Pith. sign in

REVIEW 5 major objections 5 minor 94 references

Graph-Theoretic Neural Network Fragmentation with Covariant Direct Molecular Force Learning: Enabling Coupled-Cluster Accuracy AIMD for Fluxional Systems

T0 review · 5 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read The paper claims that directly predicting nuclear force vectors on graph-theoretic fragments, instead of differentiating a learned energy surface, reproduces CCSD forces for the fluxional solvated Zundel cation H13O6+ well enough that fully

desk verdict Genuine method advance in direct vector force learning, but the 'CCSD accuracy' claim is not established: validation uses the authors' own fragment reference, whose 56 µEh/Bohr error is larger than the claimed 10 µEh/Bohr. read the letter →

arxiv 2607.21779 v1 pith:6T6M4SNE submitted 2026-07-23 physics.chem-ph cs.AIphysics.comp-ph

Xiao Zhu , Srinivasan S. Iyengar This is my paper · ORCID
classification physics.chem-phcs.AIphysics.comp-ph
keywords machine-learnedforcefieldsgraph-theoreticfragmentationcoupledcluster(CCSD)directvectorlearningprincipalaxesdescriptorsabinitiomoleculardynamicssolvatedZundelcationk-meanstessellation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that machine-learned molecular dynamics can run at coupled-cluster accuracy for systems where cheap density functionals are unreliable, using a fragment-based neural network that predicts force vectors directly rather than deriving them from a learned energy. The authors report force errors near 10 micro-Hartree/Bohr against their CCSD-derived reference and show that fully ML-predicted trajectories of the solvated Zundel cation reproduce key structural and dynamical signatures. The main contribution is the combination of covariant principal-axis force descriptors, vector-valued training that shares hidden features across force components, graph-theoretic inclusion-exclusion assembly of fragment corrections, and mini-batch k-means tessellation for training-set selection, which together stretch expensive reference data much farther than scalar force fitting.

What carries the argument

The central identity is the graph-theoretic fragmentation of post-Hartree-Fock forces, Eq. (13)/(22): full-system force corrections are assembled from fragment-level corrections weighted by inclusion-exclusion coefficients M^R derived from simplex counts, then mapped back through the Jacobian ∂x_fragment/∂x. The other key object is the covariant descriptor: each fragment is rotated so its force components align with its principal axes of inertia, producing three distance-matrix inputs and three vector outputs that preserve translational, rotational, and permutational symmetry. Vector-valued force networks share hidden-layer features across force components, cutting trainable parameters by ov

What would settle it

Compute full-system CCSD (or CCSD(T)) nuclear gradients for a set of representative solvated-Zundel configurations, including geometries with stretched and shared protons, and compare them with the graph-fragmentation reference used here. If the per-component deviation exceeds roughly 10 micro-Hartree/Bohr, or if trajectories driven by the full-system gradients diverge from those driven by the fragmented reference, the central accuracy claim fails.

Watch

Extended reading notes

Core claim

On the authors' terms, the central discovery is that nuclear forces at CCSD quality can be learned as direct vector outputs of fragment neural networks using an interatomic-distance descriptor rotated into each fragment's principal moment-of-inertia frame and projected back after prediction. This covariant construction, combined with the graph-theoretic fragmentation sum (Eq. 13) and a mini-batch k-means tessellation that selects only 10-20% of reference frames, yields full-system force errors below 0.06 mEh/Bohr for H13O6+. The resulting fully ML-predicted AIMD trajectory reproduces O-O and O-H radial distributions and the velocity autocorrelation power spectrum of the reference, indicating

Load-bearing premise

The load-bearing premise is that the paper's graph-based recipe for cutting the molecule into overlapping fragments and stitching fragment CCSD forces back together exactly reproduces what a full CCSD calculation on the whole cluster would give; the paper never runs such a full calculation, so this remains untested.

Editorial extensions

If this is right

  • Coupled-cluster-quality AIMD becomes feasible for medium water clusters and similar fluxional systems that currently require DFT-level compromise.
  • Training cost for each fragment type is independent of full system size, so the approach scales to larger clusters by reusing fragment models.
  • Direct force learning sidesteps the link-atom Jacobian and multi-fragment coupling issues that arise when energy models are differentiated after covalent bond cleavage.
  • The k-means tessellation means 80-90% of expensive CCSD fragment reference calculations can be skipped without degrading trajectory statistics.
  • Because the reference trajectories themselves use graph-fragmented CCSD rather than full-system CCSD, the method is calibrated to the fragmentation approximation, so its practical accuracy inherits that approximation's fidelity.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the graph-fragmentation reference (Eq. 10) is a faithful surrogate for full CCSD on fluxional systems, the same direct-force protocol could be applied to other strongly polarized proton-transfer clusters by retraining only the relevant fragment networks.
  • The observed slow drift of the 10%-trained trajectory toward sparsely sampled regions suggests an active-learning loop—updating k-means centroids from the ML trajectory itself—as a natural next step, which the authors mention only as future work.
  • Because force components in each principal-axis block share hidden features, the method could in principle support transfer learning between fragments of different protonation states; the paper leaves this implicit.
  • A decisive external check would compare the graph-fragmented CCSD reference against full-system CCSD gradients at several off-equilibrium configurations; the paper does not perform this comparison, so the 'CCSD accuracy' label should be read as accuracy relative to its own fragmentation reference.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. This paper introduces a machine-learning framework for direct prediction of post-Hartree-Fock nuclear forces aimed at enabling coupled-cluster-level AIMD. The method combines graph-theoretic fragmentation with fragment-level CCSD/B3LYP corrections, a principal-axis-of-inertia covariant descriptor, and vector-valued neural networks that predict force components directly instead of differentiating a learned energy surface. Training configurations are selected by mini-batch k-means tessellation. The approach is validated on the solvated Zundel cation H13O6+, using a 9326-frame reference trajectory generated from graph-fragmented CCSD/B3LYP at rank R=1. Reported fragment-level force errors are below 0.02 mEh/Bohr for most fragment types, full-system errors for the production dynamic rank-1 protocol are about 0.056 mEh/Bohr, and ML-generated trajectories are compared with the reference through radial distribution functions and velocity autocorrelation spectra.

