betaFeedback

WorksPhysics and Astronomy

IRIS: an agentic eye on symmetry-derived paths and energy landscapes of structural phase transitions

Abstract

Structural phase transitions underpin ferroelectric and multiferroic functionality, but constructing interpretable energy landscapes remains dependent on specialist expertise. Predicting this behaviour rests on Landau free energies expanded in symmetry-adapted order parameters: their coefficients classify a transition, expose the couplings between distortions, set switching paths and barriers, and parametrize phase-field and effective-Hamiltonian models at larger scales. Obtaining these coefficients from first principles means sampling the Born–Oppenheimer energy surface along symmetry-derived structural paths, which requires crystallographic analysis, reliable structure generation and extensive computation, especially for coupled multidimensional order parameters. Here we introduce IRIS, an agentic workflow that converts plain-language requests and crystal structures into symmetry-adapted energy landscapes and fitted Landau energy models. Deterministic gates govern acceptance of structures, calculations and fits, while language models interpret requests and select constrained recovery actions. Applications across six compounds include a six-dimensional BiFeO3_3 landscape that separates competition between polar and tilt amplitudes from directional locking and ranks homogeneous switching pathways. Controlled replays detected all 53 injected faults; disabling gates allowed 14 silent errors. Four prospective BaTiO3_3 runs produced fitted landscapes from 96 density-functional calculations in 34 minutes total without intervening human decisions. IRIS makes physical model construction a reusable computational capability and provides a framework for scientific automation grounded in explicit verification.

Introduction

Structural phase transitions drive functional behaviour in ferroelectrics, multiferroics, shape-memory alloys, correlated oxides and topological systems, because the collective atomic distortions that change the crystal structure couple to electronic, magnetic and mechanical degrees of freedom [1, 2, 3, 4]. The energy landscape of a transition, with its minima, barriers and couplings, constrains which phases are thermodynamically stable and sets the energetics of homogeneous switching pathways, and it supplies the coefficients of Landau and phase-field models that describe domains and switching on larger scales, where kinetics also depend on nucleation, defects, interfaces and dynamics; without first-principles landscapes, these models rest on assumed or fitted parameters. Symmetry-adapted mode analysis decomposes the distortion between two phases into displacement modes that transform as irreducible representations (irreps) of the higher-symmetry parent [2, 5, 6]. When the lower-symmetry phase is a subgroup of the parent, the distortion components along the symmetry-breaking irreps are the order parameters: vectors in irrep space, each with an amplitude, a direction and as many components as its irrep has dimensions. Primary order parameters produce the symmetry lowering, and secondary ones are induced through their couplings. A transition is then a path on the energy landscape spanned by the order parameters, and phases without a group–subgroup relation can be connected through a common higher-symmetry phase.

Computing energy landscapes in the space of these order parameters becomes harder as the order parameters gain components, grow in number or couple: the two-dimensional order parameter in hexagonal manganites needs a basis that a single decomposition does not provide; the coupled polar and antiferrodistortive three-dimensional order parameters of BiFeO3\mathrm{BiFeO}_{3} span a space too large to sample naively; and trilinear (hybrid-improper) couplings in layered perovskites tie a polar mode to non-polar distortions [7, 8].

The tools for each step, from structure standardization and symmetry-mode analysis to first-principles codes and model fitting, are mature [6, 9, 10]; the bottleneck, and the risk, lie in the workflow that chains them. Workflow engines such as AiiDA and atomate automate execution, provenance and error handling [11, 12], but not the domain knowledge: which branch the crystallography calls for (the number and dimension of the active irreps decide whether a line, a plane or a high-dimensional grid is needed) and how to verify that each generated structure is the intended one. Because each tool is internally consistent, a mismatch of conventions, such as origin, atom ordering or amplitude normalization, can pass every check and yield a landscape that is smooth but wrong.

Automating such workflows requires adapting each step to what the previous one found while keeping every adaptation crystallographically sound. Large language model (LLM) agents, which interleave reasoning with calls to external tools [13, 14] and can work as coordinated teams [15], can interpret intermediate results and recover from failures [16]; they now plan chemical syntheses and autonomous experiments [17, 18, 19], run atomistic [20, 21, 22] and continuum [23, 24, 25] simulation workflows from plain-language requests, design alloys in multi-agent teams [26], retrieve materials data [27] and draft research studies [28]. The responses of these agents, however, can be fluent and internally consistent yet physically wrong [29], the underlying language models do not reliably detect such errors in their own reasoning [30], and benchmarks of chemical knowledge, of scientific agents and of general agents measure how often a language model answers correctly or completes a task [31, 32, 33, 34] rather than whether an error is caught before it propagates; safeguards on what an agent may do have been argued to matter more than its autonomy [35], and emulated tool use shows how often agents take risky actions [36]. Hence our design principle: a language model may propose what to do next, but deterministic checks of the crystallography, the calculations and the fit decide whether a result is accepted, as in planning frameworks in which a model proposes and external verifiers decide [37] and in verifier-guided reasoning [38].

