NCSI: Computational Feedback Control Investigation for the HITSIU Device

Status	Completed
Mentor Name	Alan Kaptanoglu
Mentor's XSEDE Affiliation	Startup Allocation, PI
Mentor Has Been in XSEDE Community	Less than 1 year
Project Title	Computational Feedback Control Investigation for the HITSIU Device
Summary	The apprentice will be responsible for investigating if feedback control is feasible on the upcoming HITSIU (fusion) experiment using the diagnostic surface probes. The student will use the 3D magnetohydrodynamics (MHD) code, NIMROD, and the full simulated experiment to determine the potential for controlling the magnetic field profile in this device during a plasma discharge. The project will conclude on requirements on experimental hardware and plasma parameters necessary for magnetic field profile control.
Job Description	Considerable time will be spent learning about plasma physics, magnetohydrodynamics, and computational physics. Actual duties will include: running simulations of the experiment, writing code in Fortran/matlab/MPI/openMP on HPC architectures, data analysis including singular value decompositions, fourier transforms, and machine learning methods, and implementing basic feedback control algorithms like extremum-seeking-control (ESC) into the code. The scientific application of this work includes magnetic and current profile control in toroidal plasma experiments which means significant improved understanding of confinement concepts for fusion energy. It also would provide evidence that the "inductive helicity injectors" on the HITSIU device can be used as actuators to mitigate plasma instability formation.
Computational Resources	The student will use the XSEDE startup allocation (on Comet and Bridges) to run the 3D magnetohydrodynamic (MHD) code, NIMROD, which requires high-performance-computing (HPC) resources and parallel programming to simulates plasmas on timescale comparable to experimental ones. Typical runs take ~ a day on 50-100 cores. The combined 1.5 TB of hard drive space from Bridges pylon and Oasis will be used to do long term storage of the data output by these runs, which typically are on the order of ~ 100 GB.
Contribution to Community
Position Type	Apprentice
Training Plan	The student will be provided with some simple NIMROD test cases, data analysis scripts, and installed code on the relevant XSEDE resources. The first month will be spent on getting the student up to speed on the use of high-performance-computing (HPC) resources: getting comfortable with Unix, submitting batch jobs, and running some simple NIMROD test cases in order to get familiar with the HITSIU experiment and the data analysis scripts. Some reading will be assigned to get familiar with magnetohydrodynamics (MHD) and the specifics of HITSIU. The student already has a very strong background in different areas of physics, and some computational physics + coding experience. Because this project is fairly involved, we expect to continue this work into Fall 2019.
Student Prerequisites/Conditions/Qualifications
Duration	Summer
Start Date	06/24/2019
End Date	08/30/2019