A Singular Value Decomposition Method For Inverting Line Integrated Electron Density Measurements in Magnetically Confined Plasma Christopher Carey, The.

Slides:



Advertisements
Similar presentations
Ion Heating and Velocity Fluctuation Measurements in MST Sanjay Gangadhara, Darren Craig, David Ennis, Gennady Fiskel and the MST team University of Wisconsin-Madison.
Advertisements

A Neural Network Inversion Technique for Plasma Interferometry in Toroidal Fusion Devices Jerahmie Radder ECE 539 May 10, 2000.
9th TTF Spain September 11, 2002 B. J. Peterson, NIFS, Japan page 1 Radiative Collapse and Density Limit in the Large Helical Device.
Soft x-ray tomography on HT-7 tokamak K.Y. Chen, L.Q. Hu, Y.M. Duan HT-7.
1 Chapter 4 Interpolation and Approximation Lagrange Interpolation The basic interpolation problem can be posed in one of two ways: The basic interpolation.
Alternatives to Spherical Microphone arrays: Hybrid Geometries Aastha Gupta & Prof. Thushara Abhayapala Applied Signal Processing CECS To be presented.
Physics of fusion power Lecture 11: Diagnostics / heating.
Measuring the Wilson effect: observations and modeling with RHESSI H. Jabran Zahid M. D. Fivian H. S. Hudson.
Progress and New Results from the H-1NF Scanning Interferometer Scott Collis, George Warr, John Howard.
CISM solar wind metrics M.J. Owens and the CISM Validation and Metrics Team Boston University, Boston MA Abstract. The Center for Space-Weather Modeling.
Detecting and Tracking of Mesoscale Oceanic Features in the Miami Isopycnic Circulation Ocean Model. Ramprasad Balasubramanian, Amit Tandon*, Bin John,
Ordinary least squares regression (OLS)
Linear and generalised linear models
Physics of fusion power Lecture 10 : Running a discharge / diagnostics.
Basics of regression analysis
Outline (HIBP) diagnostics in the MST-RFP Relationship of equilibrium potential measurements with plasma parameters Simulation with a finite-sized beam.
Linear and generalised linear models Purpose of linear models Least-squares solution for linear models Analysis of diagnostics Exponential family and generalised.
Imaging Diagnostics at the H-1 National Plasma Fusion Research Facility Left: The coherence tomography system Above: Plasma emission reconstructions compared.
Physics of fusion power
Chapter 10 Real Inner Products and Least-Square (cont.)
Fast imaging of global eigenmodes in the H-1 heliac ABSTRACT We report a study of coherent plasma instabilities in the H-1 plasma using a synchronous gated.
N.Shi et al., address: Division seminar, April 9, 2014, Institute of Plasma Physics, Academy of Sciences. A novel electron density.
Physics “Advanced Electronic Structure” Pseudopotentials Contents: 1. Plane Wave Representation 2. Solution for Weak Periodic Potential 3. Solution.
10th ITPA TP Meeting - 24 April A. Scarabosio 1 Spontaneous stationary toroidal rotation in the TCV tokamak A. Scarabosio, A. Bortolon, B. P. Duval,
EISCAT Radar Summer School 15th-26th August 2005 Kiruna
Profile Measurement of HSX Plasma Using Thomson Scattering K. Zhai, F.S.B. Anderson, J. Canik, K. Likin, K. J. Willis, D.T. Anderson, HSX Plasma Laboratory,
What are helicons? Helicons are partially ionized RF discharges in a magnetic field. They are basically whistler modes confined to a cylinder. They are.
PROTO-SPHERA Diagnostics PROTO-SPHERA WORKSHOP Frascati March 18-19, 2002.
Orthogonality and Least Squares
1 B. C. Stratton, S. von Goeler, J. Robinson, and L. E. Zakharov Princeton Plasma Physics Laboratory, Princeton, New Jersey, USA D. Stutman and K. Tritz.
SUPA Advanced Data Analysis Course, Jan 6th – 7th 2009 Advanced Data Analysis for the Physical Sciences Dr Martin Hendry Dept of Physics and Astronomy.
