Pith. sign in

REVIEW 3 major objections 5 minor 80 references

Accelerated Machine Learning Force Field for Predicting Thermal Conductivity of Organic Liquids

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read A single machine-learning force field predicts the thermal conductivity of 20 organic liquids to about 14% average error—versus 78% for the standard OPLS-AA classical force field—provided the simulation is first aligned to experimental liqu

desk verdict A solid workflow paper with a reproducible 14% MAPE benchmark for thermal conductivity of 20 organic liquids, whose main caveat is that the density-alignment step is validated only on chemically similar molecules. read the letter →

arxiv 2512.01627 v1 pith:VLNX5K75 submitted 2025-12-01 physics.chem-ph physics.comp-ph

Wei Feng , Siyuan Liu , Hongyi Wang , Zhenliang Mu , Zhichen Pu , Xu Han , Tianze Zheng , Zhenze Yang
show 6 more authors
Zhi Wang Weihao Gao Yidan Cao Kuang Yu Sheng Gong Wen Yan
This is my paper
classification physics.chem-phphysics.comp-ph
keywords machinelearningforcefieldthermalconductivityorganicliquidsmoleculardynamicsdensityalignmentdifferentialattentionreversenon-equilibriumgraphneuralnetwork
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

The paper aims to show that thermal conductivity of organic liquids—a property central to cooling, energy storage, and chemical processing—can be predicted by a single machine-learned force field instead of by classical force fields or ab initio simulations. It introduces BAMBOO-TC, a workflow combining a graph equivariant differential transformer trained on DFT data with a density-alignment step that tunes intermolecular interaction strength to match experimental liquid densities. On a curated set of 20 alcohols, esters, and carbonates at room temperature, the workflow reaches a mean absolute percentage error of about 14% for thermal conductivity, compared with about 78% for OPLS-AA and about 25% before density alignment. The authors present this as the first demonstration that one MLFF can cover multiple organic liquids for thermal conductivity, and they accelerate the model with Triton kernels to make such simulations practical.

What carries the argument

The central object is BAMBOO-TC, a machine learning force field built on the Graph Equivariant Differential Transformer (GEDT): a graph neural network whose attention scores are computed as the difference between two attention functions, a 'differential attention' step meant to amplify signal and cancel attention noise. The other load-bearing component is density alignment, a physics-grounded calibration that adjusts intermolecular forces so that simulated liquid density matches experimental density, on the premise that correcting intermolecular distances also corrects properties governed by intermolecular interactions. Thermal conductivities are computed via reverse non-equilibrium molecula

What would settle it

Run BAMBOO-TC with and without density alignment on a molecule that shares the training functional groups but whose thermal transport is dominated by anisotropic shape or directional bonding (e.g., ethylene glycol versus ethanol) and compare the thermal conductivity error: if density alignment does not reduce the error from the ~25% pre-alignment level toward the ~14% range, the assumption that density correction fixes heat transport is falsified. A complementary check would be to apply the aligned model to a molecule with a similar density but a very different hydrogen-bonding network and see

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a single machine-learned force field—BAMBOO-TC—can simulate the thermal conductivity of multiple organic liquids with an average deviation of roughly 14%, substantially outperforming the off-the-shelf OPLS-AA classical force field at about 78% error. The key element is density alignment: using experimental density to adjust the strength of intermolecular interactions in the MLFF reduces thermal conductivity error from about 25% to about 14%, and this improvement transfers to molecules not used in the alignment, provided they are structurally similar to the training set. The model also introduces differential attention into the graph neu

Load-bearing premise

The load-bearing premise is that correcting a liquid's density—by tuning intermolecular interaction strength—also corrects its thermal conductivity; if this transferability fails for molecules outside the training distribution, the central claim loses support.

Editorial extensions

If this is right

  • If the central claim holds, machine-learned force fields become a practical alternative to classical force fields for transport properties of organic liquids, closing a gap that previously forced researchers to tune empirical potentials for each liquid.
  • Density—a cheap, widely available experimental quantity—could serve as a general calibration target for MLFF simulations of transport properties, extending the earlier observation for viscosity and ionic conductivity to thermal conductivity.
  • A single model covering a family of liquids enables screening of candidate coolants, electrolytes, and working fluids without retraining a force field for each candidate.
  • The roughly 8× inference speedup and about 3.7× end-to-end simulation speedup from Triton kernels bring MLFF-based transport calculations into a range where routine screening is feasible.
  • The explicit separation of DFT-trained, density-aligned, and zero-shot molecules provides a template for testing how far an MLFF's transferability extends before it needs new training data.

Reading between the lines

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

  • Inference: the density-alignment mechanism, if transferable, suggests a broader calibration principle—matching any easily measurable equilibrium property might correct systematic DFT errors in transport properties; the paper only tests density, so other properties remain open.
  • Inference: a natural stress test would be to apply the workflow to molecules with the same density but different thermal transport mechanisms (e.g., strongly hydrogen-bonded liquids vs. nonpolar liquids of similar density); if density alignment fails there, the claimed link between density and heat transport is not universal.
  • Inference: because the paper targets an average MAPE of 14% while experimental uncertainties are around 2%, the workflow is at present a screening tool rather than a metrology substitute; combining density alignment with other thermodynamic observables could push errors toward the experimental floor.
  • Inference: the limitation that transferability is confined to molecules resembling the training set implies the practical boundary of such models is defined by functional-group coverage; an actionable extension would be to report a coverage map or uncertainty estimate that tells users when a new molecule is outside the model's reliable regime.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper introduces BAMBOO-TC, a machine-learned force field (MLFF) based on a graph equivariant differential transformer (GEDT), and applies it to predict the thermal conductivity of 20 organic liquids via reverse non-equilibrium molecular dynamics (rNEMD). The model is trained on B3LYP-level DFT data for liquid clusters, and then refined using experimental liquid densities of seven of the benchmark molecules. The authors report a mean absolute percentage error (MAPE) of 14.14% for thermal conductivity after density alignment, compared with 24.85% before alignment and 78.01% for OPLS-AA. They also report a zero-shot subset of eight molecules not used in DFT training or density alignment, which shows a similar aggregate error. Finally, the GEDT model is re-implemented in Triton, yielding an about 3.7x speedup in full simulations.

