C.E. Michoski

University of Texas at Austin | UT

Extracting reliable information from diagnostic data in tokamaks is critical for understanding, analyzing, and controlling the behavior of fusion plasmas and validating models describing that behavior. Recent interest within the fusion community has focused on the use of principled statistical methods, such as Gaussian Process Regression (GPR), to...

A kernel attention module (KAM) is presented for the task of EEG-based emotion classification using neural network based models. In this study, it is shown that the KAM method can lead to more efficient and accurate models using only a single parameter design. This additional parameter can be leveraged as an interpretable scalar quantity for examin...

Data-driven science and technology offer transformative tools and methods to science. This review article highlights the latest development and progress in the interdisciplinary field of data-driven plasma science (DDPS), i.e., plasma science whose progress is driven strongly by data and data analyses. Plasma is considered to be the most ubiquitous...

In this work, a parameter efficient attention module is developed for the task of emotion classification as well as improved model interpretability based on EEG source data. Inspired by the self-attention mechanism used in transformers, we propose a Monotonicity Constrained Attention Module (MCAM) that can help incorporate different priors easily o...

In this work, a kernel attention module is presented for the task of EEG-based emotion classification with neural networks . The proposed module utilizes a self-attention mechanism by performing a kernel trick, demanding significantly fewer trainable parameters and computations than standard attention modules. The design also provides a scalar for...

In this work, a parameter-efficient attention module is presented for emotion classification using a limited, or relatively small, number of electroencephalogram (EEG) signals. This module is called the Monotonicity Constrained Attention Module (MCAM) due to its capability of incorporating priors on the monotonicity when converting features’ Gram m...

In this work, a kernel attention module is presented for the task of EEG-based emotion classification with neural networks. The proposed module utilizes a self-attention mechanism by performing a kernel trick, demanding significantly fewer trainable parameters and computations than standard attention modules. The design also provides a scalar for q...

In this work, a parameter-efficient attention module is presented for emotion classification using a limited, or relatively small, number of electroencephalogram (EEG) signals. This module is called the Monotonicity Constrained Attention Module (MCAM) due to its capability of incorporating priors on the monotonicity when converting features' Gram m...

This paper reports on the development of reduced models for electron temperature gradient (ETG) driven transport in the pedestal. Model development is enabled by a set of 61 nonlinear gyrokinetic simulations with input parameters taken from pedestals in a broad range of experimental scenarios. The simulation data have been consolidated in a new dat...

Data science and technology offer transformative tools and methods to science. This review article highlights latest development and progress in the interdisciplinary field of data-driven plasma science (DDPS). A large amount of data and machine learning algorithms go hand in hand. Most plasma data, whether experimental, observational or computatio...

DIII-D physics research addresses critical challenges for the operation of ITER and the next generation of fusion energy devices. This is done through a focus on innovations to provide solutions for high performance long pulse operation, coupled with fundamental plasma physics understanding and model validation, to drive scenario development by int...

We present a new method for formulating closures that learn from kinetic simulation data. We apply this method to phase mixing in a simple gyrokinetic turbulent system – temperature-gradient-driven turbulence in an unsheared slab. The closure, called the learned multi-mode (LMM) closure, is constructed by, first, extracting an optimal basis from a...

This paper reports on the development of reduced models for electron temperature gradient (ETG) driven transport in the pedestal. Model development is enabled by a set of 61 nonlinear gyrokinetic simulations with input parameters taken from the pedestals in a broad range of experimental scenarios. The simulation data has been consolidated in a new...

We report on a detailed study of magnetic fluctuations in the JET pedestal, employing basic theoretical considerations, gyrokinetic simulations, and experimental fluctuation data to establish the physical basis for their origin, role, and distinctive characteristics. We demonstrate quantitative agreement between gyrokinetic simulations of microtear...

In this paper we propose a dual stream neural network (DSNN) for classifying arbitrary collections of functional neuroimaging signals for the purpose of brain computer interfaces (BCIs). In the DSNN the first stream is an end-to-end classifier taking raw time-dependent signals as input and generating feature identification signatures from them. The...

We report on a detailed study of magnetic fluctuations in the JET pedestal, employing basic theoretical considerations, gyrokinetic simulations, and experimental fluctuation data, to establish the physical basis for their origin, role, and distinctive characteristics. We demonstrate quantitative agreement between gyrokinetic simulations of microtea...