M. Gelfusa 1 (16) Frascati March 2012 Validation of Polarimetric measurements on JET using advanced statistical analysis of the residuals M. Gelfusa,
A comparison of the ability of artificial neural network and polynomial fitting was carried out in order to model the horizontal deformation field. It.
Tunable, resonant heterodyne interferometer for neutral hydrogen measurements in tokamak plasmas * J.J. Moschella, R.C. Hazelton, M.D. Keitz, and C.C.
Modal Analysis of Rigid Microphone Arrays using Boundary Elements Fabio Kaiser.
Electron Density Distribution in HSX C. Deng, D.L. Brower Electrical Engineering Department University of California, Los Angeles J. Canik, S.P. Gerhardt,
Comparison of Ion Thermal Transport From GLF23 and Weiland Models Under ITER Conditions A. H. Kritz 1 Christopher M. Wolfe 1 F. Halpern 1, G. Bateman 1,
Solution of the Inverse Problem for Gravitational Wave Bursts Massimo Tinto JPL/CIT LIGO Seminar, October 12, 2004 Y. Gursel & M. Tinto, Phys. Rev. D,
Two problems with gas discharges 1.Anomalous skin depth in ICPs 2.Electron diffusion across magnetic fields Problem 1: Density does not peak near the.
APS DPP 2006 October Adaptive Extremum Seeking Control of ECCD for NTM Stabilization L. Luo 1, J. Woodby 1, E. Schuster 1 F. D. Halpern 2, G.
بسمه تعالی Fast Imaging of turbulent plasmas in the GyM device D.Iraji, D.Ricci, G.Granucci, S.Garavaglia, A.Cremona IFP-CNR-Milan 7 th Workshop on Fusion.
N OVIMIR A. P ABLANT M. B ITTER, L.F. D ELGADO -A PARICIO, M. G OTO, K.W. H ILL, S. L AZERSON, S. M ORITA, L. R OQUEMORE N OVIMIR A. P ABLANT M. B ITTER,
Physics of fusion power Lecture 12: Diagnostics / heating.
Active Control of MHDinstabilitiy 2002/11/19 S.Ohdachi et.al. Sawtooth-like phenomena in LHD S. Ohdachi, S.Yamamoto, K. Toi, K. Y.Watanabe, S.Sakakibara,
THE BEHAVIOUR OF THE HEAT CONDUCTIVITY COEFFICIENT AND THE HEAT CONVECTIVE VELOCITY AFTER ECRH SWITCH-ON (-OFF) IN T-10 V. F. Andreev, Yu. N. Dnestrovskij,
Geant4 Tracking Test (D. Lunesu)1 Daniela Lunesu, Stefano Magni Dario Menasce INFN Milano GEANT4 TRACING TESTs.
1 Reconstruction Algorithms for UHE Neutrino Events in Sea Water Simon Bevan.
045-05/rs PERSISTENT SURVEILLANCE FOR PIPELINE PROTECTION AND THREAT INTERDICTION Taming The Physics For Commercial Fusion Power Plants ARIES Team Meeting.
47th Annual Meeting of the Division of Plasma Physics, October 24-28, 2005, Denver, Colorado ECE spectrum of HSX plasma at 0.5 T K.M.Likin, H.J.Lu, D.T.Anderson,
Plasma MHD Activity Observations via Magnetic Diagnostics Magnetic islands, statistical methods, magnetic diagnostics, tokamak operation.
1) For equivalent ECRH power, off-axis heating results in lower stored energy and lower core temperature 2) Plasma flow is significantly reduced with off-axis.
1 Chapter 4 Interpolation and Approximation Lagrange Interpolation The basic interpolation problem can be posed in one of two ways: The basic interpolation.
1 ASIPP Sawtooth Stabilization by Barely Trapped Energetic Electrons Produced by ECRH Zhou Deng, Wang Shaojie, Zhang Cheng Institute of Plasma Physics,
Neutral beam ion loss measurement and modeling for NSTX D. S. Darrow Princeton Plasma Physics Laboratory American Physical Society, Division of Plasma.
Measurement of Electron Density Profile and Fluctuations on HSX C. Deng, D.L. Brower, W.X. Ding Electrical Engineering Department University of California,
Hard X-rays from Superthermal Electrons in the HSX Stellarator Preliminary Examination for Ali E. Abdou Student at the Department of Engineering Physics.