Significance. If the result holds, this is a useful demonstration that a single MLFF can predict liquid thermal conductivity with moderate accuracy for a curated set of organic liquids, and that density alignment can partially correct systematic errors inherited from DFT. The work includes several strengths: the code is available, the benchmark data are tabulated in full, three independent simulations are used for error estimates, and a zero-shot evaluation is explicitly included. The differential attention modification and Triton acceleration are useful contributions, though they are secondary to the main quantitative claim. The density-alignment transferability question is the key scientific issue: the paper provides empirical evidence for it, but the evidence base is narrow and the paper's own text acknowledges limited structural transferability.

major comments (3)
  1. [§2.2, Table 2, and Fig. 3] The headline MAPE of 14.14% is not a fully blind evaluation. Seven of the twenty benchmark molecules (MO, EO, TBUO, EF, EA, EIB, DMC in Table S1) have their experimental densities used in the alignment step, so their thermal conductivity errors are informed by calibration on those same systems. The zero-shot group (MA, MF, PRA, IPA, DEE, PRO, EV, EH) is the only set that tests extrapolation beyond both DFT training and density alignment. Please report the aggregate and zero-shot MAPE separately in the main text and state explicitly that the headline number includes calibrated molecules. From Table S2, the zero-shot MAPE after alignment is approximately 14%, which is similar to the overall number, but the current presentation invites over-reading of the headline.
  2. [§2.1 and §2.2, Figs. 3d–3e] The transferability of density alignment to thermal conductivity is the load-bearing assumption behind the improvement from 24.85% to 14.14%. The paper's own justification in §2.1 is "we expect that aligning density between MLFF and experiment would also decrease the gap," which is an expectation, not a derived mechanism. The empirical evidence is suggestive: from Table S2 the zero-shot MAPE drops from roughly 24% to roughly 14% after alignment. However, the zero-shot set contains only eight molecules, all esters, short alcohols, or an ether, and per-molecule errors remain above 20% for two of them (EV and EH). Given the paper's own caveat that transferability is not expected for structurally different molecules, the abstract and conclusion should be rephrased to state that the demonstration is for a narrow chemical neighborhood, not a general capability. If a stronger claim is intended,
  3. [§2.1 and Supplementary Information C] The density alignment procedure is described mainly by reference to BAMBOO [25]. The text says pressure adjustments are "related to inter-molecular forces" and then used to "refine the parameters of the MLFF," but it does not specify whether the refinement produces a single global set of parameters for all liquids or per-molecule corrections. If per-molecule corrections remain, the "single MLFF" claim is weakened. Please provide the explicit alignment objective, the number of alignment rounds, and a clear statement of whether the final model is one model or a family of models. The code availability is valuable, but the paper should be self-contained on this point.
minor comments (5)
  1. [SI C vs. main text] The main text and Table S1 indicate that seven molecules are used for density alignment, while Supplementary Information C says "the alignment process encompasses 8 molecules shown in the DFT + Density Alignment set." Please reconcile this inconsistency.
  2. [References] References 14 and 15 are the same paper (Cheng and Frenkel, Phys. Rev. Lett. 125, 130602). Please remove the duplicate.
  3. [§2.2, experimental uncertainty] The statement that "the mean absolute deviation for the thermal conductivity experimental measurement is 2.21%" cites Ref. [54], which is a reference correlation for methanol. This number is not representative of the uncertainty of all the experimental values used in Table S2. Please qualify this claim.
  4. [Table 1] The GEDT/GET comparison reports RMSE/R2 values without error bars or repeated-seed uncertainty. Since the differences are modest, a statement about statistical significance would be helpful.
  5. [Introduction] The sentence "water is the only system of which has been simulated by MLFF" is too strong. Reference [61] reports MLFF-based thermal conductivity calculations for polymers. Please restrict the claim to molecular liquids or adjust the wording.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: thermal conductivity is not a fitting target, and zero-shot molecules provide independent grounding.

full rationale

The claimed prediction chain is: train the GEDT model on B3LYP DFT labels; adjust intermolecular interaction strengths to match experimental densities following the same-group BAMBOO protocol; run rNEMD simulations to compute thermal conductivity; and benchmark against experimental kappa. No step defines thermal conductivity in terms of a fitted kappa target. The density-alignment step does use experimental densities for 7 of the 20 benchmark molecules, so the aggregate 14.14% MAPE is partly informed by experimental data on those same systems; however, the fitting target is density, not thermal conductivity, and the claimed improvement is not guaranteed by construction because kappa is obtained from full molecular-dynamics heat-transport simulations. The eight zero-shot molecules (MA, MF, PRA, IPA, DEE, PRO, EV, EH) were neither DFT-trained nor density-aligned and show a similar error reduction, providing genuine external grounding. The paper's self-citations [19,25] for the density-transport transferability are motivational rather than load-bearing, since the present before/after comparison independently demonstrates the effect. The manuscript also explicitly disclaims strong transferability outside the chemical neighborhood, which is a limitation but not a circular step. Therefore no circular step meeting the evidence threshold is identified.