Here we present IRIS (Irrep Route and Interpolation System), an agentic workflow built on this principle. Seven agents carry a plain-language request and two crystal structures to an energy landscape and a fitted Landau model. Branching decisions, such as which modes become axes and how many dimensions the grid needs, are explicit policies, and known failures are repaired by stored rules. A deterministic gate on every hand-off checks crystallographic consistency, mode fidelity and convergence, and every structure against the group–subgroup tree, so that no result moves on unless its gate passes. A language model acts only where no policy or stored rule decides, interpreting the request and choosing among the fixes that the code lists, and it never routes the workflow or accepts a result. We demonstrate IRIS on six compounds of increasing orderparameter complexity. The density functional theory (DFT) campaigns were run once, block by block, with the same inputs, outputs and checks as in IRIS, and failures were fixed by hand; those fixes became the stored rules. The complete gated graph was then rerun end to end on the stored energies and broken deliberately, to measure what the gates and the language model each contribute. The language model makes the workflow usable from a plain-language request and answers judgment calls without stopping, whereas failures already seen are better handled by stored rules; in these replays, the gates contained wrong results whichever of the two chose the fix. Replay measures the parts; to show the whole, we also ran IRIS prospectively on BaTiO3_3, from a typed request to the energy landscape: the structures were fetched from the Materials Project, the symmetry analysis was run live, and the three polar paths and the (PzP_z, PxP_x) plane were computed with the agent submitting and collecting its own density-functional calculations on a high-performance computing (HPC) cluster, every gate passing without a human decision.

Results

IRIS: a gated agentic workflow from two structures to a Landau model

IRIS takes a plain-language request and two crystal structures, a high-symmetry parent and a low-symmetry child, and returns a symmetry-adapted energy landscape, a fitted Landau model and an audit trail, a hashed log of every structure, calculation, gate result and decision behind each reported number (Figure 1; Methods). It is a state graph of seven agent nodes, each a component with its own inputs, outputs, decisions and gate rather than an independently reasoning language model: f0f_0 Intake, f1f_1 Crystallographer, f2f_2 Path Planner, f3f_3 Runner, f4f_4 Reviewer, f5f_5 Analyst and f6f_6 Skeptic, which validate the request, decompose the distortion into irrep modes, build a certified grid of structures, run and check the calculations, fit the Landau model and re-derive every reported number (Extended Data Table 1). Each hand-off passes a gate, deterministic code that returns a pass or an error code. A failure goes to a recovery node that tries a stored rule, then the gate’s sub-agent, which must choose from the fixes listed for that code, then a human, whose answer becomes a new rule; the same gate re-checks the result, so a language model never decides whether a step succeeded (Methods, § 4.1; Supplementary Tables S1–S3). Each node fif_i has a gate GiG_i (defined in Figure 1). Choices that shape a landscape are explicit policies, not model decisions: R1–R5 select the active modes (R5 refers a borderline selection to the sub-agent or the human), D1–D5 set the dimension and sampling of the grid (D4 treats hard secondary modes; D3 is not yet implemented) and P1–P5 fix the lattice (P1–P3 in the workflow; the anchors of P4 and the bias statement of P5 are still produced outside it; Methods, § 4.2; Supplementary Table S5).