The Moment Preserving Constrained Resampling (MPCR) algorithm for particle resampling is introduced and applied to particle-in-cell (PIC) methods to increase simulation accuracy, reduce compute cost, and/or avoid numerical instabilities. The general algorithm partitions the system space into smaller subsets and resamples the distribution within eac...

Recent work on solving partial differential equations (PDEs) with deep neural networks (DNNs) is presented. The paper reviews and extends some of these methods while carefully analyzing a fundamental feature in numerical PDEs and nonlinear analysis: irregular solutions. First, the Sod shock tube solution to the compressible Euler equations is discu...

A coordinate transformation technique between straight magnetic field line coordinate system (Ψ, θ) and Cartesian coordinate system (R, Z) is presented employing a Solov'ev solution of the Grad‐Shafranov equation. Employing the equilibrium solution, the poloidal magnetic flux Ψ(R, Z) of a diverted tokamak, magnetic field line equation is solved com...

The FY19 Theory Performance Target Understanding the relevant turbulent transport mechanisms at the edge of a high-performance tokamak is essential for predicting and optimizing the H-mode pedestal structure in future burning plasma devices. Global electromagnetic gyrokinetic simulations will be performed based on representative experimental pedest...

We present a novel phase mixing closure for a simple turbulent system-ion / electron temperature gradient driven turbulence in an unsheared slab. The closure is motivated by the simple notion that a fluid system with n degrees of freedom (i.e. n fields) may benefit from retaining n degrees of freedom in the fluid closure. This is particularly true...

An explicit implementation of the hybridizable discontinuous Galerkin (HDG) method for solving the nonlinear shallow water equations is presented. We follow the common construction of the implicit HDG for nonlinear conservation laws, and then explain the differences between the explicit formulation and the implicit version. For the implicit impleme...

As high performance computing moves towards the exascale computing regime, applications are required to expose increasingly fine grain parallelism to efficiently use next generation supercomputers. Intended as a solution to the programming challenges associated with these architectures, High Performance ParalleX (HPX) is a task-based C++ runtime, w...

Recent work has introduced a simple numerical method for solving partial differential equations (PDEs) with deep neural networks (DNNs). This paper reviews and extends the method while applying it to analyze one of the most fundamental features in numerical PDEs and nonlinear analysis: irregular solutions. First, the Sod shock tube solution to comp...

Emerging Deep Learning (DL) applications introduce heavy I/O workloads on computer clusters. The inherent long lasting, repeated, and random file access pattern can easily saturate the metadata and data service and negatively impact other users. In this paper, we present FanStore, a transient runtime file system that optimizes DL I/O on existing ha...

This paper presents experiences using Intel's KNL MIC platform on hardware that will be available in the Stampede 2 cluster launching in Summer 2017. We focus on 1) porting of existing scientific software; 2) observing performance of this software. Additionally, we comment on both the ease of use of KNL and observed performance of KNL as compared t...

In simulations of plasmas using particle-in-cell methods, it is often advantageous to resample the particle distribution function to increase simulation accuracy, to reduce the computing cost, or to avoid numerical instabilities. An algorithm for down-sampling the particles to a smaller number is proposed here. To minimize noise introduced by the d...

In this work we provide an extension of the classical von Neumann stability analysis for high-order accurate discontinuous Galerkin methods applied to generalized nonlinear convection–reaction–diffusion systems. We provide a partial linearization under which a sufficient condition emerges that guarantees stability in this context. The stability beh...

In this paper we introduce a hybridized discontinuous Galerkin (HDG) method for solving nonlinear Korteweg–de Vries type equations. Similar to a standard HDG implementation, we first express the approximate variables and numerical fluxes inside each element in terms of the approximate traces of the scalar variable (u), and its first derivative ((Fo...

A new discontinuous Galerkin (DG) method is introduced that seamlessly merges exact geometry with high-order solution accuracy. This new method is called the blended isogeometric discontinuous Galerkin (BIDG) method. The BIDG method contrasts with existing high-order accurate DG methods over curvilinear meshes (e.g. classical isoparametric DG metho...

The diffusion equation in anisotropic and nonhomogeneous media arises in the study of flow through porous media with sharp material interfaces. We discuss the solution of this problem by a hybrid discontinuous Galerkin (HDG) method. The method can be applied in three steps. First, we use a condensation technique to derive the scalar variable and th...