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

The model depends on DFT reference data, a specific MD protocol, and a density-alignment transferability assumption. The fitted density-alignment adjustments and hand-chosen hyperparameters are the main free parameters. No new physical entities are introduced.

free parameters (4)
  • density-alignment force/pressure adjustments for 7 molecules = not reported
    Used to match simulated density to experimental density for MO, EO, TBUO, EF, EA, EIB, DMC. These same molecules appear in the thermal-conductivity benchmark.
  • training loss weights (energy, force, virial) = 0.01, 0.3, 0.01
    Hand-chosen weights in Eq. S25; no sensitivity analysis is reported.
  • cutoff radius r_cut = 5 Å
    Set to 5 Å in Supplementary Eq. S4; no convergence study shown for this value.
  • number of GEDT layers = 3
    n=3 reported in Supplementary A; no optimization evidence is provided.
assumptions (5)
  • domain assumption B3LYP/def2-svpd DFT energies, forces, and virials are sufficiently accurate to train an MLFF that predicts thermal conductivity of organic liquids.
    The entire training set is built on this DFT level; no CCSD(T) or experimental validation of the reference data is provided.
  • domain assumption rNEMD with ~12,000 atoms and 2 ns production gives converged thermal conductivity for all 20 liquids at 298 K.
    The paper states size-effect tests were done but gives no details or per-system convergence evidence.
  • ad hoc to paper Density alignment transfers to thermal conductivity because correcting intermolecular interaction strengths to match experimental density also corrects heat transport.
    This is the core transferability assumption, stated as an expectation in Section 2.1, not derived or independently established for thermal conductivity in this paper.
  • domain assumption Nuclear quantum effects are negligible for these organic liquids at 298 K.
    NQE is mentioned in Section 2.1 but not quantified; it is likely weak for most of the molecules, but alcohols with hydrogen bonds could be affected.
  • domain assumption Experimental densities and thermal conductivities from handbooks and the cited literature are accurate enough to serve as ground truth.
    Many values come from reference compilations with no stated uncertainties; the paper generalizes a 2.21% experimental uncertainty from one methanol correlation to all 20 liquids.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Accelerated Machine Learning Force Field for Predicting Thermal Conductivity of Organic Liquids." pith.science (2026). https://pith.science/paper/VLNX5K75

@misc{pith2026251201627,
  author       = {Pith},
  title        = {Pith review of: Accelerated Machine Learning Force Field for Predicting Thermal Conductivity of Organic Liquids},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VLNX5K75}},
  note         = {Machine review of arXiv:2512.01627}
}
read the original abstract

The thermal conductivity of organic liquids is a vital parameter influencing various industrial and environmental applications, including energy conversion, electronics cooling, and chemical processing. However, atomistic simulation of thermal conductivity of organic liquids has been hindered by the limited accuracy of classical force fields and the huge computational demand of ab initio methods. In this work, we present a machine learning force field (MLFF)-based molecular dynamics simulation workflow to predict the thermal conductivity of 20 organic liquids. Here, we introduce the concept of differential attention into the MLFF architecture for enhanced learning ability, and we use density of the liquids to align the MLFF with experiments. As a result, this workflow achieves a mean absolute percentage error of 14% for the thermal conductivity of various organic liquids, significantly lower than that of the current off-the-shelf classical force field (78%). Furthermore, the MLFF is rewritten using Triton language to maximize simulation speed, enabling rapid prediction of thermal conductivity.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

80 extracted references · 4 canonical work pages

  1. [25]

    A predictive machine learning force-field framework for liquid electrolyte development

    Sheng Gong, Yumin Zhang, Zhenliang Mu, Zhichen Pu, Hongyi Wang, Xu Han, Zhiao Yu, Mengyi Chen, Tianze Zheng, Zhi Wang, et al. A predictive machine learning force-field framework for liquid electrolyte development. Nature Machine Intelligence, pages 1–10, 2025

    2025

  2. [1]

    Elena, Dávid P

    Ilyes Batatia, Philipp Benner, Yuan Chiang, Alin M. Elena, Dávid P. Kovács, Janosh Riebesell, Xavier R. Advincula, Mark Asta, Matthew Avaylon, William J. Baldwin, Fabian Berger, Noam Bernstein, Arghya Bhowmik, Samuel M. Blau, Vlad Cărare, James P. Darby, Sandip De, Flaviano Della Pia, Volker L. Deringer, Rokas Elijošius, Zakariya El-Machachi, Fabio Falcio...

