Support
Contact us
Cristina Rea
Cristina Rea
Principal Research Scientist
Head, Data Science Division
NW16-186
Lab Website

Cristina is a Principal Research Scientist at MIT and Leader of the Plasma Science and Fusion Center (PSFC) Disruption Studies Group. Since October 2024, she has served as Head of the PSFC Data Science Division. Her research portfolio centers on interpretable and adaptive artificial intelligence for fusion applications, with a particular focus on disruption prediction, avoidance, and mitigation. Her work spans the development of large, multi-machine databases of disruption-relevant experimental data and the creation of data-driven plasma stability metrics designed for real-time deployment on national and international tokamak facilities. She leads the CFS-funded Disruption Strategies program, advancing predictive and control solutions for SPARC.

Cristina is also contributing to the development of a national data platform for enabling open and FAIR machine learning applications in fusion. In parallel, she serves as the PSFC Liaison Officer for the IAEA Collaborating Centre on AI in Fusion and Plasma Science, fostering international collaboration and adoption of machine-learning techniques across fusion research.

In addition, Cristina is the lead organizer of the PSFC Computational Physics School for Fusion Research, a DOE-funded program that equips early-career researchers with critical skills in high-performance computing, parallel programming, uncertainty quantification, and machine learning. She also serves as Deputy Lead for the Theory and Simulation Validation Major Topical Area within the DOE Tokamak Research Program.

Education

PhD, Physics, University of Padova, 2015

MS, Physics, University of Pisa, 2011

BS, Physics, University of Bologna, 2008

Interests

• Machine learning for plasma control and disruption prevention.
• Interpretable & physics-constrained surrogate modeling.
• Real-time proximity-to-instability monitoring.
• Hybrid (differentiable) physics–ML modeling and system identification.
• Open-source scientific software and multi-tokamak data ecosystems.

Cristina Rea leads the Data Science Division at the MIT Plasma Science and Fusion Center, where her group develops interpretable machine-learning methods for magnetic confinement fusion. Her work focuses on real-time monitoring of a plasma's proximity to instability and on optimizing plasma trajectories that are resilient to disruptions and transient instabilities, validated through experiments on the DIII-D and TCV tokamaks and cross-machine studies using Alcator C-Mod data. A central theme is hybrid, physics-constrained modeling to close the sim2real gap, also thanks to differentiable programming, and high-fidelity surrogate development. Her group also builds the open-source tools behind an open-access, multi-tokamak dataset for transient and machine-learning research, enabling data-driven discovery of instability physics relevant to SPARC and future fusion power plants.

Learn more about the Disruption Group's projects: https://disruptions.mit.edu/projects/

Google Scholar Link

Selected Publications
  • Cristina Rea, S. Benjamin; "A review of machine learning-driven studies of tearing modes in tokamaks". Phys. Plasmas 1 May 2026; 33 (5): 050501. Available: https://doi.org/10.1063/5.0325461
  • Cristina Rea, “Machine learning applications enabling fusion energy: Recent developments,” Journal of Fusion Energy, 44, no. 2, p. 39, 2025.
  • M. Wang, Cristina Rea, O. So, C. Dawson, D. T. Garnier, and C. Fan, “Active ramp‑down control and trajectory design for tokamaks with neural differential equations and reinforcement learning,” Communications Physics, vol. 8, p. 231, 2025. DOi: 10.1038/s42005-025-02146-6. [Online]. Available: https://doi.org/10.1038/s42005-025-02146-6, **.
  • M. Wang, A. Pau, Cristina Rea, et al., “Learning plasma dynamics and robust rampdown trajectories with predict-first experiments at TCV,” Nature Communications, vol. 16, no. 1, p. 8877, 2025. DOi: 10.1038/s41467-025-63917-x. [Online]. Available: https://www.nature.com/articles/s41467-025-63917-x, **.
  • Keith, C. Nagpal, Cristina Rea, and R. A. Tinguely, “Risk‑aware framework development for disruption prediction: Alcator C‑Mod and DIII‑D survival analysis,” Journal of Fusion Energy, vol. 43, no. 1, p. 21, 2024, **.
  • Maris, Cristina Rea, A. Pau, et al., “Correlation of the L‑mode density limit with edge collisionality,” Nuclear Fusion, vol. 65, no. 1, p. 016 051, Dec. 2024. DOi: 10.1088/1741-4326/ad90f0, **.
  • Maris, A. Wang, Cristina Rea, R. Granetz, and E. Marmar, “The impact of disruptions on the economics of a tokamak power plant,” Fusion Science and Technology, pp. 1–17, 2023. DOi: 10.1080/15361055.2023.2229675, **.
  • Cristina Rea, Mordijck, M. Murillo, D. Humphreys, B. Spears, and M. Barbarino, “Fusion. Chapter 9,” in Artificial In-telligence for Accelerating Nuclear Applications, Science and Technology, 2022.
  • Cristina Rea, Montes, K. Erickson, R. Granetz, and R. Tinguely, “A real‑time machine learning‑based disruption pre-dictor in DIII‑D,” Nuclear Fusion, vol. 59, no. 9, p. 096 016, 2019.

Selected Invited Talks and Conference Contributions:

Summer 2025. AI‑Accelerated Fusion: Advancements in Disruption Prevention and Performance Optimization. Invited Talk: Fusion For Energy Technology Development Planning Workshop, Barcelona, Spain.

Fall 2023. A review of explainable Machine Learning accelerating Fusion science. Invited Keynote: 4th Fusion HPC Workshop, 29‑30 November, 2023, https://hpcfusion.bsc.es/.

Summer 2023. Advances in Disruption Prevention via Machine Learning: challenges and opportunities. Invited Colloquium: 2023 IPP Symposium on ”New trends in experimental fusion plasma physics and plasma wall interaction”, 28 ‑ 29 June, Garching bei Munchen, Germany.

Summer 2023. A review of explainable Machine Learning accelerating Fusion science. Invited Talk: 2023 PhDiaFusion Sum-mer School, 19 ‑ 23 June, Niepołomice Royal Castle, Poland https://phdia2023.ifj.edu.pl/.

Fall 2022. Interpretable Machine Learning Accelerating Fusion Research. Invited Tutorial: 64th Annual Meeting of the APS Division of Plasma Physics, Spokane, Washington.

Media publications, webinars, and recent news can be consulted at: https://disruptions.mit.edu/news/

---

New prediction model could improve the reliability of fusion power plants (MIT News)

PSFC researcher successfully reprise CPS-FR summer school (PSFC News)
Taming fusion with machine learning (MIT News)
Three Questions: Robert Granetz on fusion research (MIT News)