Significance. If the central accuracy claim were fully supported, the combination of graph-theoretic fragmentation and direct vector-valued force learning would be a practically useful step toward correlated-level AIMD for medium-sized water clusters, with a notable reduction in trainable parameters. The paper has clear strengths: explicit vector-valued force training with shared basis functions, the k-means-based space tessellation for data selection, and systematic error decomposition by fragment type, rank, and geometry class. However, the headline 'CCSD accuracy' is not yet established relative to full-system CCSD, and several validation details—train/test separation, trajectory agreement metrics, and the handling of principal-axis sign degeneracy—are missing. These issues affect the paper's main claims but are addressable with additional analysis and reframing.

major comments (5)
  1. [Introduction; §IV.C, Table II] The central claim of '10 micro-Hartree/Bohr accuracy with respect to CCSD' is not supported by the reported comparisons. Table II gives the dynamic rank-1 graph error for ∂∆E/∂x as 0.056 mEh/Bohr (56 µEh/Bohr), an order of magnitude larger than 10 µEh/Bohr. The full-system ML errors at the same cutoff are 0.056–0.057 mEh/Bohr for all training fractions, meaning the ML model reproduces the graph-fragmented reference but does not correct its error. If the 'exact value' axis in Fig. 7 is a full-system CCSD/B3LYP gradient, the total ML force deviation from CCSD is at least the graph error; if it is not, the manuscript never validates against a full-system CCSD force. Please state the exact reference used, report the total error against that reference, and revise the abstract/Introduction claims accordingly.
  2. [§IV.B, Eq. (24); Table I] No train/test split is described for Eq. (24). The text says models are trained on k-means-selected 10% of fragment geometries and then errors are computed over NConfig fragments, but it is not stated whether the evaluation set is disjoint from the training set. If the same configurations are used for training and evaluation, the errors in Table I and Figures 5–6 are in-sample and optimistic. Also, Table I is internally inconsistent: H2O has NConfig=8910 and training size 3891 (~44%), whereas all other fragment types use ~10%. Please clarify the data-selection procedure, report held-out errors, and correct the table.
  3. [§V.A, Figures 10–11; Abstract] The abstract claims that using only 10–20% of reference configurations yields successful fully predicted trajectories, but Section V.A reports that the 10%-training trajectory shrinks the second RDF peak near 2.7 Å and broadens the distribution near 4 Å, and that only the 40%-training trajectory aligns well with the reference. This directly qualifies the headline efficiency claim. Please provide quantitative agreement measures (e.g., integrated absolute differences or peak positions/heights) for the 10%, 20%, and 40% models and state which training fraction supports each claim.
  4. [§III.A, steps 1–6] The principal-axis descriptor is central to the covariant force representation, but the manuscript does not specify how the arbitrary signs of inertia eigenvectors and possible axis degeneracy (near-symmetric fragments) are handled. Without a canonical sign convention or a treatment of axis switching, the fragment-fixed frame can flip discontinuously between nearby configurations, producing discontinuous force predictions and trajectories. Please document the sign/order resolution and assess its effect on force continuity, especially for the fluxional water fragments studied here.
  5. [§III.B, Eq. (22); §V] Because forces are trained directly as vector outputs, the predicted force field is not guaranteed to be conservative (curl-free). The paper does not report energy conservation along the ML-driven trajectories, which is a standard check for AIMD. For NVE simulations, non-conservative forces can lead to systematic drift and biased distributions over longer times, directly relevant to the 'long-timescale' goal. Please report total-energy drift for the predicted trajectories or explain why non-conservativity is not a concern at the simulated time scales.
minor comments (5)
  1. [Various] Typos and unclear wording: 'machine trajectories' (Section V opener), 'distraibution' (Fig. 10 caption), 'calcualtions' (p.10), 'aa manifested' (p.8), and 'th-milli-Hartree/Bohr' (Section IV.B) are confusing.
  2. [Table II] The caption is truncated: 'Errors in fragment forces as per' appears to be incomplete. Please provide a complete caption including the reference for the error.
  3. [References] References [37] and [44] appear to be the same paper (Zhu and Iyengar, Phys. Rev. X 2026, 16, 011012). Please consolidate or distinguish them.
  4. [§IV.B] The statement that errors below 0.02 mEh/Bohr are 'well within' 10 micro-Hartree/Bohr is arithmetically inconsistent (0.02 mEh/Bohr = 20 µEh/Bohr). Please correct the convergence-criterion comparison or the reported accuracy claim.
  5. [§V] Details of the ML-driven trajectories (integration scheme, time step, thermostat if any, length) are not given, which hinders reproducibility of the AIMD validation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: fragment-level CCSD labels are external inputs and the full-system force prediction is a learned surrogate for an independently defined graph-theoretic target.