    2024

  3. [2]

    Density-functional exchange-energy approximation with correct asymptotic behavior.Physical review A, 38(6):3098, 1988

    Axel D Becke. Density-functional exchange-energy approximation with correct asymptotic behavior.Physical review A, 38(6):3098, 1988

    1988

  4. [3]

    Axel D. Becke. Density-functional thermochemistry. III. The role of exact exchange.J. Chem. Phys., 98(7): 5648–5652, 04 1993. ISSN 0021-9606. doi: 10.1063/1.464913

    doi:10.1063/1.464913 1993

  5. [4]

    Density-functional thermochemistry

    Axel D Becke. Density-functional thermochemistry. v. systematic optimization of exchange-correlation functionals. The Journal of chemical physics, 107(20):8554–8560, 1997

    1997

  6. [5]

    Melting points of water models: Current situation.The Journal of Chemical Physics, 156(21), 2022

    Samuel Blazquez and Carlos Vega. Melting points of water models: Current situation.The Journal of Chemical Physics, 156(21), 2022

    2022

  7. [6]

    Shapeev, and Zhuang Xiaoying

    Mortazavi Bohayra, Javvaji Brahmanandam, Shojaei Fazel, Rabczuk Timon, Alexander V. Shapeev, and Zhuang Xiaoying. Exceptional piezoelectricity, high thermal conductivity and stiffness and promising photocatalysis in two-dimensional mosi2n4 family confirmed by first-principles.Nano Energy, 82:105716, 2021

    2021

  8. [7]

    Podryabinkin Evgeny, S

    Mortazavi Bohayra, V. Podryabinkin Evgeny, S. Novikov Ivan, Rabczuk Timon, Zhuang Xiaoying, and V. Shapeev Alexander. Accelerating first-principles estimation of thermal conductivity by machine-learning interatomic potentials: A mtp/shengbte solution.Computer Physics Communications, 258:107583, 2021

    2021

Show all 80 references
  1. [8]

    A methodical selection process for the development of ketones and esters as bio-based replacements for traditional hydrocarbon solvents

    Fergal P Byrne, Bart Forier, Greet Bossaert, Charly Hoebers, Thomas J Farmer, and Andrew J Hunt. A methodical selection process for the development of ketones and esters as bio-based replacements for traditional hydrocarbon solvents. Green Chemistry, 20(17):4003–4011, 2018

    2018

  2. [9]

    URLhttps://cameochemicals.noaa.gov/chris/EFM.pdf

    CAMEO Chemicals NOAA. URLhttps://cameochemicals.noaa.gov/chris/EFM.pdf

  3. [10]

    Finite size effects in determination of thermal conductivities: comparing molecular dynamics results with simple models.J

    Patrice Chantrenne and Jean-Louis Barrat. Finite size effects in determination of thermal conductivities: comparing molecular dynamics results with simple models.J. Heat Transfer, 126(4):577–585, 2004

    2004

  4. [11]

    A universal graph deep learning interatomic potential for the periodic table.Nat

    Chi Chen and Shyue Ong. A universal graph deep learning interatomic potential for the periodic table.Nat. Comput. Sci., 2:718–728, 11 2022. doi: 10.1038/s43588-022-00349-3

    2022 doi

  5. [12]

    Junmin Chen and Kuang Yu. PhyNEO: A neural-network-enhanced physics-driven force field development workflow for bulk organic molecule and polymer simulations.Journal of Chemical Theory and Computation, 20 (1):253–265, 2024

    2024

  6. [13]

    Density and refractive index at 298.15 k and vapor- liquid equilibria at 101.3 kpa for four binary systems of methanol, n-propanol, n-butanol, or isobutanol with n-methylpiperazine

    Shuda Chen, Qunfang Lei, and Wenjun Fang. Density and refractive index at 298.15 k and vapor- liquid equilibria at 101.3 kpa for four binary systems of methanol, n-propanol, n-butanol, or isobutanol with n-methylpiperazine. Journal of Chemical & Engineering Data, 47(4):811–815, 2002

    2002

  7. [15]

    Computing the heat conductivity of fluids from density fluctuations.Physical Review Letters, 125(13):130602, 2020

    Bingqing Cheng and Daan Frenkel. Computing the heat conductivity of fluids from density fluctuations.Physical Review Letters, 125(13):130602, 2020

    2020

  8. [16]

    Bartel, and Gerbrand Ceder

    Bowen Deng, Peichen Zhong, KyuJung Jun, Janosh Riebesell, Kevin Han, Christopher J. Bartel, and Gerbrand Ceder. Chgnet as a pretrained universal neural network potential for charge-informed atomistic modelling.Nat. Mach. Intell., 5(9), 9 2023. doi: 10.1038/s42256-023-00716-3. 9

    2023 doi

  9. [17]

    Sigmoid-weighted linear units for neural network function approxi- mation in reinforcement learning.Neural networks, 107:3–11, 2018

    Stefan Elfwing, Eiji Uchibe, and Kenji Doya. Sigmoid-weighted linear units for neural network function approxi- mation in reinforcement learning.Neural networks, 107:3–11, 2018

    2018

  10. [18]

    Understanding surface interactions in aqueous miscible organic solvent treated layered double hydroxides.RSC advances, 7(9):5076–5083, 2017

    Valentina Erastova, Matteo T Degiacomi, Dermot O’Hare, and H Chris Greenwell. Understanding surface interactions in aqueous miscible organic solvent treated layered double hydroxides.RSC advances, 7(9):5076–5083, 2017

    2017

  11. [19]

    Wei Feng, Luyan Zhang, Yaobo Cheng, Jin Wu, Chunguang Wei, Junwei Zhang, and Kuang Yu. Screening and design of aqueous zinc battery electrolytes based on the multimodal optimization of molecular simulation.The Journal of Physical Chemistry Letters, 16(13):3326–3335, 2025

    2025

  12. [20]