C. Deng and D.L. Brower University of California, Los Angeles J. Canik, D.T. Anderson, F.S.B. Anderson and the HSX Group University of Wisconsin-Madison.
Profiles of density fluctuations in frequency range of (20-110)kHz Core density fluctuations Parallel flow measured by CHERS Core Density Fluctuations.
SAWTOOTH AND M=1 MODE BEHAVIOUR IN FTU PELLET ENHANCED DISCHARGES
MARTe real-time acquisition system of a Two-Color Interferometer for
Prepared BY: Helwan University Faculty Of Engineering
M4 m4 WHAT IS A TOKAMAK? Graphic: EUROfusion, Reinald Fenke, CC BY 4.0,
First Experiments Testing the Working Hypothesis in HSX:
ILC Main Linac Alignment Simulations
Inverse Matrices and Systems
5.2 Least-Squares Fit to a Straight Line
6.1 Introduction to Chi-Square Space
Alfven Oscillations in the TUMAN-3M Tokamak Ohmic Regime
Computed Tomography (C.T)
Presentation transcript:

A Singular Value Decomposition Method For Inverting Line Integrated Electron Density Measurements in Magnetically Confined Plasma Christopher Carey, The Ohio State University Ivo Furno, Los Alamos National LaboratoryIntroduction Abstract Electron transport studies in magnetically confined plasmas require high temporal and spatial resolution measurements of electron density profiles. A common method used to perform this measurement is interferometry of a coherent electromagnetic wave across the plasma cross section. To calculate the electron density profile from this line integrated measurement an inverse problem must be solved. We have developed a novel method for inverting the line integrated interferometric measurement using information from a Thomson scattering measurements. Tokamak à Configuration Variable TCV has an elongation of 3, allowing for the formation of very strongly shaped plasmas The Tokamak à Configuration Variable (TCV) is housed at the Center for Research of Plasma Physics (CRPP) in Switzerland. Due to its unique geometry with an elongation of 3, i.e. three times taller than it is wide, TCV can produce a wide array of unique plasma shapes. Thus, TCV allows for the systematic study of very strongly shaped plasmas. Some examples of these plasma configurations can be seen below. Because the performance of a fusion reactor will depend significantly on the plasma shape, TCV is a vital tool for the development of a practical fusion reactor, allowing for the study of the effect of plasma shape on confinement times. SVD Inversion Method Problem Definition As can be seen on the right, the phase shift of an interferometer chord can be related to the line integral of the electron density along the path of the beam in the plasma. The 14 interferometer measurements form a system of inhomogeneous Fredholm equations of the first kind, which is always underdetermined. One method for solving this system is to expand the unknown parameter, n e, in orthogonal basis functions that depend only on the magnetic flux surface, ρ. This assumes that the electron density is constant on a magnetic flux surface, which is valid in the case that there is not strong MHD activity in the plasma. These functions are line integrated, and the resulting system of equations can then be solved for the expansion coefficients. Singular Value Decomposition for Inversion Our novel method for inversion of the interferometer data utilizes the singular value decomposition (SVD) of the Thompson scattering data to determine the basis functions for the expansion of the electron density. The Thomson scattering data has both temporal and spatial dependence. We can arrange this data in a matrix N e such that the spatial dependence is stored along the columns and the temporal dependence along the rows, as can be seen on the left. Using the SVD, N e can be decomposed into the product of three matrices; U, S, and V. The columns of U form a set of orthogonal spatial eigenmodes that contain the spatial information of N e and are often referred to as topos. Moreover, most of this information is stored in the first three to four topos. Thus, these topos make an excellent choice of basis functions for the expansion of the local electron density. Because we are able to use only a few basis functions, solving the set of 14 equations for the expansion coefficients, a i, becomes an over determined problem for which a least squares solution can be found. We have developed a new method for the inversion of line integrated interferometer measurements using the singular value decomposition. This method is particularly powerful because it uses the Thomson scattering data to invert the interferometric data; thus, coupling the information from these two diagnostics. The SVD inversion is effective when applied to simulated electron density profiles, and is more accurate and faster than inversion by minimizing the Fisher information. This novel inversion method is also effective when applied to TCV interferometer data, and is capable of revealing the