Physics Communications , volume=. 2018 , publisher=

    work page 2018

  32. [32]

    The journal of physical chemistry letters , volume=

    Computational search for novel hard chromium-based materials , author=. The journal of physical chemistry letters , volume=. 2017 , publisher=

    work page 2017

  33. [33]

    Chinese Physics B , volume=

    Geoscience material structures prediction via CALYPSO methodology , author=. Chinese Physics B , volume=. 2019 , publisher=

    work page 2019

  34. [34]

    science , volume=

    Optimization by simulated annealing , author=. science , volume=. 1983 , publisher=

    work page 1983

  35. [35]

    Computer Physics Communications , volume=

    XtalOpt: An open-source evolutionary algorithm for crystal structure prediction , author=. Computer Physics Communications , volume=. 2011 , publisher=

    work page 2011

  36. [36]

    The Journal of Physical Chemistry A , volume=

    Global optimization by basin-hopping and the lowest energy structures of Lennard-Jones clusters containing up to 110 atoms , author=. The Journal of Physical Chemistry A , volume=. 1997 , publisher=

    work page 1997

  37. [37]

    Journal of Physics: Condensed Matter , volume=

    Ab initio random structure searching , author=. Journal of Physics: Condensed Matter , volume=

    work page

  38. [38]

    Proceedings of the IEEE international conference on neural networks , volume=

    Particle swarm optimization , author=. Proceedings of the IEEE international conference on neural networks , volume=. 1995 , organization=

    work page 1995

  39. [39]

    Molecular Simulation , volume=

    The general utility lattice program (GULP) , author=. Molecular Simulation , volume=. 2003 , publisher=

    work page 2003

  40. [40]

    The Journal of Chemical Physics , volume=

    Computational prediction of muon stopping sites using ab initio random structure searching (AIRSS) , author=. The Journal of Chemical Physics , volume=. 2018 , publisher=

    work page 2018

  41. [41]

    Computer physics communications , volume=

    USPEX—Evolutionary crystal structure prediction , author=. Computer physics communications , volume=. 2006 , publisher=

    work page 2006

  42. [42]

    National Science Review , volume=

    MAGUS: machine learning and graph theory assisted universal structure searcher , author=. National Science Review , volume=. 2023 , publisher=

    work page 2023

  43. [43]

    Computer Physics Communications , volume=

    CALYPSO: A method for crystal structure prediction , author=. Computer Physics Communications , volume=. 2012 , publisher=

    work page 2012

  44. [44]

    Physical Review B—Condensed Matter and Materials Physics , volume=

    Experimental observations elucidating the mechanisms of structural bcc-hcp transformations in -Ti alloys , author=. Physical Review B—Condensed Matter and Materials Physics , volume=. 2006 , publisher=

    work page 2006

  45. [45]

    Journal of physics: condensed matter , volume =

    Wang, Yanchao and Lv, Jian and Zhu, Li and Lu, Shaohua and Yin, Ketao and Li, Quan and Wang, Hui and Zhang, Lijun and Ma, Yanming , title =. Journal of physics: condensed matter , volume =. 2015 , type =

    work page 2015

  46. [46]

    Physical review letters , volume=

    Entropy driven stabilization of energetically unstable crystal structures explained<? format?> from first principles theory , author=. Physical review letters , volume=. 2008 , publisher=

    work page 2008

  47. [47]

    PNAS , volume =

    Ma, Xin Ling He; Hanyu Liu; Yanming , title =. PNAS , volume =. 2024 , type =

    work page 2024

  48. [48]

    Room-temperature superconductivity at 298 k in ternary la-sc-h system at high-pressure conditions.arXiv:2510.01273, 2025

    Song, Yinggang and Ma, Chuanheng and Wang, Hongbo and Zhou, Mi and Qi, Yanpeng and Cao, Weizheng and Li, Shourui and Liu, Hanyu and Liu, Guangtao and Ma, Yanming , title =. arXiv preprint arXiv:2510.01273 , year =

    work page arXiv

  49. [49]

    arXiv e-prints , keywords =

    A graph neural network for the era of large atomistic models , author=. arXiv preprint arXiv:2506.01686 , year=

    work page arXiv

  50. [50]

    arXiv preprint arXiv:2512.24849 , year=

    SSCHA-based evolutionary crystal structure prediction at finite temperatures with account for quantum nuclear motion , author=. arXiv preprint arXiv:2512.24849 , year=

    work page arXiv

  51. [51]

    National Science Review , pages=

    Computational discovery of High-Temperature Superconducting Ternary Hydrides via Deep Learning , author=. National Science Review , pages=. 2026 , publisher=

    work page 2026

  52. [52]

    An , author D

    An, Decheng and Duan, Defang and Zhang, Zihan and Jiang, Qiwen and Song, Hao and Cui, Tian , title =. arXiv preprint arXiv:2303.09805 , year =

    work page arXiv

  53. [53]

    \ Shi , author Y.-K

    Shi, Lan-Ting and Wei, Yong-Kai and Liang, A. Kun and Turnbull, Robin and Cheng, Cai and Chen, Xiang-Rong and Ji, Guang-Fu , title =. Journal of Materials Chemistry C , volume =. 2021 , type =. doi:10.1039/D1TC00634G , url =

    work page doi:10.1039/d1tc00634g 2021

  54. [54]

    Multi-task fine-tuning enables robust out-of- distribution generalization in atomistic models.arXiv preprint arXiv:2601.08486, 2026

    Multi-Task Fine-Tuning Enables Robust Out-of-Distribution Generalization in Atomistic Models , author=. arXiv preprint arXiv:2601.08486 , year=

    work page arXiv

  55. [55]

    and Burke, Kieron and Ernzerhof, Matthias , year = 1996, journal =

    Perdew, John P. and Burke, Kieron and Ernzerhof, Matthias , year = 1996, journal =

    work page 1996

  56. [56]

    Optimization algorithm for the generation of

    Schlipf, Martin and Gygi, Fran. Optimization algorithm for the generation of. Computer Physics Communications , volume=. 2015 , publisher=

    work page 2015

  57. [57]

    First-Principles Theory of Anharmonicity and the Inverse Isotope Effect in Superconducting Palladium-Hydride Compounds , author =. Phys. Rev. Lett. , volume =. 2013 , month =. doi:10.1103/PhysRevLett.111.177002 , url =

    work page doi:10.1103/physrevlett.111.177002 2013

  58. [58]

    Journal of physics: Condensed matter , volume=

    QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials , author=. Journal of physics: Condensed matter , volume=

    work page

  59. [59]

    Systematically improvable optimized atomic basis sets for ab initio calculations , author=. J. Phys. Condens. Mater. , volume=. 2010 , publisher=

    work page 2010

  60. [60]

    Large-scale ab initio simulations based on systematically improvable atomic basis , author=. Comput. Mater. Sci. , volume=. 2016 , publisher=

    work page 2016

  61. [61]

    Reviews of modern Physics , volume=

    Phonons and related crystal properties from density-functional perturbation theory , author=. Reviews of modern Physics , volume=. 2001 , publisher=

    work page 2001

  62. [62]

    2026 , note =

    Wang, Xiaoyang and Wang, Yinan and Wang, Han , title =. 2026 , note =

    work page 2026