    Measurement of the thermal conductivity of five aliphatic esters in the liquid phase.The Journal of Chemical Thermodynamics, 138:140–146, 2019

    Song Fenhong, Ju Dapeng, Fan Jing, Wang Xiaopo, and Wang Gang. Measurement of the thermal conductivity of five aliphatic esters in the liquid phase.The Journal of Chemical Thermodynamics, 138:140–146, 2019

    2019

  13. [21]

    Jecfa flavonoids - details

    Food and Agriculture Organization. Jecfa flavonoids - details. URL https://www.fao.org/food/ food-safety-quality/scientific-advice/jecfa/jecfa-flav/details/en/c/245/

  14. [22]

    Thermal conductivity of ionic liquids and ionanofluids and their feasibility as heat transfer fluids.Industrial & Engineering Chemistry Research, 57(18):6516–6529, 2018

    João MP França, Maria José V Lourenço, SM Sohel Murshed, Agílio AH Pádua, and Carlos A Nieto de Castro. Thermal conductivity of ionic liquids and ionanofluids and their feasibility as heat transfer fluids.Industrial & Engineering Chemistry Research, 57(18):6516–6529, 2018

    2018

  15. [23]

    Elsevier, 2023

    Daan Frenkel and Berend Smit.Understanding molecular simulation: from algorithms to applications. Elsevier, 2023

    2023

  16. [24]

    Pvt property measurements for some aliphatic esters from (298 to 393) k and up to 35 mpa.Journal of Chemical & Engineering Data, 52(3):737–751, 2007

    Ramesh L Gardas, Irudayaraj Johnson, David MD Vaz, Isabel MA Fonseca, and Abel GM Ferreira. Pvt property measurements for some aliphatic esters from (298 to 393) k and up to 35 mpa.Journal of Chemical & Engineering Data, 52(3):737–751, 2007

    2007

  17. [26]

    Double-hybrid density functional theory for excited electronic states of molecules

    Stefan Grimme and Frank Neese. Double-hybrid density functional theory for excited electronic states of molecules. The Journal of chemical physics, 127(15), 2007

    2007

  18. [27]

    Refining potential energy surface through dynamical properties via differentiable molecular simulation

    Bin Han and Kuang Yu. Refining potential energy surface through dynamical properties via differentiable molecular simulation. Nature Communications, 16(1):816, 2025

    2025

  19. [28]

    Haynes.CRC Handbook of Chemistry and Physics

    W.M. Haynes.CRC Handbook of Chemistry and Physics. CRC Handbook of Chemistry and Physics. CRC Press,

  20. [29]

    Arnim Hellweg and Dmitrij Rappoport. Development of new auxiliary basis functions of the karlsruhe segmented contracted basis sets including diffuse basis functions (def2-svpd, def2-tzvppd, and def2-qvppd) for ri-mp2 and ri-cc calculations. Phys. Chem. Chem. Phys., 17:1010–101...

    2015 doi

  21. [30]

    Jorgensen, David S

    William L. Jorgensen, David S. Maxwell, and Julian Tirado-Rives. Development and testing of the opls all-atom force field on conformational energetics and properties of organic liquids.Journal of the American Chemical Society, 118(45):11225–11236, 1996

    1996

  22. [31]

    Mace-off23: Transferable machine learning force fields for organic molecules.arXiv preprint arXiv:2312.15211, 2023

    Dávid Péter Kovács, J Harry Moore, Nicholas J Browning, Ilyes Batatia, Joshua T Horton, Venkat Kapil, Ioan-Bogdan Magdău, Daniel J Cole, and Gábor Csányi. Mace-off23: Transferable machine learning force fields for organic molecules.arXiv preprint arXiv:2312.15211, 2023

    2023 arXiv

  23. [32]

    Langer, J

    Marcel F. Langer, J. Thorben Frank, and Florian Knoop. Stress and heat flux via automatic differentiation. The Journal of Chemical Physics, 159(17):174105, 11 2023. ISSN 0021-9606. doi: 10.1063/5.0155760. URL https://doi.org/10.1063/5.0155760

    2023 doi

  24. [33]

    CRC Press, 2018

    Philip A Laplante, Robin Cravey, Lawrence P Dunleavy, James L Antonakos, Rodney LeRoy, Jack East, Nicholas E Buris, Christopher J Conant, Lawrence Fryda, Robert William Boyd, et al.Comprehensive dictionary of electrical engineering. CRC Press, 2018

    2018

  25. [34]

    Capturing the nuclear quantum effects in molecular dynamics for lattice thermal conductivity calculations: Using ice as example.The Journal of Chemical Physics, 153(19), 2020

    Ripeng Luo and Kuang Yu. Capturing the nuclear quantum effects in molecular dynamics for lattice thermal conductivity calculations: Using ice as example.The Journal of Chemical Physics, 153(19), 2020

    2020

  26. [35]

    URL https://www.matweb.com/search/datasheet.aspx?matguid= f2c9a5d8608e4f5aac1d570d37a54ffc&ckck=1

    Material Data Sheet. URL https://www.matweb.com/search/datasheet.aspx?matguid= f2c9a5d8608e4f5aac1d570d37a54ffc&ckck=1. 10

  27. [36]

    Scaling deep learning for materials discovery.Nature, 624:1–6, 11 2023

    Amil Merchant, Simon Batzner, Samuel Schoenholz, Muratahan Aykol, Gowoon Cheon, and Ekin Cubuk. Scaling deep learning for materials discovery.Nature, 624:1–6, 11 2023. doi: 10.1038/s41586-023-06735-9