full rationale

The derivation chain is not circular. Fragment energy/force labels are generated from CCSD and B3LYP calculations on fragment subsystems via Eq. (4), an external quantum-chemistry input independent of the ML model. Training-set selection uses unsupervised k-means tessellation over these labels, and test errors (Table I, Eq. (24)) measure ML fragment predictions against these external fragment CCSD values. Full-system forces are assembled by the fixed inclusion-exclusion sum Eq. (22), whose exact counterpart Eq. (10) is separately benchmarked in Figure 7/Table II against 'the reference full system force differences evaluated at the CCSD and B3LYP levels of theory.' Thus the ML output is not, by construction, equal to its input; it is a learned surrogate for an externally defined target. Two limitations should be distinguished from circularity: (i) the AIMD validation reference ('All ES') is itself generated 'using forces in Eq. (10)', so dynamical agreement validates the ML surrogate against the graph-theoretic reference, not against a full-system CCSD trajectory; the 'CCSD accuracy' label for trajectories is therefore inherited. (ii) The Introduction's '10 micro-Hartree/Bohr accuracy' claim is not supported by Table II, where the dynamic rank-1 graph and ML full-system errors are 0.056-0.057 mEh/Bohr (56-57 micro-Eh/Bohr). Both are accuracy/benchmarking concerns rather than circular reductions.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced. The free parameters and axioms above capture the tunable choices and modeling assumptions the central claim depends on, several of which are inherited from the authors' prior fragmentation papers rather than independently established here.

free parameters (4)
  • k (number of k-means clusters) per fragment type
    Mini-batch k-means tessellation (Section IV A) is central to constructing 10-40% training sets, but k and the distance metric are not specified; they are hand-chosen hyperparameters.
  • Graph construction cutoffs = H-O <1.4 Å; O-O edge <1.1*D_i; max 3 edges per node
    These geometric thresholds define fragment nodes and edges (Section IV). Results depend on them, and no sensitivity analysis is provided.
  • Training fraction per fragment type = 10%, 20%, 40%
    Accuracy depends strongly on this fraction (Table II: R=3 ML error 0.132 at 10% vs 0.032 at 40%); the abstract's 10-20% claim is not the setting that gives the best trajectory.
  • Neural network architecture
    Equations (14)-(18) define the network form, but layer widths, number of hidden layers, nonlinearities, optimizer, epochs, and weight initialization are not reported. m is the parameter-count driver in Figure 3.
assumptions (5)
  • domain assumption The graph-theoretic fragmentation/inclusion-exclusion formula Eq. (2)-(5) converges to the full-system CCSD correction for H13O6+.
    Invoked throughout Sections II and IV; the paper validates ML against this fragmented reference rather than full CCSD.
  • domain assumption The reference trajectory generated by Eq. (10) with dynamic rank-1 cutoff is an adequate proxy for full CCSD AIMD.
    Used as ground truth for all trajectory-comparison figures (Figures 10-12). No full CCSD trajectory is computed.
  • domain assumption Principal axes of inertia are non-degenerate and vary continuously enough for the covariant frame and atomic ordering to be well-defined on all training/inference fragments.
    Required by the descriptor construction in Section III A; eigenvalue degeneracies and sign ambiguities are not discussed.
  • domain assumption B3LYP/6-31+g(d,p) is a sufficiently systematic low level that delta-learning converges with 10-40% of fragment data.
    Underlies the delta-ML corrections in Eq. (6)-(9); no convergence study versus basis set or functional is presented.
  • domain assumption Ergodicity assumed in the velocity autocorrelation Fourier analysis Eq. (25) over a 1.86 ps trajectory.
    The ensemble average is replaced by a time integral; short trajectory length makes this questionable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph-Theoretic Neural Network Fragmentation with Covariant Direct Molecular Force Learning: Enabling Coupled-Cluster Accuracy AIMD for Fluxional Systems." pith.science (2026). https://pith.science/paper/6T6M4SNE

@misc{pith2026260721779,
  author       = {Pith},
  title        = {Pith review of: Graph-Theoretic Neural Network Fragmentation with Covariant Direct Molecular Force Learning: Enabling Coupled-Cluster Accuracy AIMD for Fluxional Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6T6M4SNE}},
  note         = {Machine review of arXiv:2607.21779}
}
read the original abstract

Accurate ab initio molecular dynamics (AIMD) simulations of complex, fluxional chemical systems are severely limited by the high computational scaling of correlated electronic structure methods. To overcome this bottleneck, we present a robust, graph-theoretic molecular fragmentation framework integrated with machine learning to directly model post-Hartree-Fock nuclear forces at coupled cluster accuracy. Bypassing the limitations of automatic differentiation on learned energy surfaces that may struggle with link-atom Jacobians, our approach directly predicts nuclear force vectors. By projecting these vectors onto fragment-fixed principal axes of inertia, we establish co-variant descriptors that naturally preserve rotational, translational, and permutational invariance. The methodology achieves exceptional high parameter efficiency through a vector-valued training protocol that reduces trainable parameters by over an order of magnitude, while an unsupervised mini-batch k-means space tessellation algorithm constructs highly representative training databases using only 10% to 20% of reference configurations. We rigorously validated this framework on the highly fluxional solvated Zundel cation H_{13}O_6^+ ). Our fully machine-learning-predicted AIMD trajectories successfully reproduced complex dynamical signatures and key structural characteristics, including radial distribution functions and the velocity autocorrelation power spectrum. Ultimately, this scalable, systematically improvable framework bridges the gap between high-level correlated wavefunction theories and long-timescale reactive sampling, laying the foundation for advanced, LLM-inspired transfer learning in modern chemical dynamics simulations.

Figures

Figures reproduced from arXiv: 2607.21779 by the authors.

