Offline Contextual Bayesian Optimization for Nuclear Fusion

Nuclear fusion is regarded as the energy of the future since it presents the possibility of unlimited clean energy. One obstacle in utilizing fusion as a feasible energy source is the stability of the reaction. Ideally, one would have a controller for the reactor that makes actions in response to the current state of the plasma in order to prolong the reaction as long as possible. In this work, we make preliminary steps to learning such a controller. Since learning on a real world reactor is infeasible, we tackle this problem by attempting to learn optimal controls offline via a simulator, where the state of the plasma can be explicitly set. In particular, we introduce a theoretically grounded Bayesian optimization algorithm that recommends a state and action pair to evaluate at every iteration and show that this results in more efficient use of the simulator.

[1]  E A Baltz,et al.  Achievement of Sustained Net Plasma Heating in a Fusion Experiment with the Optometrist Algorithm , 2017, Scientific Reports.

[2]  Kirthevasan Kandasamy,et al.  Myopic Posterior Sampling for Adaptive Goal Oriented Design of Experiments , 2019, ICML.

[3]  Zoubin Ghahramani,et al.  Pareto Frontier Learning with Expensive Correlated Objectives , 2016, ICML.

[4]  Orso Meneghini,et al.  Orchestrating TRANSP Simulations for Interpretative and Predictive Tokamak Modeling with OMFIT , 2018 .

[5]  Andreas Krause,et al.  Contextual Gaussian Process Bandit Optimization , 2011, NIPS.

[6]  Stephen O. Dean.  A Piece of the Sun: The Quest for Fusion Energy by Daniel Clery , 2013 .

[7]  Julian Kates-Harbeck,et al.  Predicting disruptive instabilities in controlled fusion plasmas through deep learning , 2019, Nature.

[8]  B. J. Xiao,et al.  PSFC / JA-19-5 Machine learning for disruption warning on Alcator C-Mod , DIII-D , and EAST , 2019 .

[9]  Kirthevasan Kandasamy,et al.  Offline Contextual Bayesian Optimization , 2019, NeurIPS.

[10]  Robert J. Goldston.  An Indispensable Truth: How Fusion Power Can Save the Planet , 2012 .

[11]  Peter Auer,et al.  Using Confidence Bounds for Exploitation-Exploration Trade-offs , 2003, J. Mach. Learn. Res..

[12]  W. R. Thompson.  ON THE LIKELIHOOD THAT ONE UNKNOWN PROBABILITY EXCEEDS ANOTHER IN VIEW OF THE EVIDENCE OF TWO SAMPLES , 1933 .

[13]  Kirthevasan Kandasamy,et al.  Parallelised Bayesian Optimisation via Thompson Sampling , 2018, AISTATS.

[14]  Christopher K. I. Williams,et al.  Gaussian Processes for Machine Learning (Adaptive Computation and Machine Learning) , 2005 .

[15]  Jet Efda Contributors,et al.  Automatic disruption classification based on manifold learning for real-time applications on JET , 2013 .

[16]  Peter I. Frazier,et al.  Bayesian Optimization with Expensive Integrands , 2018, SIAM J. Optim..

[17]  Shipra Agrawal,et al.  Thompson Sampling for Contextual Bandits with Linear Payoffs , 2012, ICML.

[18]  Jean Baccou,et al.  Bayesian Adaptive Reconstruction of Profile Optima and Optimizers , 2014, SIAM/ASA J. Uncertain. Quantification.

[19]  Benjamin Van Roy,et al.  An Information-Theoretic Analysis of Thompson Sampling , 2014, J. Mach. Learn. Res..

[20]  Jürgen Branke,et al.  Continuous multi-task Bayesian Optimisation with correlation , 2018, Eur. J. Oper. Res..

[21]  Jasper Snoek,et al.  Multi-Task Bayesian Optimization , 2013, NIPS.

[22]  Gary J. Weisel,et al.  An Indispensable Truth: How Fusion Power Can Save the Planet , 2011 .