We present an ultrafast neural network (NN) model, QLKNN, which predicts core tokamak transport heat and particle fluxes. QLKNN is a surrogate model based on a database of 300 million flux calculations of the quasilinear gyrokinetic transport model QuaLiKiz. The database covers a wide range of realistic tokamak core parameters. Physical features such as the existence of a critical gradient for the onset of turbulent transport were integrated into the neural network training methodology. We have coupled QLKNN to the tokamak modelling framework JINTRAC and rapid control-oriented tokamak transport solver RAPTOR. The coupled frameworks are demonstrated and validated through application to three JET shots covering a representative spread of H-mode operating space, predicting turbulent transport of energy and particles in the plasma core. JINTRAC-QLKNN and RAPTOR-QLKNN are able to accurately reproduce JINTRAC-QuaLiKiz T i,e and n e profiles, but 3 to 5 orders of magnitude faster. Simulations which take hours are reduced down to only a few tens of seconds. The discrepancy in the final source-driven predicted profiles between QLKNN and QuaLiKiz is on the order 1%-15%. Also the dynamic behaviour was well captured by QLKNN, with differences of only 4%-10% compared to JINTRAC-QuaLiKiz observed at mid-radius, for a study of density buildup following the L-H transition. Deployment of neural network surrogate models in multi-physics integrated tokamak modelling is a promising route towards enabling accurate and fast tokamak scenario optimization, Uncertainty Quantification, and control applications.

Ion cyclotron emission (ICE) is detected during edge localised modes (ELMs) in the KSTAR tokamak at harmonics of the proton cyclotron frequency in the outer plasma edge. The emission typically chirps downward (occasionally upward) during ELM crashes, and is driven by confined 3MeV fusion-born protons that have large drift excursions from the plasma core. We exploit fully kinetic simulations at multiple plasma densities to match the time-evolving features of the chirping ICE. This yields a unique, very high time resolution (< 1µs) diagnostic of the collapsing edge pedestal density. PACS numbers: 52.35.Hr, 52.35.Qz, 52.55.Fa, 52.55.Tn Understanding the physics of edge localised modes (ELMs) [1][2][3][4] in magnetically confined fusion (MCF) plasmas is crucial for the design of future fusion power plants. The same is true of the physics of the energetic ions born at MeV energies [5,6] from fusion reactions between fuel ions in the multi keV thermal plasma. The crash of an ELM involves impulsive relaxation of the edge magnetic field, releasing energy and particles from the plasma at levels which may not be compatible with sustained operation of the next step fusion experiment, ITER [7,8]. The confinement of fusion-born ions while they release energy, collisionally or otherwise, to the thermal plasma, was a key physics objective of the unique deuterium-tritium plasma experiments in TFTR [9] and JET [10], and will be central to ITERs research programme. Here we report an unexpected conjunction of ELM physics with fusion-born ion physic. We show how this can be exploited as a diagnostic of plasma edge density with unique, very high (< 1µs) time resolution. This is achieved through particle orbit studies combined with first principles kinetic plasma simulations that explain high-time-resolution measurements of ion cyclotron emission (ICE) from the medium-size tokamak KSTAR [11]. We show that ICE from KSTAR deuterium plasmas is driven by a small subset of the fusion-born proton population, originating in the core of the plasma and passing through the edge region where they radiate collectively through the magnetoacoustic cyclotron instability (MCI) [13][14][15][16][17][18][19][20][21][22]. The MCI can occur because of the spatially localised population inversion in velocity space that is caused by the large drift excursions of 3.0 MeV fusionborn protons on deep passing orbits. Our simulations of the MCI in its saturated nonlinear regime show that the frequency spectrum excited depends strongly on the plasma density. By comparing MCI spectra simulated at different densities with high-time-resolution measurements of ICE spectra during ELM crashes in KSTAR, we are able to infer the time evolution of the collapsing edge density at sub-microsecond resolution, which is unprecedented. Recently, ICE has been detected from the outer mid-plane of KSTAR [23-25], with spectral peak frequencies at local proton cyclotron harmonics; see e.g. Fig. 1. The only energetic protons in KSTAR plasmas are produced in deuteron-deuteron (D-D) ...

The solar wind provides a natural laboratory for observations of MHD turbulence over extended temporal scales. Here, we apply a model independent method of differencing and rescaling to identify self-similarity in the Probability Density Functions (PDF) of fluctuations in solar wind bulk plasma parameters as seen by the WIND spacecraft. Whereas the fluctuations of speed v and IMF magnitude B are multi-fractal, we find that the fluctuations in the ion density ρ, energy densities B 2 and ρv 2 as well as MHD-approximated Poynting flux vB 2 are mono-scaling on the timescales up to 26 hours. The single curve, which we find to describe the fluctuations PDF of all these quantities up to this timescale, is non-Gaussian. We model this PDF with two approaches-Fokker-Planck, for which we derive the transport coefficients and associated Langevin equation, and the Castaing distribution that arises from a model for the intermittent turbulent cascade.