Figure 1. Overview of IRIS as a gated state graph. IRIS runs from the human (left) to END through seven agent nodes, f0f_0–f6f_6: f0f_0 Intake, the only node that calls a language model directly, turns a plain-language request and two structures into a validated configuration; f1f_1 Crystallographer decomposes the distortion into irrep modes; f2f_2 Path Planner selects and certifies the modes and builds the grid; f3f_3 Runner submits the DFT jobs, and the graph pauses at a checkpoint until Slurm reports them finished (orange); f4f_4 Reviewer checks every energy; f5f_5 Analyst fits the Landau model; and f6f_6 Skeptic re-derives every reported number. Each node is followed by a gate on its outgoing edge, Gi∈{pass}∪{fail(c)}G_i \in\{\mathrm{pass}\} \cup\{\mathrm{fail}(c)\}: deterministic code (top right) that returns pass (✓, green arrows) or a failure with an error code cc: G0G_0, configuration valid; G1G_1, decomposition valid (H⊂GH \subset G, basis det =kH/kG= k_{H}/k_{G}); G2G_2, grid valid (certified mode vectors, every frame on the tree); G3G_3, jobs finished; G4G_4, energies usable; G5G_5, fit acceptable (cross-validation, coverage); G6G_6, claims verified; checks and codes in Supplementary Table S1. Any fail(c)(c) travels along the fail bus (red) to the recovery node (purple), which applies a stored rule R(i,c)\mathcal{R}(i,c) if one exists, otherwise asks the gate’s sub-agent σi\sigma_i, a language-model call that must choose one fix from the listed set Ωi(c)\Omega_i(c), and otherwise the human hh, reached through interrupt(); the answer resumes the graph through Command(decision) and is stored as a new rule, R←R∪{(i,c)↦h}\mathcal{R} \leftarrow\mathcal{R} \cup\{(i,c) \mapsto h\}. The fix is applied, fif_i is re-run and GiG_i re-checks it; fail(coverage) at G5G_5 returns to f2f_2. Top left: the chat interface, illustrated with a CsSnI3_3 exchange (request, a human decision at a gate, report) composed for the figure; the interface itself is the one used in the prospective BaTiO3_3 run of § 2.6, in which no decision went to the human; the intake was evaluated on text requests (§ 2.5), the front end was not. Right: the output (Figure 2b). Full description: Methods (§ 4.1) and Supplementary Section S1.1.

Unlike a scripted interpolation between two structures, every coordinate and structure is fixed by symmetry (the two constructions are compared in Supplementary § 12.2.2). Each axis is one component of an active order parameter, the amplitude of an irrep along a fixed direction of its irrep space, so a two- or three-component order parameter spans two or three axes; each frame is the parent plus certified mode vectors, so its space group is known before any calculation and is checked against the group–subgroup tree. Certification removes errors that symmetry checks cannot see (mismatched atoms, the rigid translation carried by polar modes, mis-scaled amplitudes), and the grid size follows the number of free order-parameter components (Extended Data Figures 1 and 2; Supplementary Figure S8).

Figure S8. From a mode decomposition to a certified landscape: what each IRIS agent adds. (a) The stages of the workflow coloured by the agent that runs them (f0f_0 purple, f1f_1 teal, f2f_2 orange, f3/f4f_3/f_4 red, f5f_5 blue, f6f_6 grey), with the gate (GiG_i) or Planner check (cic_i) that closes each one beneath; the mode decomposition is one stage of f1f_1. (b) For each requirement of a landscape: what the decomposition step returns on its own, what IRIS adds and which agent adds it, and the failure in the campaigns of this work that made the addition necessary (Extended Data Table 5). (c) Summary: one decomposition per parent–child pair against an audited grid, its energies and a Landau model from two CIFs.

Validation on perovskite tilt transitions

The six compounds, CsSnI3_3, Ca3_3Ti2_2O7_7, InMnO3_3, ErMnO3_3, BiFeO3_3 and HfO2_2, are ordered by order-parameter complexity, so that each exercises a part of the workflow that the previous one does not

(Extended Data Table 2). We first establish that the structures IRIS generates are the intended ones, using two perovskite tilt transitions in which two primary irreps span a plane (Figure 2; Supplementary Table S12; decompositions in Supplementary Table S7 and Supplementary Table S18; DFT settings for all systems in Supplementary Table S11). The two transitions test the construction in complementary ways. In cubic CsSnI3_3 (Pm3ˉm→PnmaPm\bar{3}m \rightarrow Pnma), which adopts the most common perovskite tilt pattern, the two irreps move disjoint displacement components; in the layered hybrid improper ferroelectric Ca3_3Ti2_2O7_7 (I4/mmm→A21amI4/mmm \rightarrow A2_{1}am) they share 16 of the 48 atoms, and only the child is available, so IRIS must first construct the parent by symmetrizing it (Supplementary Table S17). Symmetry also fixes what each plane contains. In CsSnI3_3 the rotation M3+M_{3}^{+} and tilt R4+R_{4}^{+} have isotropy subgroups that intersect in PnmaPnma, so every interior point has the ground-state symmetry; in Ca3_3Ti2_2O7_7, by contrast, the tilt X3−X_{3}^{-} and rotation X2+X_{2}^{+} condense on one arm, allowing their trilinear coupling to the polar Γ5−\Gamma_{5}^{-} [7, 8]. All 961 frames of the CsSnI3_3 plane and all 1,372 of the Ca3_3Ti2_2O7_7 plane (and all 10,201 nodes of its finer check grid) have the expected space groups, yet certification caught errors that leave the space group unchanged: four in Ca3_3Ti2_2O7_7 (permuted atoms, mis-scaled vectors, norms short by 2\sqrt{2}, caught by the norm check of G1G_{1}, and a rigid translation making up 75% of the squared norm of the raw Γ5−\Gamma_{5}^{-} vector; Supplementary Fig. S4b) and mis-scaled vectors in CsSnI3_3, besides a single-mode frame on the child’s sheared lattice that the three-way symmetry check caught (Supplementary Fig. S5c; Supplementary § 12.2.1).

