Search

281. Visual Analogy Extrapolation Challenge (VAEC)

Author(s):
Webb, Taylor; Dulberg, Zachary; Frankland, Steven; Petrov, Alexander; O'Reilly, Randall; Cohen, Jonathan
Abstract:
Extrapolation -- the ability to make inferences that go beyond the scope of one's experiences -- is a hallmark of human intelligence. By contrast, the generalization exhibited by contemporary neural network algorithms is largely limited to interpolation between data points in their training corpora. In this paper, we consider the challenge of learning representations that support extrapolation. We introduce a novel visual analogy benchmark that allows the graded evaluation of extrapolation as a function of distance from the convex domain defined by the training data. We also introduce a simple technique, context normalization, that encourages representations that emphasize the relations between objects. We find that this technique enables a significant improvement in the ability to extrapolate, considerably outperforming a number of competitive techniques.
Type:
Dataset and Image
Issue Date:
2020

282. Machine Learning Characterization of Alfvénic and Sub-Alfvénic Chirping and Correlation With Fast-Ion Loss at NSTX

Author(s):
Woods, B. J. Q.; Duarte, V. N.; Fredrickson, E. D.; Gorelenkov, N. N.; Podestà, M.; Vann, R. G. L.
Abstract:
Abrupt large events in the Alfvenic and sub-Alfvenic frequency bands in tokamaks are typically correlated with increased fast-ion loss. Here, machine learning is used to speed up the laborious process of characterizing the behavior of magnetic perturbations from corresponding frequency spectrograms that are typically identified by humans. The analysis allows for comparison between different mode character (such as quiescent, fixed frequency, and chirping, avalanching) and plasma parameters obtained from the TRANSP code, such as the ratio of the neutral beam injection (NBI) velocity and the Alfven velocity (v_inj./v_A), the q-profile, and the ratio of the neutral beam beta and the total plasma beta (beta_beam,i / beta). In agreement with the previous work by Fredrickson et al., we find a correlation between beta_beam,i and mode character. In addition, previously unknown correlations are found between moments of the spectrograms and mode character. Character transition from quiescent to nonquiescent behavior for magnetic fluctuations in the 50200-kHz frequency band is observed along the boundary v_phi ~ (1/4)(v_inj. - 3v_A), where v_phi is the rotation velocity.
Type:
Dataset
Issue Date:
December 2019

283. Machine Learning Characterization of Alfvénic and Sub-Alfvénic Chirping and Correlation With Fast-Ion Loss at NSTX

Author(s):
Woods, B. J. Q.; Duarte, V. N.; Fredrickson, E. D.; Gorelenkov, N. N.; Podestà, M.; Vann, R. G. L.
Abstract:
Abrupt large events in the Alfvenic and sub-Alfvenic frequency bands in tokamaks are typically correlated with increased fast-ion loss. Here, machine learning is used to speed up the laborious process of characterizing the behavior of magnetic perturbations from corresponding frequency spectrograms that are typically identified by humans. The analysis allows for comparison between different mode character (such as quiescent, fixed frequency, and chirping, avalanching) and plasma parameters obtained from the TRANSP code, such as the ratio of the neutral beam injection (NBI) velocity and the Alfven velocity (v_inj./v_A), the q-profile, and the ratio of the neutral beam beta and the total plasma beta (beta_beam,i / beta). In agreement with the previous work by Fredrickson et al., we find a correlation between beta_beam,i and mode character. In addition, previously unknown correlations are found between moments of the spectrograms and mode character. Character transition from quiescent to nonquiescent behavior for magnetic fluctuations in the 50200-kHz frequency band is observed along the boundary v_phi ~ (1/4)(v_inj. - 3v_A), where v_phi is the rotation velocity.
Type:
Dataset
Issue Date:
December 2019