    2023 doi

  28. [37]

    Advances in the improvement of thermal- conductivity of phase change material-based lithium-ion battery thermal management systems: An updated review

    Abhijeet Mitra, Rajan Kumar, Dwesh Kumar Singh, and Zafar Said. Advances in the improvement of thermal- conductivity of phase change material-based lithium-ion battery thermal management systems: An updated review. Journal of Energy Storage, 53:105195, 2022

    2022

  29. [38]

    Advances in liquid coolant technologies for electronics cooling

    Satish C Mohapatra and Daniel Loikits. Advances in liquid coolant technologies for electronics cooling. Semiconductor Thermal Measurement and Management IEEE Twenty First Annual IEEE Symposium, 2005., pages 354–360, 2005

    2005

  30. [39]

    A simple nonequilibrium molecular dynamics method for calculating the thermal conduc- tivity

    Florian Müller-Plathe. A simple nonequilibrium molecular dynamics method for calculating the thermal conduc- tivity. The Journal of chemical physics, 106(14):6082–6085, 1997

    1997

  31. [40]

    Thermodynamic properties of four ester-hydrocarbon mixtures.Journal of Chemical and Engineering Data, 25(3):283–286, 1980

    Tatsuhiko Ohta and Isamu Nagata. Thermodynamic properties of four ester-hydrocarbon mixtures.Journal of Chemical and Engineering Data, 25(3):283–286, 1980

    1980

  32. [41]

    Densities and refractive indices of pure alcohols as a function of temperature.Journal of Chemical and Engineering Data, 27(3):312–317, 1982

    Juan Ortega. Densities and refractive indices of pure alcohols as a function of temperature.Journal of Chemical and Engineering Data, 27(3):312–317, 1982

    1982

  33. [42]

    Pytorch: An imperative style, high-performance deep learning library

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, et al. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32, 2019

    2019

  34. [43]

    Generalized gradient approximation made simple

    John P Perdew, Kieron Burke, and Matthias Ernzerhof. Generalized gradient approximation made simple. Physical review letters, 77(18):3865, 1996

    1996

  35. [44]

    Atomistic simulations of the thermal conductivity of liquids.Physical Review Materials, 4(5):053801, 2020

    Marcello Puligheddu and Giulia Galli. Atomistic simulations of the thermal conductivity of liquids.Physical Review Materials, 4(5):053801, 2020

    2020

  36. [45]

    A full coupled-cluster singles and doubles model: The inclusion of disconnected triples

    George D Purvis and Rodney J Bartlett. A full coupled-cluster singles and doubles model: The inclusion of disconnected triples. The Journal of Chemical Physics, 76(4):1910–1918, 1982

    1910

  37. [46]

    A fifth-order perturbation comparison of electron correlation theories.Chemical Physics Letters, 157(6):479–483, 1989

    Krishnan Raghavachari, Gary W Trucks, John A Pople, and Martin Head-Gordon. A fifth-order perturbation comparison of electron correlation theories.Chemical Physics Letters, 157(6):479–483, 1989

    1989

  38. [47]

    S Babu Sanker and Rajesh Baby. Phase change material based thermal management of lithium ion batteries: A review on thermal performance of various thermal conductivity enhancers.Journal of Energy Storage, 50:104606, 2022

    2022

  39. [48]

    Nandhibatla V Sastry, Sunil R Patel, and Saurabh S Soni. Excess molar volumes, excess isentropic compressibilities, excess viscosities, relative permittivity and molar polarization deviations for methyl acetate+, ethyl acetate+, butyl acetate+, isoamyl acetate+, methyl propion...

    2013

  40. [49]

    Reformulation of the d3(becke–johnson) dispersion correction without resorting to higher than c6 dispersion coefficients.J

    Heiner Schröder, Anne Creon, and Tobias Schwabe. Reformulation of the d3(becke–johnson) dispersion correction without resorting to higher than c6 dispersion coefficients.J. Chem. Theory Comput., 11(7):3163–3170, 6 2015. doi: 10.1021/acs.jctc.5b00400

    2015 doi

  41. [50]

    Size effects in molecular dynamics thermal conductivity predictions.Physical Review B—Condensed Matter and Materials Physics, 81 (21):214305, 2010

    Daniel P Sellan, Eric S Landry, JE Turney, Alan JH McGaughey, and Cristina H Amon. Size effects in molecular dynamics thermal conductivity predictions.Physical Review B—Condensed Matter and Materials Physics, 81 (21):214305, 2010

    2010

  42. [51]

    Serijan and Paul H

    Kasper T. Serijan and Paul H. Wise. Dicyclic hydrocarbons. iv. synthesis and physical properties ofα, α- and α,ι-diphenyl- and dicyclohexyl- pentanes and hexanes1.Journal of the American Chemical Society, 73, 11: 5191–5193, 1951

    1951

  43. [52]

    Approaching coupled cluster accuracy with a general- purpose neural network potential through transfer learning.Nature communications, 10(1):2903, 2019

    Justin S Smith, Benjamin T Nebgen, Roman Zubatyuk, Nicholas Lubbers, Christian Devereux, Kipton Barros, Sergei Tretiak, Olexandr Isayev, and Adrian E Roitberg. Approaching coupled cluster accuracy with a general- purpose neural network potential through transfer learning.Natur...