References

  1. [1]Michael A. Carpenter and Christopher J. Howard. Symmetry rules and strain/order-parameter relationships for coupling between octahedral tilting and cooperative Jahn–Teller transitions in ABX₃ perovskites. I. Theory. Acta Crystallographica Section B, 65:134–146, 2009.DOI
  2. [2]J. M. Perez-Mato, D. Orobengoa, and M. I. Aroyo. Mode crystallography of distorted structures. Acta Crystallographica Section A, 66:558–590, 2010.DOI
  3. [3]Sang-Wook Cheong and Maxim Mostovoy. Multiferroics: a magnetic twist for ferroelectricity. Nature Materials, 6:13–20, 2007.pubmed.ncbi.nlm.nih.gov/17199121
  4. [4]Masatoshi Imada, Atsushi Fujimori, and Yoshinori Tokura. Metal–insulator transitions. Reviews of Modern Physics, 70:1039–1263, 1998.DOI
  5. [5]Christopher J. Howard and Harold T. Stokes. Group-theoretical analysis of octahedral tilting in perovskites. Acta Crystallographica Section B, 54:782–789, 1998.
  6. [6]Branton J. Campbell, Harold T. Stokes, David E. Tanner, and Dorian M. Hatch. ISODISPLACE: a web-based tool for exploring structural distortions. Journal of Applied Crystallography, 39:607–614, 2006.DOI
  7. [7]Nicole A. Benedek and Craig J. Fennie. Hybrid improper ferroelectricity: a mechanism for controllable polarization-magnetization coupling. Physical Review Letters, 106:107204, 2011.
  8. [8]Elizabeth A. Nowadnick and Craig J. Fennie. Domains and ferroelectric switching pathways in Ca₃Ti₂O₇ from first principles. Physical Review B, 94:104105, 2016.
  9. [9]Harold T. Stokes and Dorian M. Hatch. FINDSYM: program for identifying the space-group symmetry of a crystal. Journal of Applied Crystallography, 38:237–238, 2005.DOI
  10. [10]G. Kresse and J. Furthmüller. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Physical Review B, 54:11169–11186, 1996.pubmed.ncbi.nlm.nih.gov/9984901
  11. [11]Sebastiaan P. Huber, Spyros Zoupanos, Martin U h rin, Leopold Talirz, Leonid Kahle, Rico Häuselmann, Dominik Gresch, Tiziano Müller, Aliaksandr V. Yakutovich, Casper W. Andersen, et al. AiiDA 1.0, a scalable computational infrastructure for automated reproducible workflows and data provenance. Scientific Data, 7:300, 2020. doi: 10.1038/s41597-020-00638-4.DOI
  12. [12]Kiran Mathew, Joseph H. Montoya, Alireza Fagh aninia, Shyam S. Dwaraknath, Muratahan Aykol, Hanmei Tang, Iek-heng Chu, Tess Smidt, Brandon Bocklund, Matthew Horton, et al. Automate: A high-level interface to generate, execute, and analyze computational materials science workflows. Computational Materials Science, 139:140–152, 2017. doi: 10.1016/j.commatsci.2017.07.030.DOI
  13. [13]Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R. Narasimhan, and Yuan Cao. ReAct: Synergizing reasoning and acting in language models. In International Conference on Learning Representations (ICLR), 2023.arxiv.org/abs/2210.03629
  14. [14]Timo Schick, Jane Dwivedi-Yu, Roberto Dessi, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. Toolformer: Language models can teach themselves to use tools. In Advances in Neural Information Processing Systems, volume 36, pages 68539–68551, 2023. doi: 10.52202/075280-2997.DOI
  15. [15]Sirui Hong, Mingchen Zhuge, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, Lingfeng Xiao, Chenglin Wu, and Jürgen Schmidhuber. MetaGPT: Meta programming for a multi-agent collaborative framework. In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2308.00352
  16. [16]Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. Reflexion: Language agents with verbal reinforcement learning. In Advances in Neural Information Processing Systems, volume 36, pages 8634–8652, 2023. doi: 10.52202/075280-0377.DOI