Figure 1
Figure 1. FIG. 1. The three principle mass moment of inertia axis, [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. A demonstration of architecture of neural network [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 4
Figure 4. FIG. 4. Distribution functions that determine the range of [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figures from the paper (10 more)
Figure 5
Figure 5. Figure 5: FIG. 5. Error as shown in Eq. ( [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Distribution of fragment force errors. Given the [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Full system force component and graph force compo [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 9
Figure 9. Figure 9: FIG. 9. Increasing training set size systematically improve [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: FIG. 10. Oxygen-oxygen distance and oxygen-oxygen-oxygen [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: FIG. 11. Oxygen-hydrogen distance distribution comparison [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 13
Figure 13. Figure 13: FIG. 13. Relative computational complexity for training sets, [PITH_FULL_IMAGE:figures/full_fig_p014_13.png]
Figure 14
Figure 14. Figure 14: FIG. 14. Distribution of fragment force errors for the given [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: FIG. 15. Distribution of fragment force errors for the given [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]
Figure 16
Figure 16. Figure 16: FIG. 16. Distribution of fragment force errors for the given [PITH_FULL_IMAGE:figures/full_fig_p017_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

94 extracted references · 1 linked inside Pith

  1. [1]

    Wang, I. S. Y.; Karplus, M. Dynamics of Organic Reac- tions. J. Amer. Chem. Soc.1973, 95, 8160

    1973

  2. [2]

    Classical Trajectories Using the Full Ab Initio Potential Energy Surface H −+CH4 →CH 4+H−

    Leforestier, C. Classical Trajectories Using the Full Ab Initio Potential Energy Surface H −+CH4 →CH 4+H−. J. Chem. Phys.1978, 68, 4406

    1978

  3. [3]

    Unified Approach for Molecular Dynamics and Density-Functional Theory

    Car, R.; Parrinello, M. Unified Approach for Molecular Dynamics and Density-Functional Theory. Phys. Rev. Lett.1985, 55, 2471. 18

    1985

  4. [4]

    B.; Millam, J

    Schlegel, H. B.; Millam, J. M.; Iyengar, S. S.; Voth, G. A.; Daniels, A. D.; Scuseria, G. E.; Frisch, M. J. Ab Initio Molecular Dynamics: Propagating the Density Matrix with Gaussian Orbitals. J. Chem. Phys.2001, 114, 9758

    2001

  5. [5]

    Recent Advances and Perspectives on Nonadiabatic Mixed Quantum-Classical Dynamics

    Crespo-Otero, R.; Barbatti, M. Recent Advances and Perspectives on Nonadiabatic Mixed Quantum-Classical Dynamics. Chem. Rev.2018, 118, 7026

    2018

  6. [6]

    Escaping Free-Energy Minima

    Laio, A.; Parrinello, M. Escaping Free-Energy Minima. Proc. Natl. Acad. Sci. USA2002, 99, 12562

  7. [7]

    Efficient Explo- ration of Reactive Potential Energy Surfaces Using Car- Parrinello Molecular Dynamics

    Iannuzzi, M.; Laio, A.; Parrinello, M. Efficient Explo- ration of Reactive Potential Energy Surfaces Using Car- Parrinello Molecular Dynamics. Phys. Rev. Lett.2003, 90, 238302

    2003

  8. [8]

    Phatak, P.; Sumner, I.; Iyengar, S. S. Gauging the Flexibility of the Active Site in Soybean Lipoxygenase- 1 (SLO-1) Through an Atom-Centered Density Matrix Propagation (ADMP) Treatment That Facilitates the Sampling of Rare Events. J. Phys. Chem. B2012, 116, 10145

Show all 94 references
  1. [9]

    Phatak, P.; Venderley, J.; Debrota, J.; Li, J.; Iyen- gar, S. S. Active Site Dynamical Effects That Fa- cilitate the Hydrogen Transfer Process in Soybean Lipoxygenase-1 (SLO-1): Isotope Effects. J. Phys. Chem. B2015, 119, 9532

  2. [10]

    Gordon, R. G. Correlation Functions for Molecular Mo- tion. Adv. Mag. Resonance1968, 3, 1

  3. [11]

    McQuarrie, D. A. Statistical Mechanics; University Sci- ence Books, Sausalito, CA, 2000

    2000

  4. [12]

    S.; Petersen, M

    Iyengar, S. S.; Petersen, M. K.; Day, T. J.; Burn- ham, C. J.; Teige, V. E.; Voth, G. A. The properties of ion-water clusters. I. The protonated 21-water cluster. J. Chem. Phys.2005, 123, 084309

    2005

  5. [13]

    Iyengar, S. S. Dynamical effects on vibrational and elec- tronic spectra of hydroperoxyl radical water clusters. J. Chem. Phys.2005, 123, 084310

    2005

  6. [14]

    Iyengar, S. S. Further analysis of the dynamically aver- aged vibrational spectrum for the “magic” protonated 21-water cluster. J. Chem. Phys.2007, 126, 216101

    2007

  7. [15]

    T.; Iyengar, S

    Li, X.; Moore, D. T.; Iyengar, S. S. Insights from first principles molecular dynamics studies toward infrared multiple-photon and single-photon action spectroscopy: Case study of the proton-bound dimethyl ether dimer. J. Chem. Phys.2008, 128, 184308

    2008

  8. [16]

    R.; Moore, D

    Li, X.; Oomens, J.; Eyler, J. R.; Moore, D. T.; Iyen- gar, S. S. Isotope dependent, temperature regulated, en- ergy repartitioning in a low-barrier, short-strong hydro- gen bonded cluster. J. Chem. Phys.2010, 132, 244301

    2010

  9. [17]

    M.; Iyengar, S

    Dietrick, S. M.; Iyengar, S. S. Constructing Periodic Phase Space Orbits from Ab Initio Molecular Dynam- ics Trajectories to Analyze Vibrational Spectra: Case Study of the Zundel (H 5O+ 2 ) Cation. J. Chem. Theory and Comput.2012, 8, 4876

