PhyMLP: an automated strategy for machine-learning potential construction via data fusion and adaptive point-sampling

Abstract

This paper presents an automated strategy for constructing machine-learning potential (MLP) datasets through a physics-strengthened point-sampling scheme called physics-strengthened machine-learning potential (PhyMLP). By integrating heterogeneous data sources - including the Rose equation of state, experimental pressure-volume (P-V) measurements, traditional empirical potentials, and first-principles calculations - this method effectively circumvents computational bottlenecks and reduces the reliance on the exhaustive density functional theory (DFT) computations inherent in conventional training-set construction. The PhyMLP employs an adaptive, physics-guided sampling strategy that leverages intrinsic material responses and requires only a limited number of critical DFT calculations to efficiently characterize the potential energy surface. Using body-centered cubic tungsten as a benchmark system, the moment tensor potential trained on the PhyMLP-generated dataset exhibited exceptional predictive accuracy across a wide spectrum of material properties, ranging from fundamental physical constants to complex defect energetics and kinetic behavior. Ultimately, this study established a systematic and computationally efficient paradigm for developing accurate transferable MLPs, thereby offering a robust framework for large-scale atomistic simulations and advanced material modeling.

Keywords

Machine-learning potentials / multi-source data fusion / automated dataset construction / tungsten / molecular dynamics

References

[1] Ge W,Fan Y,Stenzel MH.Machine learning in polymer research.Adv Mater2025;37:2413695

[2] Bartók AP,Bernstein N.Machine learning a general-purpose interatomic potential for silicon.Phys Rev X2018;8:041048

[3] Rohskopf A,Gordiz K,Henry A.Empirical interatomic potentials optimized for phonon properties.Npj Comput Mater2017;3:27

[4] Finnis MW.A simple empirical N-body potential for transition metals.Philos Mag A1984;50:45-55

[5] Daw MS.Embedded-atom method: derivation and application to impurities, surfaces, and other defects in metals.Phys Rev B1984;29:6443

[6] Baskes MI.Modified embedded-atom potentials for cubic materials and impurities.Phys Rev B1992;46:2727

[7] Fedik N,Kulichenko M.Extending machine learning beyond interatomic potentials for predicting molecular properties.Nat Rev Chem2022;6:653-72

[8] Mortazavi B,Rabczuk T.Atomistic modeling of the mechanical properties: the rise of machine learning interatomic potentials.Mater Horiz2023;10:1956-68

[9] Kulichenko M,Lubbers N.Data generation for machine learning interatomic potentials and beyond.Chem Rev2024;124:13681-714

[10] Chen C.A universal graph deep learning interatomic potential for the periodic table.Nat Comput Sci2022;2:718-28

[11] Ito K,Hyodo K.Machine learning interatomic potential with DFT accuracy for general grain boundaries in α-Fe.Npj Comput Mater2024;10:255

[12] Ji Y,Xu Z.Machine-learning interatomic potentials for long-range systems.Phys Rev Lett2025;135:178001

[13] Wan K,Shi X.Construction of high accuracy machine learning interatomic potential for surface/interface of nanomaterials - a review.Adv Mater2024;36:2305758

[14] Jordan MI.Machine learning: trends, perspectives, and prospects.Science2015;349:255-60

[15] Hinton GE.Reducing the dimensionality of data with neural networks.Science2006;313:504-7

[16] Sharifani K.Machine learning and deep learning: a review of methods and applications.World Inf Technol Eng J2023;10:3897-904https://ssrn.com/abstract=4458723. (accessed on 23 Jul 2026)

[17] Carleo G,Cranmer K.Machine learning and the physical sciences.Rev Mod Phys2019;91:045002

[18] Wang Q.Harnessing machine learning for high-entropy alloy catalysis: a focus on adsorption energy prediction.Npj Comput Mater2025;11:91

[19] Liu X,Pei Z.Machine learning for high-entropy alloys: progress, challenges and opportunities.Prog Mater Sci2023;131:101018