    2019

  44. [53]

    Strongly constrained and appropriately normed semilocal density functional

    Jianwei Sun, Adrienn Ruzsinszky, and John P Perdew. Strongly constrained and appropriately normed semilocal density functional. Physical review letters, 115(3):036402, 2015. 11

    2015

  45. [54]

    Reference correlation of the thermal conductivity of methanol from the triple point to 660 k and up to 245 mpa.Journal of Physical and Chemical Reference Data, 42(4), 2013

    EA Sykioti, Marc J Assael, Marcia L Huber, and Richard A Perkins. Reference correlation of the thermal conductivity of methanol from the triple point to 660 k and up to 245 mpa.Journal of Physical and Chemical Reference Data, 42(4), 2013

    2013

  46. [55]

    New thin lithium-ion batteries using a liquid electrolyte with thermal stability.Journal of power sources, 97:677–680, 2001

    Norio Takami, Masahiro Sekino, Takahisa Ohsaki, Motoya Kanda, and Masao Yamamoto. New thin lithium-ion batteries using a liquid electrolyte with thermal stability.Journal of power sources, 97:677–680, 2001

    2001

  47. [56]

    URLhttps://thermtestasia.cn/material-database

    Thermtest Asia Material Database. URLhttps://thermtestasia.cn/material-database

  48. [57]

    Lammps-a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales.Computer Physics Communications, 271:108171, 2022

    Aidan P Thompson, H Metin Aktulga, Richard Berger, Dan S Bolintineanu, W Michael Brown, Paul S Crozier, Pieter J In’t Veld, Axel Kohlmeyer, Stan G Moore, Trung Dac Nguyen, et al. Lammps-a flexible simulation tool for particle-based materials modeling at the atomic, meso, and c...

    2022

  49. [58]

    Triton: an intermediate language and compiler for tiled neural network computations

    Philippe Tillet, Hsiang-Tsung Kung, and David Cox. Triton: an intermediate language and compiler for tiled neural network computations. InProceedings of the 3rd ACM SIGPLAN International Workshop on Machine Learning and Programming Languages, pages 10–19, 2019

    2019

  50. [59]

    Heat transport in liquid water from first-principles and deep neural network simulations.Physical Review B, 104(22):224202, 2021

    Davide Tisi, Linfeng Zhang, Riccardo Bertossa, Han Wang, Roberto Car, and Stefano Baroni. Heat transport in liquid water from first-principles and deep neural network simulations.Physical Review B, 104(22):224202, 2021

    2021

  51. [60]

    Vanegas, and Marino Arroyo

    Alejandro Torres-Sánchez, Juan M. Vanegas, and Marino Arroyo. Geometric derivation of the microscopic stress: A covariant central force decomposition.J. Mech. Phys. Solids, 93:224–239, 2016. ISSN 0022-5096. doi: https://doi.org/10.1016/j.jmps.2016.03.006. Special Issue in hono...

    2016 doi

  52. [61]

    Enhancing thermal conductivity computation of polymers via machine learning techniques.The Journal of Physical Chemistry B, 129(33):8593–8602, 2025

    Chengyang Tu, Xin Li, Junmin Chen, Bo Sun, and Kuang Yu. Enhancing thermal conductivity computation of polymers via machine learning techniques.The Journal of Physical Chemistry B, 129(33):8593–8602, 2025

    2025

  53. [62]

    Physnet: A neural network for predicting energies, forces, dipole moments and partial charges

    Oliver Unke and Markus Meuwly. Physnet: A neural network for predicting energies, forces, dipole moments and partial charges. J. Chem. Theory Comput., 15, 05 2019. doi: 10.1021/acs.jctc.9b00181

    2019 doi

  54. [63]

    Machine learning force fields.Chem

    Oliver T Unke, Stefan Chmiela, Huziel E Sauceda, Michael Gastegger, Igor Poltavsky, Kristof T Schuütt, Alexandre Tkatchenko, and Klaus-Robert Müller. Machine learning force fields.Chem. Rev., 121(16):10142–10186, 2021

    2021

  55. [64]

    Gromacs: fast, flexible, and free.Journal of computational chemistry, 26(16):1701–1718, 2005

    David Van Der Spoel, Erik Lindahl, Berk Hess, Gerrit Groenhof, Alan E Mark, and Herman JC Berendsen. Gromacs: fast, flexible, and free.Journal of computational chemistry, 26(16):1701–1718, 2005

    2005

  56. [65]

    Thermochemistry of ionic liquid heat-transfer fluids.Thermochimica Acta, 425(1-2):181–188, 2005

    Michael E Van Valkenburg, Robert L Vaughn, Margaret Williams, and John S Wilkes. Thermochemistry of ionic liquid heat-transfer fluids.Thermochimica Acta, 425(1-2):181–188, 2005

    2005

  57. [66]

    Attention is all you need.Advances in neural information processing systems, 30, 2017

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in neural information processing systems, 30, 2017

    2017

  58. [67]

    Modeling thermal conductivity of concentrated and mixed-solvent electrolyte systems

    Peiming Wang and Andrzej Anderko. Modeling thermal conductivity of concentrated and mixed-solvent electrolyte systems. Industrial & engineering chemistry research, 47(15):5698–5709, 2008

    2008

  59. [68]

    Material properties of porous asphalt pavement cold patch mixtures with different solvents.Journal of Materials in Civil Engineering, 32(10):06020015, 2020

    Xiang Wang, Xueqin Chen, Qiao Dong, and Ahmad Jahanzaib. Material properties of porous asphalt pavement cold patch mixtures with different solvents.Journal of Materials in Civil Engineering, 32(10):06020015, 2020

    2020

  60. [69]

    Dmff: An open-source automatic differentiable platform for molecular force field development and molecular dynamics simulation.J

    Xinyan Wang, Jichen Li, Lan Yang, Feiyang Chen, Yingze Wang, Junhan Chang, Junmin Chen, Wei Feng, Linfeng Zhang, and Kuang Yu. Dmff: An open-source automatic differentiable platform for molecular force field development and molecular dynamics simulation.J. Chem. Theory Comput....