Drift wave turbulence driven by the steep electron and ion temperature gradients in H-mode divertor tokamaks produce scattering of the RF waves used for heating and current drive. The X-ray emission spectra produced by the fast electrons require the turbulence broaden RF wave spectrum. Both the 5 GHz Lower Hybrid waves and the 170 GHz electron cycl...

In this work, we consider the application of Discontinuous Galerkin (DG) solutions to open channel flow problems, governed by two-dimensional shallow water equations (SWE), with solid curved wall boundaries on which the no-normal flow boundary conditions are prescribed. A commonly used approach consists of straightforwardly imposing the no-normal f...

Nonlinear systems of equations demonstrate complicated regularity features that are often obfuscated by overly diffuse numerical methods. Using a discontinuous Galerkin finite element method, we study a nonlinear system of advection-diffusion-reaction equations and aspects of its regularity. For numerical regularization, we present a family of solu...

A new parallel discontinuous Galerkin solver, called ArcOn, is developed to describe the intermittent turbulent transport of filamentary blobs in the scrape-off layer (SOL) of fusion plasma. The model is comprised of an elliptic subsystem coupled to two convection-dominated reaction–diffusion–convection equations. Upwinding is used for a class of n...

We present a class of chemical reactor systems, modeled numerically using a fractional multistep method between the reacting and diffusing modes of the system, subsequently allowing one to utilize algebraic techniques for the resulting reactive subsystems. A mixed form discontinuous Galerkin method is presented with implicit and explicit (IMEX) tim...

Naturally occurring error fields as well as resonant magnetic perturbations applied for stability control are known to cause magnetic field-line chaos in the scrape-off layer (SOL) region of tokamaks. Here, 2D simulations with the BOUT++ simulation framework are used to investigate the effect of the field-line chaos on the SOL and in particular on...

We present a comprehensive assessment of nodal and hybrid modal/nodal discontinuous Galerkin (DG) finite element solutions on a range of unstructured meshes to nonlinear shallow water flow with smooth solutions. The nodal DG methods on triangles and a tensor-product nodal basis on quadrilaterals are considered. The hybrid modal/nodal DG methods uti...

We present numerical methods for a system of equations consisting of the two dimensional Saint–Venant shallow water equations (SWEs) fully coupled to a completely generalized Exner formulation of hydrodynamically driven sediment discharge. This formulation is implemented by way of a discontinuous Galerkin (DG) finite element method, using a Roe Flu...

Storm surge due to hurricanes and tropical storms can result in significant loss of life, property damage, and long-term damage to coastal ecosystems and landscapes. Computer modeling of storm surge can be used for two primary purposes: forecasting of surge as storms approach land for emergency planning and evacuation of coastal populations, and hi...

We present a family of p-enrichment schemes. These schemes may be separated into two basic classes: the first, called \emph{fixed tolerance schemes}, rely on setting global scalar tolerances on the local regularity of the solution, and the second, called \emph{dioristic schemes}, rely on time-evolving bounds on the local variation in the solution....

We introduce a fully generalized quiescent chemical reactor system in arbitrary space $\vdim =1,2$ or 3, with $n\in\mathbb{N}$ chemical constituents $\alpha_{i}$, where the character of the numerical solution is strongly determined by the relative scaling between the local reactivity of species $\alpha_{i}$ and the local functional diffusivity $\ma...

We present a solution to the conservation form (Eulerian form) of the quantum hydrodynamic equations which arise in chemical dynamics by implementing a mixed/discontinuous Galerkin (MDG) finite element numerical scheme. We show that this methodology is stable, showing good accuracy and a remarkable scale invariance in its solution space. In additio...

We present a generalized discontinuous Galerkin method for a multicomponent compressible barotropic Navier–Stokes system of equations. The system presented has a functional viscosity ν which depends on the pressure p=p(ρ,μi) of the flow, with the density ρ and the local concentration μi. High order Runge–Kutta time-discretization techniques are emp...

We prove the global existence and uniqueness of strong solutions for a compressible multifluid described by the barotropic Navier-Stokes equations in dim = 1. The result holds when the diffusion coefficient depends on the pressure. It relies on a global control in time of the L 2 norm of the space derivative of the density, via a new kind of entrop...