    2012

  10. [18]

    B.; Raghavachari, K.; Iyengar, S

    Li, J.; Pacheco, A. B.; Raghavachari, K.; Iyengar, S. S. A Grotthuss-like proton shuttle in the anomalousC 2H + 3 carbocation: energetic and vibrational properties for iso- topologues. Phys. Chem. Chem. Phys.2016, 18, 29395– 29411

    2016

  11. [19]

    M.; Iyengar, S

    Sager, L. M.; Iyengar, S. S. Proton relays in anomalous carbocations dictate spectroscopy, stability, and mecha- nisms: case studies on C2H5+ and C3H3+. Phys. Chem. Chem. Phys.2017, 19, 27801–27816

    2017

  12. [20]

    E.; Iyengar, S

    Li, X.; Teige, V. E.; Iyengar, S. S. Can the four- coordinated, penta-valent oxygen in hydroxide water clusters be detected through experimental vibrational spectroscopy? J. Phys. Chem. A2007, 111, 4815–4820

  13. [21]

    Li, X.; Iyengar, S. S. Quantum Wavepacket Ab Ini- tio Molecular Dynamics for Extended Systems. J. Phys. Chem. A2011, 115, 6269

  14. [22]

    S.; Day, T

    Iyengar, S. S.; Day, T. J.; Voth, G. A. On the amphiphilic behavior of the hydrated proton: anab initiomolecu- lar dynamics study. International Journal of Mass Spec- trometry2005, 241, 197–204

  15. [23]

    The Quest for a Universal Den- sity Functional: The Accuracy of Density Functionals Across a Broad Spectrum of Databases in Chemistry and Physics

    Peverati, R.; Truhlar, D. The Quest for a Universal Den- sity Functional: The Accuracy of Density Functionals Across a Broad Spectrum of Databases in Chemistry and Physics. Philos. Trans. Royal Soc. A2014, 372, 10120476

  16. [24]

    Perspective: Advances and challenges in treating van der Waals dispersion forces in density functional theory

    Klimes, J.; Michaelides, A. Perspective: Advances and challenges in treating van der Waals dispersion forces in density functional theory. J. Chem. Phys.2012, 137, 120901

    2012

  17. [25]

    Sadhukhan, S.; Munoz, D.; Adamo, C.; Scuseria, G. E. Predicting Proton Transfer Barriers with Density Func- tional Methods. Chem. Phys. Lett.1999, 306, 83

    1999

  18. [26]

    B.; Schaefer, H

    Xie, Y.; Remington, R. B.; Schaefer, H. F., III The Pro- tonated Water Dimer: Extensive Theoretical Studies of H5O+ 2 . J. Chem. Phys.1994, 101, 4878

    1994

  19. [27]

    J.; Mori-S´ anchez, P.; Yang, W

    Cohen, A. J.; Mori-S´ anchez, P.; Yang, W. Challenges for Density Functional Theory. Chem. Revs.2012, 112, 289

    2012

  20. [28]

    E.; Poltavsky, I.; Schuett, K

    Chmiela, S.; Tkatchenko, A.; Sauceda, H. E.; Poltavsky, I.; Schuett, K. T.; Mueller, K.-R. Machine learning of accurate energy-conserving molecular force fields. SCIENCE ADV ANCES2017, 3

  21. [29]

    B.; Batzner, S.; Xie, Y.; Sun, L.; Kolpak, A

    Vandermause, J.; Torrisi, S. B.; Batzner, S.; Xie, Y.; Sun, L.; Kolpak, A. M.; Kozinsky, B. On-the-fly active learning of interpretable Bayesian force fields for atom- istic rare events. NPJ COMPUTATIONAL MATERI- ALS2020, 6

  22. [30]

    P.; Simm, G

    Batatia, I.; Kovacs, D. P.; Simm, G. N. C.; Ort- ner, C.; Csanyi, G. MACE: Higher Order Equivariant Message Passing Neural Networks for Fast and Accurate Force Fields. ADV ANCES IN NEURAL INFORMA- TION PROCESSING SYSTEMS 35, NEURIPS 2022. Koyejo, S., Mohamed, S., Agarwal, A., ...

    2022

  23. [31]

    Generalized neural-network representation of high-dimensional potential-energy sur- faces

    Behler, J.; Parrinello, M. Generalized neural-network representation of high-dimensional potential-energy sur- faces. Phys. Rev. Lett.2007, 98

    2007

  24. [32]

    DeePMD-kit: A deep learning package for many-body potential energy representation and molecular dynamics

    Wang, H.; Zhang, L.; Han, J.; E, W. DeePMD-kit: A deep learning package for many-body potential energy representation and molecular dynamics. COMPUTER PHYSICS COMMUNICATIONS2018, 228, 178–184

  25. [33]

    S.; Isayev, O.; Roitberg, A

    Smith, J. S.; Isayev, O.; Roitberg, A. E. ANI-1: an ex- tensible neural network potential with DFT accuracy at force field computational cost. CHEMICAL SCIENCE 2017, 8, 3192–3203

    2017

  26. [34]

    Four Generations of High-Dimensional Neu- ral Network Potentials

    Behler, J. Four Generations of High-Dimensional Neu- ral Network Potentials. Chem. Rev.2021, 121, 10037– 19 10072

    2021

  27. [35]

    Automated Fitting of Neural Network Potentials at Coupled Cluster Accuracy: Protonated Water Clusters as Testing Ground

    Schran, C.; Behler, J.; Marx, D. Automated Fitting of Neural Network Potentials at Coupled Cluster Accuracy: Protonated Water Clusters as Testing Ground. J. Chem. Theory Comput.2020, 16, 88–99

    2020

  28. [36]

    Transferability of ma- chine learning potentials: Protonated water neural net- work potential applied to the protonated water hexamer

    Schran, C.; Brieuc, F.; Marx, D. Transferability of ma- chine learning potentials: Protonated water neural net- work potential applied to the protonated water hexamer. J. Chem. Phys.2021, 154

    2021

  29. [38]

    S.; Riley, P

    Gilmer, J.; Schoenholz, S. S.; Riley, P. F.; Vinyals, O.; Dahl, G. E. Neural Message Passing for Quantum Chem- istry. INTERNATIONAL CONFERENCE ON MA- CHINE LEARNING, VOL 70. Precup, D., Teh, Y., Eds.; Proceedings of Machine Learning Research; 2017; Vol. 70; 34th International ...

