GENIUS: An agentic AI framework for autonomous design and execution of simulation protocols
Open accessAbstract
Abstract Predictive atomistic simulations have propelled materials discovery, yet routine setup and debugging still demand computer specialists. This know-how gap limits the use of Integrated Computational Materials Engineering (ICME), where state-of-the-art codes exist but remain cumbersome for non-experts. We address this bottleneck with GENIUS, an AI-agentic workflow that fuses a smart Quantum ESPRESSO knowledge graph with a tiered hierarchy of large language models supervised by a finite-state error-recovery machine. Here we show that GENIUS translates free-form human-generated prompts into Quantum ESPRESSO input files that pass early execution validation for ≈ 80% of 295 diverse benchmarks. Zero-shot generation succeeds for 14.2% of all prompts, and among cases that do not succeed initially, 76.3% are autonomously recovered by the automated error-handling loop, with the attempt-wise success rate decaying exponentially toward a 7% baseline. Compared with LLM-only baselines, GENIUS increases inference and computational efficiency and virtually eliminates hallucinations. The framework democratizes electronic-structure DFT simulations by intelligently automating protocol generation, validation, and repair, enabling large-scale screening and accelerating ICME design loops worldwide across academia and industry.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The corpus already demonstrates several concrete successes: autonomous VASP execution across routine electronic-structure tasks, large-scale Quantum ESPRESSO protocol generation and repair, substantial reduction of adsorption-search effort without collapsing energy quality, adaptive alloy optimization under repeated MD feedback, catalyst discovery with experimental confirmation, and one genuine多智能
“Importantly, the corpus already demonstrates several concrete successes: autonomous VASP execution across routine electronic-structure tasks [25], large-scale Quantum ESPRESSO protocol generation and repair [42], substantial reduction of adsorption-search effort without collapsing energy quality [28], adaptive alloy optimization under repeated MD feedback [33], catalyst discovery with experimental”
From From Scientific Copilots to Tool-Grounded Autonomy: AI Agents in Simulation-Driven Materials Discovery (2026) · cites this paper as [42]
Split into 2 atoms — 1 confirmed, 0 partially supported, 1 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredGENIUS benchmarked autonomous Quantum ESPRESSO protocol generation and repair at scale.
“GENIUS benchmarked autonomous Quantum ESPRESSO protocol generation and repair at scale [42].”
From From Scientific Copilots to Tool-Grounded Autonomy: AI Agents in Simulation-Driven Materials Discovery (2026) · cites this paper as [42]
GENIUS passed early execution validation for about 80% of 295 benchmark prompts and autonomously recovered 76.3% of initially failed cases.
“GENIUS passed early execution validation for about 80% of 295 benchmark prompts and autonomously recovered 76.3% of initially failed cases [42]”
From From Scientific Copilots to Tool-Grounded Autonomy: AI Agents in Simulation-Driven Materials Discovery (2026) · cites this paper as [42]
Papers such as SmartCIF, DynaMate, VASPilot, Matty, GENIUS, and ToPolyAgent show that a substantial share of the current advance lies not in replacing scientific judgment with text generation, but in making brittle infrastructural steps programmable and recoverable.
“Papers such as SmartCIF, DynaMate, VASPilot, Matty, GENIUS, and ToPolyAgent show that a substantial share of the current advance lies not in replacing scientific judgment with text generation, but in making these brittle infrastructural steps programmable and recoverable [21,25,37-39,42].”