[20] Zhang Y,Chen W.DP-GEN: a concurrent learning platform for the generation of reliable deep learning based potential energy models.Comput Phys Commun2020;253:107206

[21] Galib M,Ponga M. AtomProNet: data flow to and from machine learning interatomic potentials in materials science. arXiv 2025, arXiv:2501.14039. Available online: https://doi.org/10.48550/arXiv.2501.14039. (accessed on 23 Jul 2026)

[22] Li W,Chen Y.DeePKS+ ABACUS as a bridge between expensive quantum mechanical models and machine learning potentials.J Phys Chem A2022;126:9154-64

[23] Chen C,Zhao R.NepTrain and NepTrainKit: automated active learning and visualization toolkit for neuroevo-lution potentials.Comput Phys Commun2025;317:109859

[24] Cao Y,Freitas R.Capturing short-range order in high-entropy alloys with machine learning potentials.npj Comput Mater2025;11:268

[25] Röcken S.Accurate machine learning force fields via experimental and simulation data fusion.npj Comput Mater2024;10:69

[26] Sauceda HE,Chmiela S,Müller KR.BIGDML - towards accurate quantum machine learning force fields for materials.Nat Commun2022;13:3733

[27] Shapeev AV.Moment tensor potentials: a class of systematically improvable interatomic potentials.Multiscale Model Simul2016;14:1153-73

[28] Liu J,Fan Z,Su Y.Large-scale machine-learning molecular dynamics simulation of primary radiation damage in tungsten.Phys Rev B2023;108:054312

[29] Kang S,Park T.Toward fast and accurate machine learning interatomic potentials for atomic layer deposition precursors.Mater Today Adv2024;21:100474

[30] Mortazavi B,Podryabinkin EV.Exploring phononic properties of two-dimensional materials using machine learning interatomic potentials.Appl Mater Today2020;20:100685

[31] Rose JH,Guinea F.Universal features of the equation of state of metals.Phys Rev B1984;29:2963

[32] Hixson RS.Shock compression of tungsten and molybdenum.J Appl Phys1992;71:1721-8

[33] Ma PW.Universality of point defect structure in body-centered cubic metals.Phys Rev Mater2019;3:013605

[34] Ma PW.Effect of stress on vacancy formation and migration in body-centered-cubic metals.Phys Rev Mater2019;3:063601

[35] Mason DR,Becquart CS.An empirical potential for simulating vacancy clusters in tungsten.J Phys Condens Matter2017;29:505501

[36] Alexander R,Proville L.Ab initio scaling laws for the formation energy of nanosized interstitial defect clusters in iron, tungsten, and vanadium.Phys Rev B2016;94:024103

[37] Bonny G,Bakaev A,Van Neck D.Many-body central force potentials for tungsten.Model Simul Mater Sci Eng2014;22:053001

[38] Gehringer D,Holec D.Models of configurationally-complex alloys made simple.Comput Phys Commun2023;286:108664

[39] Gilbert MR.Neutron-induced transmutation effects in W and W-alloys in a fusion environment.Nucl Fusion2011;51:043005

[40] Romaner L,Pippan R.Effect of rhenium on the dislocation core structure in tungsten.Phys Rev Lett2010;104:195503

[41] Tanno T,Nogami S.Microstructure development in neutron irradiated tungsten alloys.Mater Trans2011;52:1447-51

[42] Li YH,Liang L.Transition from ductilizing to hardening in tungsten: the dependence on rhenium distribution.Acta Mater2019;181:110-23

[43] Szlachta WJ,Csányi G.Accuracy and transferability of Gaussian approximation potential models for tungsten.Phys Rev B2014;90:104108

[44] Muzyk M,Kurzydłowski KJ,Dudarev SL.Phase stability, point defects, and elastic properties of WV and W-Ta alloys.Phys Rev B2011;84:104115

[45] Bonny G,Terentyev D.Interatomic potential to study plastic deformation in tungsten-rhenium alloys.J Appl Phys2017;121:165107