    2017

  30. [39]

    S.; Freeman, W

    Yedidia, J. S.; Freeman, W. T.; Weiss, Y. Understanding Belief Propagation and Its Generalizations. In Exploring Artificial Intelligence in the New Millennium; Morgan Kaufmann Publishers Inc.: San Francisco, CA, USA, 2003; p 239–269

    2003

  31. [40]

    C.; Zhu, X.; Iyengar, S

    Ricard, T. C.; Zhu, X.; Iyengar, S. S. Capturing weak interactions in surface adsorbate systems at coupled clus- ter accuracy: a graph-theoretic molecular fragmentation approach improved through machine learning. J. Chem. Theory Comput.2023, 19, 8541

    2023

  32. [41]

    C.; Haycraft, C.; Iyengar, S

    Ricard, T. C.; Haycraft, C.; Iyengar, S. S. Adaptive, geometric networks for efficient coarse-grainedab initio molecular dynamics with post-Hartree-Fock accuracy. J. Chem. Theory Comput.2018, 14, 2852

    2018

  33. [42]

    C.; Iyengar, S

    Ricard, T. C.; Iyengar, S. S. An efficient and accurate ap- proach to estimate hybrid functional and large basis set contributions to condensed phase systems and molecule- surface interactions. J. Chem. Theory Comput.2020, 16, 4790

    2020

  34. [43]

    Zhu, X.; Iyengar, S. S. Graph Theoretic Molecular Frag- mentation for Multidimensional Potential Energy Sur- faces Yield an Adaptive and General Transfer Machine Learning Protocol. J. Chem. Theory Comput.2022, 18, 5125–5144

    2022

  35. [44]

    Zhu, X.; Iyengar, S. S. Large Language Model-Type Ar- chitecture for High-Dimensional Molecular Potential En- ergy Surfaces. Phys. Rev. X2026, 16, 011012

  36. [45]

    N.; Kaiser, L

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, L. u.; Polosukhin, I. Attention is All you Need. 31st Conference on Neu- ral Information Processing Systems (NIPS 2017), Long Beach, CA, USA. Guyon, I., Luxburg, U. V., Bengio, S., Wallach, H...

    2017

  37. [46]

    S.; Ricard, T

    Iyengar, S. S.; Ricard, T. C.; Zhu, X. A reformulation of all ONIOM-type molecular fragmentation approaches using graph theory-based projection operators: Applica- tions to dynamics, molecular potential surfaces, and ma- chine learning and quantum computing. J. Phys. Chem. A20...

  38. [47]

    S.; Saha, D.; Dwivedi, A.; Lopez-Ruiz, M

    Iyengar, S. S.; Saha, D.; Dwivedi, A.; Lopez-Ruiz, M. A.; Kumar, A.; Zhang, J. H.; Ricard, T. C.; Richerme, P.; Sabry, A. Quantum Algorithms for the Study of Elec- tronic Structure and Molecular Dynamics: Novel Com- putational Protocols. In Comprehensive Computational Chemistr...

    2023

  39. [48]

    Li, J.; Iyengar, S. S. Ab initio Molecular Dynamics using Recursive, Spatially Separated, Overlapping Model Sub- systems Mixed Within an ONIOM Based Fragmentation Energy Extrapolation Technique. J. Chem. Theory Com- put.2015, 11, 3978–3991

    2015

  40. [49]

    Li, J.; Haycraft, C.; Iyengar, S. S. Hybrid, Extended La- grangian – Born-OppenheimerAb InitioMolecular Dy- namics using Fragment-Based Electronic Structure. J. Chem. Theory Comput.2016, 12, 2493

    2016

  41. [50]

    On-the-fly

    Haycraft, C.; Li, J.; Iyengar, S. S. Efficient, “On-the-fly” Born–Oppenheimer and Car–Parrinello–type Dynamics with coupled cluster accuracy through Fragment Based Electronic Structure. J. Chem. Theory Comput.2017, 13, 21887

    2017

  42. [51]

    on-the-fly

    Ricard, T. C.; Iyengar, S. S. Efficiently capturing weak interactions inab initiomolecular dynamics through “on-the-fly” basis set extrapolation. J. Chem. Theory Comput.2018, 14, 5535

    2018

  43. [52]

    Kumar, A.; Iyengar, S. S. Fragment-based electronic structure for potential energy surfaces using a super- position of fragmentation topologies. J. Chem. Theory Comput.2019, 15, 5769

    2019

  44. [53]

    C.; Kumar, A.; Iyengar, S

    Ricard, T. C.; Kumar, A.; Iyengar, S. S. Embedded, graph-theoretically defined many-body approximations for wavefunction-in-DFT and DFT-in-DFT: Applica- tions to gas- and condensed-phase ab initio molecular dynamics, and potential surfaces for quantum nuclear effects. Int. J. ...