    2023 doi

  61. [70]

    Enhancing gpu-acceleration in the python-based simulations of chemistry frameworks.WIREs Computational Molecular Science, 15(2):e70008, 2025

    Xiaojie Wu, Qiming Sun, Zhichen Pu, Tianze Zheng, Wenzhi Ma, Wen Yan, Yu Xia, Zhengxiao Wu, Mian Huo, Xiang Li, Weiluo Ren, Sheng Gong, Yumin Zhang, and Weihao Gao. Enhancing gpu-acceleration in the python-based simulations of chemistry frameworks.WIREs Computational Molecular...

    2025 doi

  62. [71]

    Accurate prediction of heat conductivity of water by a neuroevolution potential.The Journal of Chemical Physics, 158(20), 2023

    Ke Xu, Yongchao Hao, Ting Liang, Penghua Ying, Jianbin Xu, Jianyang Wu, and Zheyong Fan. Accurate prediction of heat conductivity of water by a neuroevolution potential.The Journal of Chemical Physics, 158(20), 2023. 12

    2023

  63. [72]

    Nep-mb-pol: A unified machine-learned framework for fast and accurate prediction of water’s thermodynamic and transport properties

    Ke Xu, Ting Liang, Nan Xu, Penghua Ying, Shunda Chen, Ning Wei, Jianbin Xu, and Zheyong Fan. Nep-mb-pol: A unified machine-learned framework for fast and accurate prediction of water’s thermodynamic and transport properties. arXiv preprint arXiv:2411.09631, 2024

    2024 arXiv

  64. [73]

    Elsevier, 1995

    Carl L Yaws.Handbook of thermal conductivity,volume 2: organic compounds C5 To C7. Elsevier, 1995

    1995

  65. [74]

    Elsevier, 2009

    Carl L Yaws.Thermal Conductivity of Liquid–Organic Compounds. Elsevier, 2009

    2009

  66. [75]

    Elsevier, 1997

    CL Yaws.Handbook of thermal conductivity,Volume 1: Organic Compounds C1 to C4. Elsevier, 1997

    1997

  67. [76]

    Development of the electrolyte in lithium-ion battery: a concise review on its thermal hazards.Journal of Thermal Analysis and Calorimetry, pages 1–20, 2024

    Jia-Chi Ye, Yen-Wen Lai, Xin-Hao Huang, Zhi-Xiang Chang, Yi-Hung Chung, and Chi-Min Shu. Development of the electrolyte in lithium-ion battery: a concise review on its thermal hazards.Journal of Thermal Analysis and Calorimetry, pages 1–20, 2024

    2024

  68. [77]

    Differential transformer.arXiv preprint arXiv:2410.05258, 2024

    Tianzhu Ye, Li Dong, Yuqing Xia, Yutao Sun, Yi Zhu, Gao Huang, and Furu Wei. Differential transformer.arXiv preprint arXiv:2410.05258, 2024

    2024 arXiv

  69. [78]

    Study of thermal conductivity of nanofluids for the application of heat transfer fluids.Thermochimica Acta, 455(1-2):66–69, 2007

    Dae-Hwang Yoo, KS Hong, and Ho-Soon Yang. Study of thermal conductivity of nanofluids for the application of heat transfer fluids.Thermochimica Acta, 455(1-2):66–69, 2007

    2007

  70. [79]

    Thermal conductivity of water at extreme conditions.The Journal of Physical Chemistry B, 127(31):7011–7017, 2023

    Cunzhi Zhang, Marcello Puligheddu, Linfeng Zhang, Roberto Car, and Giulia Galli. Thermal conductivity of water at extreme conditions.The Journal of Physical Chemistry B, 127(31):7011–7017, 2023

    2023

  71. [80]

    York, Shi Liu, Tong Zhu, Zhicheng Zhong, Jian Lv, Jun Cheng, Weile Jia, Mohan Chen, Guolin Ke, Weinan E, Linfeng Zhang, and Han Wang

    Duo Zhang, Xinzijian Liu, Xiangyu Zhang, Chengqian Zhang, Chun Cai, Hangrui Bi, Yiming Du, Xuejian Qin, Jiameng Huang, Bowen Li, Yifan Shan, Jinzhe Zeng, Yuzhi Zhang, Siyuan Liu, Yifan Li, Junhan Chang, Xinyan Wang, Shuo Zhou, Jianchuan Liu, Xiaoshan Luo, Zhenyu Wang, Wanrun J...

    2023

  72. [81]

    embeddingMLP

    Meimei Zhang, Enrico Lussetti, Luís ES de Souza, and Florian Müller-Plathe. Thermal conductivities of molecular liquids by reverse nonequilibrium molecular dynamics.The Journal of PhysicalChemistry B, 109(31):15060–15067, 2005. 13 Appendix Supplementary Information Accelerated...

    2005

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.