From From Scientific Copilots to Tool-Grounded Autonomy: AI Agents in Simulation-Driven Materials Discovery (2026) · cites this paper as [42]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials
Giannozzi, P., Stefano Baroni, A., Bonini, N., Calandra, M., Car, R., Cavazzoni, C., Ceresoli, D., Chiarotti, G., Cococcioni, M., Dabo, I., et al. (2009) · Journal of Physics Condensed Matter · doi:10.1088/0953-8984/21/39/395502 · Cited by 29,336 · see citing works
Kohonen, T. (1990) · Proceedings of the IEEE · doi:10.1109/5.58325 · Cited by 8,414 · see citing works
Knowledge Graph Embedding: A Survey of Approaches and Applications
Wang, Q., Mao, Z., Wang, B., & Guo, L. (2017) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2017.2754499 · Cited by 2,654 · see citing works
DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
Guo, D., Yang, D., Zhang, H., Song, J., Wang, P., Zhu, Q., Xu, R., Zhang, R., Ma, S., Bi, X., et al. (2025) · Nature · doi:10.1038/s41586-025-09422-z · Cited by 1,701 · see citing works
Reproducibility in density functional theory calculations of solids
Lejaeghere, K., Bihlmayer, G., Björkman, T., Blaha, P., Blügel, S., Blum, V., Caliste, D., Castelli, I. E., Clark, S. J., Dal Corso, A., et al. (2016) · Science · doi:10.1126/science.aad3000 · Cited by 1,649 · see citing works
Bock, F. E., Aydin, R. C., Cyron, C. J., Huber, N. W., Kalidindi, S. R., & Klusemann, B. (2019) · Frontiers in Materials · doi:10.3389/fmats.2019.00110 · Cited by 410 · see citing works
Self-citationToward Computational Materials Design: The Impact of Density Functional Theory on Materials Research
Häfner, J., Wolverton, C., & Ceder, G. (2006) · MRS Bulletin · doi:10.1557/mrs2006.174 · Cited by 370 · see citing works
Integrated computational materials engineering: A new paradigm for the global materials profession
Allison, J. E., Backman, D., & Christodoulou, L. (2006) · JOM · doi:10.1007/s11837-006-0223-5 · Cited by 248 · see citing works
Jablonka, K. M., Ai, Q., Al-Feghali, A., Badhwar, S., Bocarsly, J. D., Bran, A. M., Bringuier, S., Brinson, L. C., Choudhary, K., Circi, D., et al. (2023) · Digital Discovery · doi:10.1039/d3dd00113j · Cited by 218 · see citing works
Is Cosine-Similarity of Embeddings Really About Similarity?
Steck, H., Ekanadham, C., & Kallus, N. (2024) · International World Wide Web Conference (WWW) · doi:10.1145/3589335.3651526 · Cited by 161 · see citing works
Integrated computational materials engineering of corrosion resistant alloys
Taylor, C. D., Lu, P., Saal, J. E., Frankel, G. S., & Scully, J. R. (2018) · npj Materials Degradation · doi:10.1038/s41529-018-0027-4 · Cited by 157 · see citing works
A family of dual-anion-based sodium superionic conductors for all-solid-state sodium-ion batteries
Lin, X., Zhang, S., Yang, M., Xiao, B., Zhao, Y., Luo, J., Fu, J., Wang, C., Li, X., Li, W. S., et al. (2024) · Nature Materials · doi:10.1038/s41563-024-02011-x · Cited by 152 · see citing works
The fundamental principles of reproducibility
Gundersen, O. E. (2021) · Philosophical Transactions of the Royal Society A Mathematical Physical and Engineering Sciences · doi:10.1098/rsta.2020.0210 · Cited by 101 · see citing works
Han, X., Alameh, M., Gong, N., Xue, L., Ghattas, M., Bojja, G., Xu, J., Zhao, G., Warzecha, C. C., Padilla, M. S., et al. (2024) · Nature Chemistry · doi:10.1038/s41557-024-01557-2 · Cited by 96 · see citing works
Shen, S. C.-Y., Khare, E., Lee, N., Saad, M. K., Kaplan, D. L., & Buehler, M. J. (2023) · Chemical Reviews · doi:10.1021/acs.chemrev.2c00479 · Cited by 78 · see citing works
Expansion of the Materials Cloud 2D Database
Campi, D., Mounet, N., Gibertini, M., Pizzi, G., & Marzari, N. (2023) · ACS Nano · doi:10.1021/acsnano.2c11510 · Cited by 75 · see citing works