    2020

  45. [54]

    Kumar, A.; DeGregorio, N.; Iyengar, S. S. Graph- Theory-Based Molecular Fragmentation for Efficient and Accurate Potential Surface Calculations in Multiple Di- mensions. J. Chem. Theory Comput.2021, 17, 6671– 6690

    2021

  46. [55]

    H.; Ricard, T

    Zhang, J. H.; Ricard, T. C.; Haycraft, C.; Iyen- gar, S. S. Weighted-Graph-Theoretic Methods for Many- Body Corrections within ONIOM: Smooth AIMD and the Role of High-Order Many-Body Terms. J. Chem. Theory Comput.2021, 17, 2672–2690

    2021

  47. [56]

    H.; Iyengar, S

    Zhang, J. H.; Iyengar, S. S. Graph-|Q⟩ ⟨C|: A Graph- based Quantum-classical algorithm for efficient elec- tronic structure on hybrid quantum/classical hardware systems: Improved quantum circuit depth performance. J. Chem. Theory Comput.2022, 18, 2885

    2022

  48. [57]

    Kumar, A.; DeGregorio, N.; Ricard, T.; Iyengar, S. S. Graph-Theoretic Molecular Fragmentation for Potential Surfaces Leads Naturally to a Tensor Network Form and Allows Accurate and Efficient Quantum Nuclear Dynam- ics. J. Chem. Theory Comput.2022, 18, 7243

    2022

  49. [58]

    S.; Zhang, J

    Iyengar, S. S.; Zhang, J. H.; Saha, D.; Ricard, T. C. Graph-|Q⟩ ⟨C|: A Quantum Algorithm with Reduced Quantum Circuit Depth for Electronic Structure. J. Phys. Chem. A2023, 127, 9334

  50. [59]

    K.; Shah, N

    Dey, T. K.; Shah, N. R. On the number of simplicial complexes in Rd. Comput. Geom.1997, 8, 267

    1997

  51. [60]

    C.; Franzosa, R

    Adams, C. C.; Franzosa, R. D. Introduction to topology: pure and applied; 2008

    2008

  52. [61]

    20 Affine spaces

    Berger, M.; Pansu, P.; Berry, J.-P.; Saint-Raymond, X. 20 Affine spaces. In Problems in Geometry; Springer, 1984; p 11

    1984

  53. [62]

    Armstrong, M. A. Basic topology; Springer Science & Business Media, 2013

    2013

  54. [63]

    IMOMM: A new integrated ab initio + molecular mechanics geometry optimization scheme of equilibrium structures and transition states

    Maseras, F.; Morokuma, K. IMOMM: A new integrated ab initio + molecular mechanics geometry optimization scheme of equilibrium structures and transition states. J. Comput. Chem.1995, 16, 1170

    1995

  55. [64]

    L.; Conte, R.; Bow- man, J

    Nandi, A.; Qu, C.; Houston, P. L.; Conte, R.; Bow- man, J. M. ∆-machine learning for potential energy sur- faces: A PIP approach to bring a DFT-based PES to CCSD(T) level of theory. J. Chem. Phys.2021, 154, 051102

    2021

  56. [65]

    O.; Rupp, M.; von Lilien- feld, O

    Ramakrishnan, R.; Dral, P. O.; Rupp, M.; von Lilien- feld, O. A. Big Data Meets Quantum Chemistry Approx- imations: The ∆-Machine Learning Approach. J. Chem. Theory Comput.2015, 11, 2087–2096

    2015

  57. [66]

    ∆-Machine learning- driven discovery of double hybrid organic–inorganic per- ovskites

    Chen, J.; Xu, W.; Zhang, R. ∆-Machine learning- driven discovery of double hybrid organic–inorganic per- ovskites. J. Mater. Chem. A2022, 10, 1402–1413

  58. [67]

    A.; Greenwell, C

    Unzueta, P. A.; Greenwell, C. S.; Beran, G. J. O. Predict- ing Density Functional Theory-Quality Nuclear Mag- netic Resonance Chemical Shifts via ∆-Machine Learn- ing. J. Chem. Theory Comput.2021, 17, 826–840

    2021

  59. [68]

    L.; Qu, C.; Yu, Q.; Conte, R.; Tkatchenko, A.; Bowman, J

    Nandi, A.; Pandey, P.; Houston, P. L.; Qu, C.; Yu, Q.; Conte, R.; Tkatchenko, A.; Bowman, J. M. ∆-Machine Learning to Elevate DFT-Based Potentials and a Force Field to the CCSD(T) Level Illustrated for Ethanol. J. Chem. Theory Comput.2024, 20, 8807–8819

    2024

  60. [69]

    Neural Machine Translation by Jointly Learning to Align and Translate

    Bahdanau, D.; Cho, K.; Bengio, Y. Neural Machine Translation by Jointly Learning to Align and Translate. 2016;https://arxiv.org/abs/1409.0473

    2016 arXiv

  61. [70]

    S.; Voth, G

    Rega, N.; Iyengar, S. S.; Voth, G. A.; Schlegel, H. B.; Vreven, T.; Frisch, M. J. Hybrid Ab-Initio/Empirical Molecular Dynamics: Combining the ONIOM Scheme with the Atom-Centered Density Matrix Propagation (ADMP) Approach. J. Phys. Chem. B2004, 108, 4210

  62. [71]

    Hydration and Intermolecular Interaction: Infrared Investigations with Polyelectrolyte Membranes; Academic Press, 2012

    Zundel, G. Hydration and Intermolecular Interaction: Infrared Investigations with Polyelectrolyte Membranes; Academic Press, 2012

    2012

  63. [72]

    E.; Ungar, P

    Tuckerman, M. E.; Ungar, P. J.; Vonrosenvinge, T.; Klein, M. L. Ab Initio Molecular Dynamics Simulations. J. Phys. Chem.1996, 100, 12878

