arXiv subjects

Rory Conlin

Publications and source records attributed to Rory Conlin.

24 records · Page 2Linked to original sources

arXiv2023

The DESC Stellarator Code Suite Part II: Perturbation and continuation methods

A new perturbation and continuation method is presented for computing and analyzing stellarator equilibria. The method is formally derived from a series expansion about the equilibrium condition $F \equiv J \times B - \nabla p = 0$, and an efficient algorithm for computing solutions to 2nd and 3rd order perturbations is developed. The method has been implemented in the DESC stellarator equilibrium code, using automatic differentiation to compute the required derivatives. Examples are shown demonstrating its use for computing complicated equilibria, perturbing a tokamak into a stellarator, and performing parameter scans in pressure, rotational transform and boundary shape in a fraction of the time required for a full solution.

physics.plasm-ph↗

arXiv2023

The DESC Stellarator Code Suite Part I: Quick and accurate equilibria computations

3D equilibrium codes are vital for stellarator design and operation, and high-accuracy equilibria are also necessary for stability studies. This paper details comparisons of two three-dimensional equilibrium codes, VMEC, which uses a steepest-descent algorithm to reach a minimum-energy plasma state, and DESC, which minimizes the magnetohydrodynamic (MHD) force error in real space directly. Accuracy as measured by satisfaction of MHD force balance is presented for each code, along with the computation time. It is shown that DESC is able to achieve more accurate solutions, especially near-axis. The importance of higher accuracy equilibria is shown in DESC's better agreement of stability metrics with asymptotic formulae. DESC's global Fourier-Zernike basis also yields the solution with analytic derivatives explicitly everywhere in the plasma volume, provides improved accuracy in the radial direction versus conventional finite differences, and allows for exponential convergence. Further, DESC can compute the same accuracy solution as VMEC in an order of magnitude less time.

physics.plasm-ph↗

arXiv2022

The DESC Stellarator Code Suite Part III: Quasi-symmetry optimization

The DESC stellarator optimization code takes advantage of advanced numerical methods to search the full parameter space much faster than conventional tools. Only a single equilibrium solution is needed at each optimization step thanks to automatic differentiation, which efficiently provides exact derivative information. A Gauss-Newton trust-region optimization method uses second-order derivative information to take large steps in parameter space and converges rapidly. With just-in-time compilation and GPU portability, high-dimensional stellarator optimization runs take orders of magnitude less computation time with DESC compared to other approaches. This paper presents the theory of the DESC fixed-boundary local optimization algorithm along with demonstrations of how to easily implement it in the code. Example quasi-symmetry optimizations are shown and compared to results from conventional tools. Three different forms of quasi-symmetry objectives are available in DESC, and their relative advantages are discussed in detail. In the examples presented, the triple product formulation yields the best optimization results in terms of minimized computation time and particle transport. This paper concludes with an explanation of how the modular code suite can be extended to accommodate other types of optimization problems.

physics.plasm-ph↗

arXiv2022

Exploration via Planning for Information about the Optimal Trajectory

Many potential applications of reinforcement learning (RL) are stymied by the large numbers of samples required to learn an effective policy. This is especially true when applying RL to real-world control tasks, e.g. in the sciences or robotics, where executing a policy in the environment is costly. In popular RL algorithms, agents typically explore either by adding stochasticity to a reward-maximizing policy or by attempting to gather maximal information about environment dynamics without taking the given task into account. In this work, we develop a method that allows us to plan for exploration while taking both the task and the current knowledge about the dynamics into account. The key insight to our approach is to plan an action sequence that maximizes the expected information gain about the optimal trajectory for the task at hand. We demonstrate that our method learns strong policies with 2x fewer samples than strong exploration baselines and 200x fewer samples than model free methods on a diverse set of low-to-medium dimensional control tasks in both the open-loop and closed-loop control settings.

cs.LG↗

arXiv2022

Greedy permanent magnet optimization

A number of scientific fields rely on placing permanent magnets in order to produce a desired magnetic field. We have shown in recent work that the placement process can be formulated as sparse regression. However, binary, grid-aligned solutions are desired for realistic engineering designs. We now show that the binary permanent magnet problem can be formulated as a quadratic program with quadratic equality constraints (QPQC), the binary, grid-aligned problem is equivalent to the quadratic knapsack problem with multiple knapsack constraints (MdQKP), and the single-orientation-only problem is equivalent to the unconstrained quadratic binary problem (BQP). We then provide a set of simple greedy algorithms for solving variants of permanent magnet optimization, and demonstrate their capabilities by designing magnets for stellarator plasmas. The algorithms can a-priori produce sparse, grid-aligned, binary solutions. Despite its simple design and greedy nature, we provide an algorithm that outperforms the state-of-the-art algorithms while being substantially faster, more flexible, and easier-to-use.

physics.plasm-ph↗

arXiv2022

Implementation of AI/Deep Learning Disruption Predictor into a Plasma Control System

This paper reports on advances to the state-of-the-art deep-learning disruption prediction models based on the Fusion Recurrent Neural Network (FRNN) originally introduced a 2019 Nature publication. In particular, the predictor now features not only the disruption score, as an indicator of the probability of an imminent disruption, but also a sensitivity score in real-time to indicate the underlying reasons for the imminent disruption. This adds valuable physics-interpretability for the deep-learning model and can provide helpful guidance for control actuators now that it is fully implemented into a modern Plasma Control System (PCS). The advance is a significant step forward in moving from modern deep-learning disruption prediction to real-time control and brings novel AI-enabled capabilities relevant for application to the future burning plasma ITER system. Our analyses use large amounts of data from JET and DIII-D vetted in the earlier NATURE publication. In addition to when a shot is predicted to disrupt, this paper addresses reasons why by carrying out sensitivity studies. FRNN is accordingly extended to use many more channels of information, including measured DIII-D signals such as (i) the n1rms signal that is correlated with the n =1 modes with finite frequency, including neoclassical tearing mode and sawtooth dynamics, (ii) the bolometer data indicative of plasma impurity content, and (iii) q-min, the minimum value of the safety factor relevant to the key physics of kink modes. The additional channels and interpretability features expand the ability of the deep learning FRNN software to provide information about disruption subcategories as well as more precise and direct guidance for the actuators in a plasma control system.

physics.plasm-ph↗