[46] Mrovec M,Bailey AG,Elsässer C.Bond-order potential for simulations of extended defects in tungsten.Phys Rev B2007;75:104119

[47] Featherston FH.Elastic constants of tantalum, tungsten, and molybdenum.Phys Rev1963;130:1324

[48] Wang Y,Zhang X.Calculated equation of state of Al, Cu, Ta, Mo, and W to 1000 GPa.Phys Rev Lett2000;84:3220

[49] Ferroni F,Arakawa K,Edmondson PD.High temperature annealing of ion irradiated tungsten.Acta Mater2015;90:380-93

[50] Chen Y,Gao N.New interatomic potentials of W, Re and W-Re alloy for radiation defects.J Nucl Mater2018;502:141-53

[51] Wang X,Zhang L,Wang H.A tungsten deep neural-network potential for simulating mechanical property degradation under fusion service environment.Nucl Fusion2022;62:126013

[52] Byggmästar J,Nordlund K.Machine-learning interatomic potential for radiation damage and defects in tungsten.Phys Rev B2019;100:144105

[53] Lide DR. CRC handbook of chemistry and physics: a ready-reference book of chemical and physical data. CRC Press; 1995. https://books.google.com/books/about/CRC_Handbook_of_Chemistry_and_Physics.html?id=q2qJId5TKOkC&utm_source=chatgpt.com. (accessed on 23 Jul 2026)

[54] Qi X,Wang S.Thermoelastic properties of tungsten at simultaneous high pressure and temperature.J Appl Phys2020;128:105105

[55] Renault PO,Bimbault L,Elkaım E.Poisson’s ratio measurement in tungsten thin films combining an x-ray diffractometer with in situ tensile tester.Appl Phys Lett1998;73:1952-4

[56] Kittel P.Enthalpy, entropy, and exergy flow losses in pulse tube cryocoolers. In Cryocoolers 13. Springer; 2005. pp. 343-52.

[57] Simmons G. Single crystal elastic constants and calculated aggregate properties. Cambridge, MA: MIT Press; 1971. https://scholar.smu.edu/cgi/viewcontent.cgi?article=1029&context=journal_grc. (accessed on 23 Jul 2026)

[58] You Y,Wu F.Principal component analysis enables the design of deep learning potential precisely capturing LLZO phase transitions.npj Comput Mater2024;10:57

[59] He S,El Atwani O.Complex dislocation loop networks as natural extensions of the sink efficiency of saturated grain boundaries in irradiated metals.Sci Adv2024;10:eadj8395

[60] Borges PPPO,Asta M.Electronic descriptors for dislocation deformation behavior and intrinsic ductility in bcc high-entropy alloys.Sci Adv2024;10:eadp7670

[61] Hou J,Kong XS.Hydrogen modulated dislocation reaction and defect accumulation in bcc metals.Acta Mater2025:121524

[62] Parakh A,Harkins KA.Nucleation of dislocations in 3.9 nm nanocrystals at high pressure.Phys Rev Lett2020;124:106104

[63] Li G,Zhang Y.Electromechanical coupled modulation of dislocation nucleation and annihilation in ferroelectric oxide films.Phys Rev B2025;111:224101

[64] Kanhaiya K,Im W.Accurate simulation of surfaces and interfaces of ten FCC metals and steel using Lennard–Jones potentials.npj Comput Mater2021;7:17

[65] Zhu LF,Gong Y.Melting properties of the refractory metals V and W and the binary VW alloy fully from first principles.Phys Rev B2024;109:094110

[66] Kresse G.Ab initio molecular-dynamics simulation of the liquid-metal–amorphous-semiconductor transition in germanium.Phys Rev B1994;49:14251

[67] Kresse G.Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set.Phys Rev B1996;54:11169

[68] Plimpton S.Fast parallel algorithms for short-range molecular dynamics.J Comput Phys1995;117:1-19