    1996

  64. [73]

    Tuckerman, M.; Laasonen, K.; Sprik, M.; Parrinello, M. J. Phys. Chem.1995, 99, 5749

    1995

  65. [74]

    W.; Voth, G

    Schmitt, U. W.; Voth, G. A. The computer simulation of proton transport in water. J. Chem. Phys.1999, 111, 9361–9381

    1999

  66. [75]

    W.; Voth, G

    Schmitt, U. W.; Voth, G. A. Multistate empirical va- lence bond model for proton transport in water. J. Phys. Chem. B1998, 102, 5547–5551

  67. [76]

    Dynamics and Infrared Spectroscopy of the Protonated Water Dimer

    Vendrell, O.; Gatti, F.; Meyer, H.-D. Dynamics and Infrared Spectroscopy of the Protonated Water Dimer. Ang. Chem. Intl. Ed.2007, 46, 6918

    2007

  68. [77]

    Structure and Dynamics of a Proton Wire: A Theoretical Study ofH + Translocation Along the Single-File Water Chain in the Gramicidin a Chan- nel

    Pomes, R.; Roux, B. Structure and Dynamics of a Proton Wire: A Theoretical Study ofH + Translocation Along the Single-File Water Chain in the Gramicidin a Chan- nel. Biophys J.1996, 71, 19

    1996

  69. [78]

    J.; Phillips, L

    McEwan, M. J.; Phillips, L. F. Chemistry of the Atmo- sphere; Eward Arnold: London, 1975

    1975

  70. [79]

    Wayne, R. P. Chemistry of the Atmosphere; Clarendon Press: Oxford, 1994

    1994

  71. [80]

    Theoretical study of H+ transloca- tion along a model proton wire

    Pomes, R.; Roux, B. Theoretical study of H+ transloca- tion along a model proton wire. J. Phys. Chem.1996, 100, 2519–2527

    1996

  72. [81]

    Solva- tion and Hydrogen-Bonding Effects on Proton Wires

    Decornez, H.; Drukker, K.; Hammes-Schiffer, S. Solva- tion and Hydrogen-Bonding Effects on Proton Wires. J. Phys. Chem. A1999, 103, 2891

  73. [82]

    L.; Schmitt, U

    Brewer, M. L.; Schmitt, U. W.; Voth, G. A. The For- mation and Dynamics of Proton Wires in Channel En- vironments. Biophys J.2001, 80, 1691

    2001

  74. [83]

    Water structure of a hydrophobic protein at atomic resolution: Pentagon rings of water molecules in crystals of crambin

    Teeter, M. Water structure of a hydrophobic protein at atomic resolution: Pentagon rings of water molecules in crystals of crambin. Proceedings of the National Academy of Sciences1984, 81, 6014–6018

  75. [84]

    M.; Shieh, H

    Neidle, S.; Berman, H. M.; Shieh, H. S. Highly Struc- tured Water Network in Crystals of a Deoxydinucleoside- Drug Complex. Nature1980, 288, 129

  76. [85]

    A.; Peek, M

    Lipscomb, L. A.; Peek, M. E.; Zhou, F. X.; Bertrand, J. A.; VanDerveer, D.; Williams, L. D. Wa- ter ring structure at DNA interfaces: hydration and dy- namics of DNA-anthracycline complexes. Biochemistry 1994, 33, 3649–3659

    1994

  77. [86]

    S.; Tripp, B

    Tu, C.; Rowlett, R. S.; Tripp, B. C.; Ferry, J. G.; Silver- man, D. N. Chemical rescue of proton transfer in catal- ysis by carbonic anhydrases in theβ-andγ-class. Bio- chemistry2002, 41, 15429–15435

  78. [87]

    The Grotthuss Mechanism

    Agmon, N. The Grotthuss Mechanism. Chem.Phys.Lett. 1995, 244, 456

    1995

  79. [88]

    A.; Wong, M

    Hartigan, J. A.; Wong, M. A. Algorithm AS 136: A k-means clustering algorithm. J. R. Stat. Soc. series c (applied statistics)1979, 28, 100–108

    1979

  80. [89]

    Some methods for classification and anal- ysis of multivariate observations

    MacQueen, J. Some methods for classification and anal- ysis of multivariate observations. Proceedings of the fifth Berkeley symposium on mathematical statistics and probability. 1967; Vol. 1; pp 281–297

    1967

  81. [90]

    Computing Dirichlet tessellations

    Bowyer, A. Computing Dirichlet tessellations. Comput. J.1981, 24, 162–166

    1981

  82. [91]

    Voronoi Diagrams — A survey of a fundamental geometric data structure

    Aurenhammer, F. Voronoi Diagrams — A survey of a fundamental geometric data structure. ACM Comput. Survey1991, 23, 345

  83. [92]

    Project, T. C. CGAL User and Reference Manual, 4th ed.; CGAL Editorial Board, 2018

    2018

  84. [93]

    Web-scale k-means clustering

    Sculley, D. Web-scale k-means clustering. WWW ’10: Proceedings of the 19th international conference on World wide web. New York, NY, USA, Association for Computing Machinery: New York, NY, USA, 2010; pp 1177–1178

    2010

  85. [94]

    J.; Trucks, G

    Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuse- ria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Petersson, G. A.; Nakatsuji, H.; Li, X.; Car- icato, M.; Marenich, A. V.; Bloino, J.; Janesko, B. G.; Gomperts, R.; Mennucci, B.; Hratchian, H. P.; Or- ti...

    2016

  86. [95]

    H.; Teukolsky, S

    Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flan- nery, B. P. Numerical Recipes in C; Cambridge Univer- sity Press, New York, 1992

    1992

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.