277

Papers

10/10

Roadblocks Active

1

Connections

Unsupervised anomaly detection methods struggle to distinguish near-normal anomalies from normal patterns; Existing methods are vulnerable to noise contamination within normal samples

██████████ 0.7 plasma-disruption Preprint

ReadSaveConnections

Magnetic islands at the top and bottom of the H-mode pedestal, combined with strong screening of resonant fields in the gradient region, account for ELM suppression by n=2 RMPs in low-collisionality DIII-D plasmas; Narrow magnetic islands and enhanced collisional transport in the resistive foot of the pedestal produce the observed density pump-out (density reduction) in DIII-D

██████████ 0.9 elm-control 🔗 60 cited Preprint

ReadSaveConnections

Physics-aware Neural Operator Transformer (PNOT) enables fast, real-time reconstruction of divertor temperature fields with improved accuracy over conventional methods; Graph attention mechanism explicitly captures spatial physical dependencies in boundary heat-flux relations

██████████ 0.9 divertor-thermal Preprint

ReadSaveConnections

Resonant magnetic perturbations (RMPs) increase particle confinement and overall performance in DIII-D H-mode plasmas over a robust range of counter-current rotation; RMP-induced density pump-in effect is caused by reduction of turbulent transport

██████████ 0.8 elm-control 🔗 5 cited Preprint

ReadSaveConnections

Bidirectional autoregressive latent diffusion can predict evolution of multiple MHD fields (mass density, pressure, velocity, magnetic field components); Bidirectional flow enables self-supervised consistency metric for uncertainty and error estimation without ground truth access

██████████ 0.7 plasma-disruption Preprint

ReadSaveConnections

Static resonant magnetic perturbations (RMPs) can stabilize resistive tearing modes (m/n = 2/1) at moderate amplitudes without causing mode locking; RMP amplitude is critical in determining plasma response and tearing mode behavior; excessive amplitude leads to mode locking

██████████ 0.8 plasma-disruption 🔗 78 cited Preprint

ReadSaveConnections

Generative Meta-Learning with Human Feedback (GMHF) framework successfully bridges domain gaps when target domain data is unavailable or partial; Theoretical analysis derives generalization error bounds showing that aligning generated data distribution with expert beliefs regarding target physics significantly reduces risk

██████████ 0.5 turbulence-modeling Preprint

ReadSaveConnections

Toroidally oriented pump inlet passively exploits intrinsic neutral flows to reduce back-flow and improve particle-exhaust performance in tokamak divertors; Multi-species toroidal neutral 'wind' exists in detached tokamak divertors with strong toroidal alignment and kilometre-per-second velocities

██████████ 0.7 divertor-thermal Preprint

ReadSaveConnections

Semi-implicit MHD formulation enables robust nonlinear time-dependent computation in stellarators with large timesteps; 2D nodal spectral elements over poloidal plane combined with Fourier representation in toroidal angle enables efficient 3D modeling of non-axisymmetric configurations

██████████ 0.8 plasma-disruption Preprint

ReadSaveConnections

Compressive fluctuations (acoustic waves and compressible turbulence) increase fusion reaction rates through three mechanisms: hydrodynamic clumping, two-temperature effects, and kinetic streaming; Fusion power amplification from waves can exceed the power gain achieved by using equivalent wave energy for direct heating

██████████ 0.7 q-engineering Preprint

ReadSaveConnections