284. Intrinsic Rotation in Axisymmetric Devices

Author(s):
T Stoltzfus-Dueck
Abstract:
Toroidal rotation is critical for fusion in tokamaks, since it stabilizes instabilities that can otherwise cause disruptions or degrade confinement. Unlike present-day devices, ITER might not have enough neutral-beam torque to easily avoid these instabilities. We must therefore understand how the plasma rotates intrinsically, that is, without applied torque. Experimentally, torque-free plasmas indeed rotate, with profiles that are often non-flat and even non-monotonic. The rotation depends on many plasma parameters including collisionality and plasma current, and exhibits sudden bifurcations (rotation reversals) at critical parameter values.Since toroidal angular momentum is conserved in axisymmetric systems, and since experimentally inferred momentum transport is much too large to be neoclassical, theoretical work has focused on rotation drive by nondiffusive turbulent momentum fluxes. In the edge, intrinsic rotation relaxes to a steady state in which the total momentum outflux from the plasma vanishes. Ion drift orbits, scrape-off-layer flows, separatrix geometry, and turbulence intensity gradient all play a role. In the core, nondiffusive and viscous momentum fluxes balance to set the rotation gradient at each flux surface. Although many mechanisms have been proposed for the nondiffusive fluxes, most are treated in one of two distinct but related gyrokinetic formulations. In a radially local fluxtube, appropriate for rho star <<1, the lowest-order gyrokinetic formulations exhibit a symmetry that prohibits nondiffusive momentum flux for nonrotating plasmas in an up- down symmetric magnetic geometry with no ExB shear. Many symmetry-breaking mechanisms have been identified, but none have yet been conclusively demonstrated to drive a strong enough flux to explain commonly observed experimental rotation profiles. Radially global gyrokinetic simulations naturally include many symmetry-breaking mechanisms, and have shown cases with experimentally relevant levels of nondiffusive flux. These promising early results motivate further work to analyze, verify, and validate.This article provides a pedagogical introduction to intrinsic rotation in axisymmetric devices. Intended for both newcomers to the topic and experienced practitioners, the article reviews a broad range of topics including experimental and theoretical results for both edge and core rotation, while maintaining a focus on the underlying concepts.
Type:
Dataset
Issue Date:
November 2019

285. Deep convolutional neural networks for multi-scale time-series classification and application to disruption prediction in fusion devices

Author(s):
Churchill, R.M; the DIII-D team
Abstract:
The multi-scale, mutli-physics nature of fusion plasmas makes predicting plasma events challenging. Recent advances in deep convolutional neural network architectures (CNN) utilizing dilated convolutions enable accurate predictions on sequences which have long-range, multi-scale characteristics, such as the time-series generated by diagnostic instruments observing fusion plasmas. Here we apply this neural network architecture to the popular problem of disruption prediction in fusion tokamaks, utilizing raw data from a single diagnostic, the Electron Cyclotron Emission imaging (ECEi) diagnostic from the DIII-D tokamak. ECEi measures a fundamental plasma quantity (electron temperature) with high temporal resolution over the entire plasma discharge, making it sensitive to a number of potential pre-disruptions markers with different temporal and spatial scales. Promising, initial disruption prediction results are obtained training a deep CNN with large receptive field ({$\sim$}30k), achieving an $F_1$-score of {$\sim$}91\% on individual time-slices using only the ECEi data.
Type:
Dataset
Issue Date:
October 2019

286. Calibrated VIC Model Parameters over CONUS

Author(s):
Yang, Yuan; Pan, Ming; Beck, Hylke; Fisher, Colby; Beighley, R. Edward; Kao, Shih-Chieh; Hong, Yang; Wood, Eric
Abstract:
Conventional basin-by-basin approaches to calibrate hydrologic models are limited to gauged basins and typically result in spatially discontinuous parameter fields. Moreover, the consequent low calibration density in space falls seriously behind the need from present-day applications like high resolution river hydrodynamic modeling. In this study we calibrated three key parameters of the Variable Infiltration Capacity (VIC) model at every 1/8° grid-cell using machine learning-based maps of four streamflow characteristics for the conterminous United States (CONUS), with a total of 52,663 grid-cells. This new calibration approach, as an alternative to parameter regionalization, applied to ungauged regions too. A key difference made here is that we tried to regionalize physical variables (streamflow characteristics) instead of model parameters whose behavior may often be less well understood. The resulting parameter fields no longer presented any spatial discontinuities and the patterns corresponded well with climate characteristics, such as aridity and runoff ratio. The calibrated parameters were evaluated against observed streamflow from 704/648 (calibration/validation period) small-to-medium-sized catchments used to derive the streamflow characteristics, 3941/3809 (calibration/validation period) small-to-medium-sized catchments not used to derive the streamflow characteristics) as well as five large basins. Comparisons indicated marked improvements in bias and Nash-Sutcliffe efficiency. Model performance was still poor in arid and semiarid regions, which is mostly due to both model structural and forcing deficiencies. Although the performance gain was limited by the relative small number of parameters to calibrate, the study and results here served as a proof-of-concept for a new promising approach for fine-scale hydrologic model calibrations.
Type:
Dataset
Issue Date:
4 September 2019

287. Analytic stability boundaries for compressional and global Alfven eigenmodes driven by fast ions. I. Interaction via ordinary and anomalous cyclotron resonances.