Bridges and Mechanisms: Integrating Systems Science Thinking into Implementation Research
Luke, D. A., Powell, B. J., & Paniagua-Ávila, A. (2023) · Annual Review of Public Health · doi:10.1146/annurev-publhealth-060922-040205 · Cited by 56 · see citing works
Workflow Engineering in Materials Design within the BATTERY 2030 + Project
Schaarschmidt, J., Yuan, J., Strunk, T., Kondov, I., Huber, S. P., Pizzi, G., Kahle, L., Bölle, F. T., Castelli, I. E., Vegge, T., et al. (2021) · Advanced Energy Materials · doi:10.1002/aenm.202102638 · Cited by 54 · see citing works
Self-citationYu, Z., Singh, B., Yu, Y., & Nazar, L. F. (2025) · Nature Materials · doi:10.1038/s41563-025-02238-2 · Cited by 49 · see citing works
Common workflows for computing material properties using different quantum engines
Huber, S. P., Bosoni, E., Bercx, M., Bröder, J., Degomme, A., Dikan, V., Eimre, K., Flage−Larsen, E., Garcı́a, A., Genovese, L., et al. (2021) · npj Computational Materials · doi:10.1038/s41524-021-00594-6 · Cited by 40 · see citing works
SimStack: An Intuitive Workflow Framework
Rêgo, C. R. C., Schaarschmidt, J., Schlöder, T., Penaloza-Amion, M., Bag, S., Neumann, T., Strunk, T., & Wenzel, W. (2022) · Frontiers in Materials · doi:10.3389/fmats.2022.877597 · Cited by 38 · see citing works
Self-citationAn active learning approach to model solid-electrolyte interphase formation in Li-ion batteries
Soleymanibrojeni, M., Rêgo, C. R. C., Esmaeilpour, M., & Wenzel, W. (2023) · Journal of Materials Chemistry A · doi:10.1039/d3ta06054c · Cited by 31 · see citing works
Self-citationTowards high-throughput many-body perturbation theory: efficient algorithms and automated workflows
Bonacci, M., Qiao, J., Spallanzani, N., Marrazzo, A., Pizzi, G., Molinari, E., Varsano, D., Ferretti, A., & Prezzi, D. (2023) · npj Computational Materials · doi:10.1038/s41524-023-01027-2 · Cited by 20 · see citing works
Bekemeier, S., Rêgo, C. R. C., Mai, H. L., Saikia, U., Waseda, O., Apel, M., Arendt, F., Aschemann, A., Bayerlein, B., Courant, R., et al. (2025) · Advanced Engineering Materials · doi:10.1002/adem.202402149 · Cited by 19 · see citing works
Self-citationArtificial Intelligence, Scientific Discovery, and Product Innovation
Toner-Rodgers, A. (2024) · arXiv · doi:10.48550/arxiv.2412.17866 · Cited by 17 · see citing works
Guedes‐Sobrinho, D., Silveira, D. N., de Araújo, L. O., Dalmedico, J. F., Wenzel, W., Pramudya, Y., Piotrowski, M. J., & Rêgo, C. R. C. (2023) · Scientific Reports · doi:10.1038/s41598-023-31220-8 · Cited by 14 · see citing works
Self-citationAutomated workflow for analyzing thermodynamic stability in polymorphic perovskite alloys
de Araújo, L. O., Rêgo, C. R. C., Wenzel, W., Piotrowski, M. J., Dias, A. C., & Guedes‐Sobrinho, D. (2024) · npj Computational Materials · doi:10.1038/s41524-024-01320-8 · Cited by 12 · see citing works
Self-citationTÜLU 3: Pushing Frontiers in Open Language Model Post-Training
Lambert, N., Morrison, J., Pyatkin, V., Huang, S., Ivison, H., Brahman, F., Miranda, L. J. V., Liu, A., Dziri, N., Lyu, S., et al. (2024) · arXiv · doi:10.48550/arxiv.2411.15124 · Cited by 12 · see citing works
Kimi k1.5: Scaling Reinforcement Learning with LLMs
Kimi Team, Du, A., Gao, B., Xing, B., Jiang, C., Chen, C., Li, C., Xiao, C., Du, C., Liao, C., et al. (2025) · arXiv · doi:10.48550/arxiv.2501.12599 · Cited by 12 · see citing works
Dalmedico, J. F., Silveira, D. N., de Araújo, L. O., Wenzel, W., Rêgo, C. R. C., Dias, A. C., Guedes‐Sobrinho, D., & Piotrowski, M. J. (2024) · ChemPhysChem · doi:10.1002/cphc.202400118 · Cited by 10 · see citing works
Self-citationFour ways to power-up AI for drug discovery
King, A. (2025) · Nature · doi:10.1038/d41586-025-00602-5 · Cited by 9 · see citing works
Materials Cloud three-dimensional crystals database (MC3D)