  17. [17]Daniil A. Boiko, Robert MacKnight, Ben Kline, and Gabe Gomes. Autonomous chemical research with large language models. Nature, 624:570–578, 2023.DOI
  18. [18]Andres M. Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D. White, and Philippe Schwaller. Augmenting large language models with chemistry tools. Nature Machine Intelligence, 6:525–535, 2024.DOI
  19. [19]Nathan J. Szymanski et al. An autonomous laboratory for the accelerated synthesis of inorganic materials. Nature, 624:86–91, 2023.DOI
  20. [20]Yunheng Zou, Austin H. Cheng, Abdulrahman Aldossary, Jiaru Bai, Shi Xuan Leong, Jorge Arturo Campos-Gonzalez-Angulo, Changhyeok Choi, Cher Tian Ser, Gary Tom, Andrew Wang, et al. El Agente: An autonomous agent for quantum chemistry. Matter, 8(7):102263, 2025. doi: 10.1016/j.matt.2025.102263.DOI
  21. [21]Thang D. Pham, Aditya Tanikanti, and Murat Keçeli. ChemGraph as an agentic framework for computational chemistry workflows. Communications Chemistry, 9:33, 2026. doi: 10.1038/s42004-025-01776-9.DOI
  22. [22]Quintina Campbell, Sam Cox, Jorge Medina, Brittany Watterson, and Andrew D. White. MDCrow: Automating molecular dynamics workflows with large language models. arXiv:2502.09565, 2025.arxiv.org/abs/2502.09565
  23. [23]Sukriti Manna, Henry Chan, and Subramanian K. R. S. Sankaranarayanan. AutoMOOSE: An agentic AI for autonomous phase-field simulation. arXiv preprint arXiv:2603.20986, 2026.arxiv.org/abs/2603.20986
  24. [24]Bo Ni and Markus J. Buehler. MechAgents: Large language model multi-agent collaborations can solve mechanics problems, generate new data, and integrate knowledge. Extreme Mechanics Letters, 67:102131, 2024.arxiv.org/abs/2311.08166
  25. [25]Sandeep Pandey, Ran Xu, Wenkang Wang, and Xu Chu. OpenFOAMGPT: A retrieval-augmented large language model (LLM) agent for OpenFOAM-based computational fluid dynamics. Physics of Fluids, 37(3):035120, 2025. doi: 10.1063/5.0257555.DOI
  26. [26]Alireza Ghafarollahi and Markus J. Buehler. Automating alloy design and discovery with physics-aware multimodal multiagent AI. Proceedings of the National Academy of Sciences, 122(4):e2414074122, 2025. doi: 10.1073/pnas.2414074122.DOI
  27. [27]Yuan Chiang, Elvis Hsieh, Chia-Hong Chou, and Janosh Riebesell. LLaMP: Large language model made powerful for high-fidelity materials knowledge retrieval. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 25189–25221. Association for Computational Linguistics, 2025. doi: 10.18653/v1/2025.emnlp-main.1280.
  28. [28]Chris Lu, Cong Lu, Robert Tjarko Lange, Yutaro Yamada, Shengran Hu, Jakob Foerster, David Ha, and Jeff Clune. Towards end-to-end automation of AI research. Nature, 651(8107):914–919, 2026. doi: 10.1038/s41586-026-10265-5.DOI
  29. [29]Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezhen Yu, Dan Su, Yan Xu, Etsuko Ishii, Yejin Bang, Delong Chen, Wenliang Dai, et al. Survey of hallucination in natural language generation. ACM Computing Surveys, 2023. doi: 10.1145/3571730.DOI
  30. [30]Jie Huang, Xinyun Chen, Swaroop Mishra, Huaixiu Steven Zheng, Adams Wei Yu, Xinying Song, and Denny Zhou. Large language models cannot self-correct reasoning yet. In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2310.01798
  31. [31]Adrian Mirza, Nawaf Alamp ara, Sreekanth Kunchapu, Martiño Ríos-García, Benedict Emoekabu, Aswanth Krishnan, Tanya Gupta, Mara Schilling-Wilhelmi, Macjonathan Okereke, Anagha Aneesh, et al. A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists. Nature Chemistry, 17:1027–1034, 2025. doi: 10.1038/s41557-025-01815-x.DOI
  32. [32]Ziru Chen, Shijie Chen, Yuting Ning, Qianheng Zhang, Boshi Wang, Botao Yu, Yifei Li, Zeyi Liao, Chen Wei, Zitong Lu, et al. ScienceAgentBench: Toward rigorous assessment of language agents for data-driven scientific discovery. In International Conference on Learning Representations (ICLR), 2025.arxiv.org/abs/2410.05080