Author(s):
Lestz J.B., Gorelenkov N.N., Belova E.V., Tang S.X., Crocker N.A.
Abstract:
Conditions for net fast ion drive are derived for beam-driven, sub-cyclotron compressional (CAE) and global (GAE) Alfven eigenmodes, such as those routinely observed in spherical tokamaks such as NSTX(-U) and MAST. Both co- and counter-propagating CAEs and GAEs are investigated, driven by the ordinary and anomalous Doppler-shifted cyclotron resonance with fast ions. Whereas prior results were restricted to vanishingly narrow distributions in velocity space, broad parameter regimes are identified in this work which enable an analytic treatment for realistic fast ion distributions generated by neutral beam injection. The simple, approximate conditions derived in these regimes for beam distributions of realistic width compare well to the numerical evaluation of the full analytic expressions for fast ion drive. Moreover, previous results in the very narrow beam case are corrected and generalized to retain all terms in omega/omega_{ci} and k_{||}/kperp, which are often assumed to be small parameters but can significantly modify the conditions of drive and damping when they are non-negligible. Favorable agreement is demonstrated between the approximate stability criterion, simulation results, and a large database of NSTX observations of cntr-GAEs.
Type:
Dataset
Issue Date:
September 2019

288. Quasioptical modeling of wave beams with and without mode conversion: II. Numerical simulations of single-mode beams

Author(s):
K. Yanagihara, I. Y. Dodin, and S. Kubo
Abstract:
This work continues a series of papers where we propose an algorithm for quasioptical modeling of electromagnetic beams with and without mode conversion. The general theory was reported in the first paper of this series, where a parabolic partial differential equation was derived for the field envelope that may contain one or multiple modes with close group velocities. Here, we present a corresponding code PARADE (PAraxial RAy DEscription) and its test applications to single-mode beams in vacuum and also in inhomogeneous magnetized plasma. The numerical results are compared, respectively, with analytic formulas from Gaussian-beam optics and also with cold-plasma ray tracing. Quasioptical simulations of mode-converting beams are reported in the next, third paper of this series.
Type:
Dataset
Issue Date:
August 2019

289. Quasioptical modeling of wave beams with and without mode conversion: III. Numerical simulations of mode-converting beams

Author(s):
K. Yanagihara, I. Y. Dodin, and S. Kubo
Abstract:
This work continues a series of papers where we propose an algorithm for quasioptical modeling of electromagnetic beams with and without mode conversion. The general theory was reported in the first paper of this series, where a parabolic partial differential equation was derived for the field envelope that may contain one or multiple modes with close group velocities. In the second paper, we presented a corresponding code PARADE (PAraxial RAy DEscription) and its test applications to single-mode beams. Here, we report quasioptical simulations of mode-converting beams for the first time. We also demonstrate that PARADE can model splitting of two-mode beams. The numerical results produced by PARADE show good agreement with those of one-dimensional full-wave simulations and also with conventional ray tracing (to the extent that one-dimensional and ray-tracing simulations are applicable).
Type:
Dataset
Issue Date:
August 2019

290. Reduced Model for Direct Induction Startup Scenario Development on MAST-U and NSTX-U

Author(s):
Battaglia, D.J.; Thornton, A.J.; Gerhardt, S.P.; Kirk, A.; Kogan, L; Menard, J.E.
Abstract:
A reduced semi-empirical model using time-dependent axisymmetric vacuum field calculations is used to develop the prefill and feed-forward coil current targets required for reliable direct induction (DI) startup on the new MA-class spherical tokamaks, MAST-U and NSTX-U. The calculations are constrained by operational limits unique to each device, such as the geometry of the conductive elements and active coils, power supply specifications and coil heating and stress limits. The calculations are also constrained by semi-empirical models for sufficient breakdown, current drive, equilibrium and stability of the plasma developed from a shared database. A large database of DI startup on NSTX and NSTX-U is leveraged to quantify the requirements for achieving a reliable breakdown (Ip ~ 20 kA). It is observed that without pre-ionization, STs access the large E/P regime at modest loop voltage (Vloop) where the electrons in the weakly ionized plasma are continually accelerating along the open field lines. This ensures a rapid (order millisecond) breakdown of the neutral gas, even without pre-ionization or high-quality field nulls. The timescale of the initial increase in Ip on NSTX is reproduced in the reduced model provided a mechanism for impeding the applied electric field is included. Most discharges that fail in the startup phase are due to an inconsistency in the evolution of the plasma current (Ip) and equilibrium field or loss of vertical stability during the burn-through phase. The requirements for the self-consistent evolution of the fields in the weakly and full-ionized plasma states are derived from demonstrated DI startup on NSTX, NSTX-U and MAST. The predictive calculations completed for MAST-U and NSTX-U illustrate that the maximum Ip ramp rate (dIp/dt) in the early startup phase is limited by the voltage limits on the poloidal field coils on MAST-U and passive vertical stability on NSTX-U.
Type:
Dataset
Issue Date:
August 2019