hollowness of density profiles which previous methods would have destroyed.Conclusion Diagnostics TCV is equipped with a 14 chord interferometer and a Thomson scattering system for the measurement of electron density profiles. Data from the interferometer can be acquired at a rate of 25 kHz, which is fast enough to resolve features of interest in the density profiles. However, the interferometer gives a line integrated measurement of the electron density along the path of each laser in the plasma. Thus, to determine the local electron density profiles this measurement must be inverted. The Thomson scattering system can acquire data at a rate of up to 20 Hz, which is far to slow to resolve features in the electron density. Nevertheless, the Thomson scattering system has the advantage of giving a local measurement of the electron density at 25 positions along a vertical line in the plasma cross section. Far infrared interferometer and Thomson scattering on TCV. The lines represent the 14 laser beams probing the plasma and the red boxes show the 25 scattering volumes. Results on Artificial Data Cont. Inversion Time One of the the greatest strengths of the SVD inversion method is the small number of computations it requires as compared to other inversion methods. This results in a much faster inversion, which can be done for a whole tokamak discharge during data acquisition. A comparison of the time required for the SVD inversion to that required for the minimum Fisher inversion, as the number of temporal points inverted is varied, can be seen in the table to the right. Results on Artificial Data Artificial Density Profiles The performance of the SVD inversion has been tested using a phantom density profile, which can be used to calculate pseudo-measurements, i.e. what a detector would measure if the phantom were the true density. The reconstruction of the pseudo- measurements can be compared to the phantom. The SVD inversion has been compared to another commonly used statistical inversion method, minimizing the Fisher information, using the phantom density profile. Inversion Accuracy The accuracy of the SVD inversion has been compared to the minimum Fisher inversion using the phantom density profile. A comparison of both inversions as a function of the flux radius at two different times can be seen below. Because the SVD inversion does not impose artificial constraints on the solution, as opposed to the artificial smoothing imposed by minimizing the Fisher information, it preserves features in the density profile that minimizing the Fisher information destroys. A comparison of both inversion methods as a function of time can be seen below as well. The χ 2 has been calculated in ρ space at each time sample for both methods. It is clear in this plot that the SVD inversion produces less cumulative error. SVD InversionMin. Fisher Information Density ρρ Inversion of artificial electron density as a function of ρ. Actual density profile is shown as dashed red lines and inversions as solid blue lines. Inverted electron densities at (a) ρ = 0 and (b) ρ = Error of inversion methods in ρ space is shown in (c) as a function of time. Time evolution of the actual electron density is shown as a dashed red line, SVD inversion is shown in blue, and minimum Fisher inversion is shown in magenta. Density χ2χ2 t / sawtooth period (a) (b) (c) Application to TCV Data n e [m ] t [s] Standard SawteethGiant Sawteeth (a) (b) (c) (d) Standard SawteethGiant Sawteeth t [s] n e [m ] (a) (b) (c) (d) Standard SawteethGiant Sawteeth n e [m ] ρρ TCV Interferometer Data On TCV, different types of central relaxation oscillations are observed in the presence of electron cyclotron heating (ECH) depending on the location of the deposition power. Normal sawteeth are observed with central ECH power deposition, while giant sawteeth occur when heating close to the sawtooth inversion surface. Examples of line integrated interferometer measurements can be seen for both of these oscillations to the right. The seventh interferometer chord, at the plasma center, is shown in (a) and (c). The first interferometer chord, at the plasma edge, is shown in (b) and (d). Inversion of TCV Data The inversion of the interferometer data shown above, using the SVD method, can be seen below. The SVD inversion reveals the sawtooth oscillation and clearly shows a difference between the standard sawteeth and the giant sawteeth. Furthermore, when plotted as a function of ρ we see that in the case of the giant sawtooth oscillation, the SVD inversion reveals the hollowness of the electron density profile after the sawtooth crash. Inversion using the SVD of TCV interferometer data as a function of time for different radial positions: (a) and (c) on axis, (b) and (d) outside of the inversion radius at ρ = 0.8. Inversion using the SVD of TCV interferometer data as a function of ρ for the times marked in the figure to the left.