  33. [33]Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, et al. AgentBench: Evaluating LLMs as agents. In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2308.03688
  34. [34]Carlos E. Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, and Karthik Narasimhan. SWE-bench: Can language models resolve real-world GitHub issues? In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2310.06770
  35. [35]Xiangru Tang, Qiao Jin, Kunlun Zhu, Tongxin Yuan, Yichi Zhang, Wangchunshu Zhou, Meng Qu, Yilun Zhao, Jian Tang, Zhuosheng Zhang, et al. Risks of AI scientists: prioritizing safeguarding over autonomy. Nature Communications, 16:8317, 2025. doi: 10.1038/s41467-025-63913-1.DOI
  36. [36]Yangjun Ruan, Honghua Dong, Andrew Wang, Silviu Pitis, Yongchao Zhou, Jimmy Ba, Yann Dubois, Chris J. Maddison, and Tatsunori Hashimoto. Identifying the risks of LM agents with an LM-emulated sandbox. In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2309.15817
  37. [37]Subbarao Kambhampati, Karthik Valmeekam, Lin Guan, Mudit Verma, Kaya Stechly, Siddhant Bhambri, Lucas Paul Saldyt, and Anil B. Murthy. Position: LLMs can’t plan, but can help planning in LLM-modulo frameworks. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 22895–22907. PMLR, 2024.arxiv.org/abs/2402.01817
  38. [38]Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step. In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2305.20050
  39. [39]Anubhav Jain, Shyue Ping Ong, Geoffrey Hautier, Wei Chen, William Davidson Richards, Stephen Dacek, Shreyas Cholia, Dan Gunter, David Skinner, Gerbrand Ceder, and Kristin A. Persson. Commentary: The Materials Project: A materials genome approach to accelerating materials innovation. APL Materials, 1(1):011002, 2013.DOI
  40. [40]Christopher Grote and Robert F. Berger. Strain tuning of tin–halide and lead–halide perovskites: a first-principles atomic and electronic structure study. The Journal of Physical Chemistry C, 119(40): 22832–22837, 2015. doi: 10.1021/acs.jpcc.5b07446.DOI
  41. [41]Quintin N. Meier, Adrien Stucky, Jeremie Teyssier, Sinéad M. Griffin, Dirk van der Marel, and Nicola A. Spaldin. Manifestation of structural Higgs and Goldstone modes in the hexagonal manganites. Physical Review B, 102:014102, 2020.DOI
  42. [42]Sergey Artyukhin, Kris T. Delaney, Nicola A. Spaldin, and Maxim Mostovoy. Landau theory of topological defects in multiferroic hexagonal manganites. Nature Materials, 13:42–49, 2014.DOI
  43. [43]Craig J. Fennie and Karin M. Rabe. Ferroelectric transition in YMnO₃ from first principles. Physical Review B, 72:100103(R), 2005.DOI
  44. [44]Didrik R. Småbråten, Quintin N. Meier, Sandra H. Skjærvø, Katherine Inzani, Dennis Meier, and Sverre M. Selbach. Charged domain walls in improper ferroelectric hexagonal manganites and gallates. Physical Review Materials, 2:114405, 2018.DOI
  45. [45]J. B. Neaton, C. Ederer, U. V. Waghmare, N. A. Spaldin, and K. M. Rabe. First-principles study of spontaneous polarization in multiferroic BiFeO₃. Physical Review B, 71:014113, 2005.DOI
  46. [46]Claude Ederer and Nicola A. Spaldin. Weak ferromagnetism and magnetoelectric coupling in bismuth ferrite. Physical Review B, 71:060401(R), 2005.DOI
  47. [47]Bin Xu, Vincent Garcia, Stéphane Fusil, Manuel Bibes, and L. Bellaiche. Intrinsic polarization switching mechanisms in BiFeO₃. Physical Review B, 95:104104, 2017.DOI
  48. [48]Oswaldo Diéguez, O. E. González-Vázquez, Jacek C. Wojdel, and Jorge Íñiguez. First-principles predictions of low-energy phases of multiferroic BiFeO₃. Physical Review B, 83:094105, 2011.DOI