Huber, S. P., Bercx, M., Hörmann, N., Uhrin, M., Pizzi, G., & Marzari, N. (2022) · NCCR MARVEL · doi:10.24435/materialscloud:rw-t0 · Cited by 9 · see citing works
Detecting LLM Hallucinations Using Monte Carlo Simulations on Token ProbabilitiesPreprint
Ledger, G., & Mancinni, R. (2024) · doi:10.36227/techrxiv.171822396.61518693/v1 · Cited by 7 · see citing works
An Automatized Simulation Workflow for Powder Pressing Simulations Using SimStack
Mieller, B., Valavi, M., & Rêgo, C. R. C. (2024) · Advanced Engineering Materials · doi:10.1002/adem.202400872 · Cited by 7 · see citing works
Self-citationA Systematic Study on the Potentials and Limitations of LLM-assisted Software Development
Michelutti, C., Eckert, J., Monecke, M., Klein, J., & Glesner, S. (2024) · International Conference on Foundation and Large Language Models (FLLM) · doi:10.1109/fllm63129.2024.10852455 · Cited by 6 · see citing works
Bastos, C. M. O., Dias, A. C., de Brito, A. C. F., Barcelos, I. D., da Rosa, A. L., Silveira, D. N., Piotrowski, M. J., Wenzel, W., Rêgo, C. R. C., & Guedes‐Sobrinho, D. (2025) · Journal of Materials Chemistry A · doi:10.1039/d5ta02721g · Cited by 3 · see citing works
Self-citation
Flagged first, then by citations
Research works printed but not linked: 1
- Brown, T. et al. Language models are few-shot learners. Adv. Neural Inf. Process. Syst. 33, 1877-1901 (2020).
Cited by
Sorted by citations
Masgent: an AI-assisted materials simulation agent
Liu, G., Yang, S., & Zhong, Y. (2026) · Digital Discovery · doi:10.1039/d6dd00043f · Cited by 2 · see citing works
Liu, G., Ma, X., Zhang, L., Pascasio, J., Yang, J., Chen, Y., & Li, W.-L. (2026) · Digital Discovery · doi:10.1039/d6dd00273k · Cited by 2 · see citing works
Computational materials agents: from task demonstrations to executable scientific workflows
Hu, C., & Wang, Z. (2026) · AI Agent · doi:10.20517/aiagent.2026.38 · Cited by 2 · see citing works
Alviz-Meza, A., Valencia-Arias, A., Rojas-Flores, S., & Diaz, F. (2026) · Data · doi:10.3390/data11070180
Silva, G. R. D., Salah, M., Machado, M. M., Dias, A. C., Bastos, C. M. O., Piotrowski, M. J., Rêgo, C. R. C., & Guedes‐Sobrinho, D. (2026) · The Journal of Physical Chemistry A · doi:10.1021/acs.jpca.6c05562
Self-citation
- Authors
- Soleymanibrojeni, M., Aydin, R., Guedes-Sobrinho, D., Dias, A. C., Piotrowski, M. J., Wenzel, W., & Rêgo, C. R. C.
- Affiliations
- Helmholtz-Zentrum Hereon (DE)
- German Research Centre for Artificial Intelligence (DE)
- Universidade Federal do Paraná (BR)
- Universidade de Brasília (BR)
- Universidade Federal de Pelotas (BR)
- Karlsruhe Institute of Technology (DE)
- Corresponding author
- Rêgo, Celso Ricardo Caldeira
- ORCID
- Venue
- Communications Materials
- Published
- 2026-04-29
- Volume
- 7
- Issue
- 1
- ISSN
- 2662-4443 (ISSN-L)
- OpenAlex
- W7159588102
- Cited by
- 5
- References
- 36
- Citations by year
- 2026: 5
- FWCI
- 6.78
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- Communications Materials (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion, cc-by)
- Repository KITopen (Karlsruhe Institute of Technology) (submittedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- arXiv (Cornell University) (cc-by)
- KITopen (cc-by)
- Funders
- Helmholtz Association (HGF ZT-I-PF-5-261 GENIUS)
- Topics
- Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
- Social Sciences › Decision Sciences › Information Systems and Management › Scientific Computing and Data Management
- Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Advanced Chemical Physics Studies
- Keywords
- agentic AI, Quantum ESPRESSO, knowledge graphs, large language models, error recovery, zero-shot generation, integrated computational materials engineering, materials discovery, hallucination mitigation, workflow automation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