  49. [49]Natalya S. Fedorova, Dmitri E. Nikonov, Hai Li, Ian A. Young, and Jorge Íñiguez. First-principles Landau-like potential for BiFeO₃ and related materials. Phys. Rev. B, 106:165122, 2022. doi: 10.1103/PhysRevB.106.165122.DOI
  50. [50]F. Kubel and H. Schmid. Structure of a ferroelectric and ferroelastic monodomain crystal of the perovskite BiFeO₃. Acta Crystallogr. B, 46:698–702, 1990.DOI
  51. [51]D. Lebeugle, D. Colson, A. Forget, and M. Viret. Very large spontaneous electric polarization in BiFeO₃ single crystals at room temperature and its evolution under cycling fields. Appl. Phys. Lett., 91:022907, 2007. doi: 10.1063/1.2753390.DOI
  52. [52]Ronald E. Cohen. Origin of ferroelectricity in perovskite oxides. Nature, 358(6382):136–138, 1992.DOI
  53. [53]R. D. King-Smith and David Vanderbilt. First-principles investigation of ferroelectricity in perovskite compounds. Physical Review B, 49(9):5828–5844, 1994.pubmed.ncbi.nlm.nih.gov/10011559
  54. [54]T. S. Böscke, J. Müller, D. Bräuhaus, U. Schröder, and U. Böttger. Ferroelectricity in hafnium oxide thin films. Applied Physics Letters, 99:102903, 2011.DOI
  55. [55]Johannes Müller, Tim S. Böscke, Uwe Schröder, Stefan Mueller, Dennis Bräuhaus, Ulrich Böttger, Lothar Frey, and Thomas Mikolajick. Ferroelectricity in simple binary ZrO₂ and HfO₂. Nano Letters, 12:4318–4323, 2012.DOI
  56. [56]Min Hyuk Park, Young Hwan Lee, Han Joon Kim, Yu Jin Kim, Taehwan Moon, Keum Do Kim, Johannes Müller, Alfred Kersch, Uwe Schroeder, Thomas Mikolajick, and Cheol Seong Hwang. Ferroelectricity and antiferroelectricity of doped thin HfO₂-based films. Advanced Materials, 27:1811–1831, 2015.DOI
  57. [57]Suraj S. Cheema et al. Enhanced ferroelectricity in ultrathin films grown directly on silicon. Nature, 580:478–482, 2020.DOI
  58. [58]Aldo Raelijarijaoana and R. E. Cohen. Hafnia HfO₂ is a proper ferroelectric. Physical Review B, 108 (9):094109, 2023.
  59. [59]F. Delodovici, P. Barone, and S. Picozzi. Trilinear-coupling-driven ferroelectricity in HfO₂. Physical Review Materials, 5:064405, 2021. doi: 10.1103/PhysRevMaterials.5.064405.DOI
  60. [60]Sukriti Manna et al. Equivariant neural network potential for hafnia. submitted, 2026.
  61. [61]LangChain, Inc. LangGraph: a library for building stateful, multi-actor applications with language models. https://github.com/langchain-ai/langgraph, 2026.
  62. [62]Luca Beurer-Kellner, Marc Fischer, and Martin Vechev. Guiding LLMs the right way: Fast, non-invasive constrained generation. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 3658–3673. PMLR, 2024.
  63. [63]Atsushi Togo, Kohei Shinohara, and Isao Tanaka. Spglib: a software library for crystal symmetry search. Science and Technology of Advanced Materials: Methods, 4:2384822, 2024. doi: 10.1080/27660400.2024.2384822.DOI
  64. [64]H. T. Stokes, D. M. Hatch, and B. J. Campbell. ISOTROPY software suite. https://iso.byu.edu, 2026.
  65. [65]Ilyes Batatia, Dávid P. Kovács, Gregor N. C. Simm, Christoph Ortner, and Gábor Csányi. MACE: Higher order equivariant message passing neural networks for fast and accurate force fields. In Advances in Neural Information Processing Systems, volume 35, pages 11423–11436, 2022. doi: 10.52202/068431-0830.DOI
  66. [66]Andy B. Yoo, Morris A. Jette, and Mark Grondona. SLURM: Simple Linux utility for resource management. In Job Scheduling Strategies for Parallel Processing, volume 2862 of Lecture Notes in Computer Science, pages 44–60. Springer, 2003.DOI
  67. [67]William Burnside. Theory of Groups of Finite Order. Cambridge University Press, Cambridge, 2nd edition, 1911.
  68. [68]Peter M. Neumann. A lemma that is not Burnside’s. The Mathematical Scientist, 4(2):133–141, 1979.
  69. [69]P. E. Blöchl. Projector augmented-wave method. Physical Review B, 50:17953–17979, 1994.pubmed.ncbi.nlm.nih.gov/9976227
  70. [70]G. Kresse and D. Joubert. From ultrasoft pseudopotentials to the projector augmented-wave method. Physical Review B, 59:1758–1775, 1999.DOI
  71. [71]John P. Perdew, Kieron Burke, and Matthias Ernzerhof. Generalized gradient approximation made simple. Physical Review Letters, 77:3865–3868, 1996.pubmed.ncbi.nlm.nih.gov/10062328
  72. [72]John P. Perdew, Adrienn Ruzsinszky, Gábor I. Csonka, Oleg A. Vydrov, Gustavo E. Scuseria, Lucian A. Constantin, Xiaolan Zhou, and Kieron Burke. Restoring the density-gradient expansion for exchange in solids and surfaces. Physical Review Letters, 100:136406, 2008.pubmed.ncbi.nlm.nih.gov/18517979
  73. [73]S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys, and A. P. Sutton. Electron-energy-loss spectra and the structural stability of nickel oxide: an LSDA+U study. Physical Review B, 57:1505–1509, 1998.DOI
  74. [74]D. M. Ceperley and B. J. Alder. Ground state of the electron gas by a stochastic method. Physical Review Letters, 45(7):566–569, 1980.
  75. [75]J. P. Perdew and Alex Zunger. Self-interaction correction to density-functional approximations for many-electron systems. Physical Review B, 23(10):5048–5079, 1981.DOI
  76. [76]A. I. Liechtenstein, V. I. Anisimov, and J. Zaanen. Density-functional theory and strong interactions: orbital ordering in Mott-Hubbard insulators. Physical Review B, 52:R5467–R5470, 1995.pubmed.ncbi.nlm.nih.gov/9981806
  77. [77]Igor A. Kornev, L. Bellaiche, P.-E. Janolin, B. Dkhil, and E. Suard. Phase diagram of Pb(Zr,Ti)O₃ solid solutions from first principles. Physical Review Letters, 97:157601, 2006.
  78. [78]Weinan E, Weiqing Ren, and Eric Vanden-Eijnden. String method for the study of rare events. Physical Review B, 66(5):052301, 2002.DOI
  79. [79]Jean Duchon. Splines minimizing rotation-invariant semi-norms in Sobolev spaces. In Constructive Theory of Functions of Several Variables, volume 571 of Lecture Notes in Mathematics, pages 85–100. Springer, Berlin, 1977.DOI
  80. [80]Anthropic. Claude Sonnet 5.5 (API model claude-sonnet-5-5). Models overview, Claude Platform Docs, https://platform.claude.com/docs/en/models/overview, 2026. Documentation accessed 19 Oct 2026; model used 17–29 Oct 2026.
  81. [81]Anthropic. Claude Opus 5.5 (API model claude-opus-5-5). Models overview, Claude Platform Docs, https://platform.claude.com/docs/en/models/overview, 2026. Documentation accessed 19 Oct 2026; model used 17–29 Oct 2026.
  82. [82]OpenAI. GPT-6.1 Sol (API model gpt-6.1-sol). OpenAI API documentation, model page, https://developers.openai.com/api/docs/models/gpt-6.1-sol, 2026. Documentation accessed 19 Oct 2026; model used 17–29 Oct 2026.
  83. [83]Google. Gemini 4 Argon (API model gemini-4-argon). Gemini API, Google AI for Developers, https://ai.google.dev/gemini-api/docs/models/gemini-4-argon, 2026. Documentation accessed 19 Oct 2026; model used 17–29 Oct 2026.
  84. [84]Chujie Zheng, Hao Zhou, Fandong Meng, Jie Zhou, and Minlie Huang. Large language models are not robust multiple choice selectors. In International Conference on Learning Representations (ICLR), 2024.arxiv.org/abs/2309.03882
  85. [85]Bradley Efron. Bootstrap methods: another look at the jackknife. The Annals of Statistics, 7(1):1–26, 1979.
  86. [86]Quinn McNemar. Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika, 12(2):153–157, 1947.pubmed.ncbi.nlm.nih.gov/20254758

Paper details

vlnc:W_z7uu93gp
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude

View statement
Looking for related papers…
Contents