Gwanggeun Seo

Skip to content

About

I work on three-dimensional magnetic fields as an actuator in tokamaks: how the plasma responds to small non-axisymmetric fields, and how to choose those fields to control instabilities.

The same actuator cuts both ways: an uncorrected intrinsic error field drives locked modes and disruptions, while a well-chosen resonant magnetic perturbation suppresses ELMs. My work so far has been about measuring and predicting the plasma response to these fields, and using it to optimize the coil currents. I am an M.S. student in the 3DST Lab at Seoul National University, advised by Prof. Jong-Kyu Park. Getting there has meant doing whatever the problem needed at the time: plasma-response modeling with GPEC and DCON, equilibrium and coil design with EFIT and FOCUS, building and installing 3D coils on VEST, and leading 3D-field experiment sessions on KSTAR. Recent work includes a re-assessment of KSTAR's intrinsic error field using profile diagnostics, an application of our group's quasi-symmetric (QSMP) error field correction approach to ITER in collaboration with the ITER Organization, and a physics-guided surrogate model of the plasma response toward within-shot RMP optimization.

I came to fusion because I want it to become a practical energy source. I expect to complete my M.S. in February 2027 and am applying to Ph.D. programs for Fall 2027. What I hope to carry forward is less the specific problem than the way of working — closing the loop between model, experiment, and control — into instability and disruption control more broadly, combining physics models with machine learning to make those decisions in real time.

Research interests

Control of instabilities and disruptions with 3D magnetic fields: error field correction and RMP optimization
Data-driven and physics-guided models for real-time plasma control
3D MHD equilibrium, stability, and plasma response to non-axisymmetric fields
Optimization of 3D coil systems under engineering constraints, and stellarator–tokamak hybrid configurations

News

Nov 2026UpcomingContributed oral at the 68th Annual Meeting of the APS Division of Plasma Physics, Chicago: revised intrinsic error field assessment in KSTAR.
Oct 2026Poster Prize at the 10th Asia-Pacific Conference on Plasma Physics (AAPPS-DPP 2026), Busan.
Oct 2026First-author paper on KSTAR error field assessment published in Nuclear Fusion.
Sep 2026Preprint on quasi-symmetric error field correction for ITER posted on arXiv.
Jul 2026Poster at the 52nd EPS Conference on Plasma Physics, Edinburgh.
May 2026Co-authored preprint on a programmable stellarator–tokamak hybrid (with USTC) posted on arXiv.
Aug 2025Two months at PPPL and Columbia University as part of the SNU–PPPL summer collaboration.
Mar 2025Started the M.S. program in the 3DST Lab at SNU.

Outside the lab

When I'm not in the lab, I'm usually fishing. I also volunteer as an education tutor at the Samsung Sori-Saem Welfare Center in Seoul.

Research

Most of my work sits between modeling and experiment. The common thread is the plasma response to 3D fields: measuring it, predicting it, and using it to choose coil currents.

Intrinsic error field in KSTAR

KSTAR · 2024 – present · with KFE, PPPL, and Columbia University

KSTAR's intrinsic error field reference dates from the 2015 compass scans, and the device has undergone substantial hardware changes since then. This work re-assesses the intrinsic error field under present operating conditions. It grew out of a KSTAR experimental proposal I submitted; the dedicated error field sessions were run together with the KSTAR team at KFE, with PPPL and Columbia collaborating on the analysis, and I served as session leader for the 3D field sessions.

Found an intrinsic error field about an order of magnitude above the 2015 reference value, though still modest in absolute terms for a device of KSTAR's size.
Introduced MID–OFFMID compass scans with rotating-RMP cross-validation.
Used ECE temperature fluctuations and CES rotation data to separate mode locking from field penetration.

Three compass scan plots (MID, OFFMID, H-mode) with fitted circles whose centers mark the same error field direction — Compass scans in three independent coil configurations converge on one intrinsic error field direction.

Paper: Nuclear Fusion 66, 126043 (2026) · APS DPP 2026 oral · AAPPS-DPP 2026 oral and poster (Poster Prize) · Related: Bursch et al., arXiv:2609.30452 (co-author)

Quasi-symmetric error field correction for ITER

ITER · 2024 – present · with the ITER Organization

Conventional error field correction typically targets only the resonant components. Building on the quasi-symmetric magnetic perturbation (QSMP) concept developed in our group, I applied it to ITER as a torque-aware correction framework that also accounts for non-resonant components, and compared it against overlap- and resonance-based schemes.

A correction optimized at 7.5 MA transfers to the 15 MA baseline scenario while staying below the modeled locking-risk threshold.
Evaluated robustness against PF, TF, and CS coil-misalignment error fields.

Bar chart of residual NTV torque for each PF error field source under three correction strategies — Residual NTV torque by correction strategy: resonance-based schemes leave very different torques; the torque-optimal (QSMP) solution controls it.

Preprint: arXiv:2609.29849 · EPS 2026 poster · APS DPP 2025 and the 29th Workshop on MHD Stability Control (talks by J.-K. Park)

Physics-guided surrogate model toward real-time RMP optimization

KSTAR · 2026 – present

A full EFIT–DCON–GPEC calculation is suited to between-shot analysis, not to use within a discharge. I built a surrogate of the GPEC coil-coupling response operator so that RMP coil currents can be recommended in tens of milliseconds, as a step toward within-shot optimization.

Automated, restartable EFIT–DCON–GPEC workflow on HPC; 17,734 validated KSTAR equilibria with shot-separated splits.
Rational-surface-aware equilibrium features cut the median held-out error from 39.2% to 10.7%.
SVD-based n-RMP and constrained E-RMP optimizers under the 5 kA in-vessel control coil (IVCC) current limit, about 40 ms per equilibrium.

3D coil design and installation on VEST

VEST · 2023 – 2025 · undergraduate thesis project

Designed, screened, and optimized a 3D coil system for the VEST spherical tokamak with GPEC, EFIT, CHEASE, and FOCUS, then led the power, wiring, and switching integration.

Built an in-house switching system in place of a commercial unit quoted at about KRW 10 million (about USD 7,000), with a faster rise time.

Related: Plasma Physics and Controlled Fusion 67, 115021 (2025) · iFPC 2024 oral · iFPC 2025 poster

Error field correction and RMP coil optimization for CPD

CPD · 2025 – present

CPD (Compact Pilot Device) is a next-step fusion device being designed in Korea. I am responsible for the error field correction (EFC) side of the design: quantifying the error field risk from coil misalignment and optimizing the correction-coil layout, using GPEC plasma-response calculations.

Quantified PF/CS coil-misalignment error field risk with 24 single-coil GPEC responses and a 30,000-sample Monte Carlo scan: 3.77% of cases exceed the 8.2 G engineering reference.
Compared three-row EFC coil geometries over poloidal span and vessel offset; correcting all four rational surfaces to 0.82 G needs 117.1 kA versus 18.9 kA for a core-only target, a limit set by the actuator rank of a three-row design.

Failure probability versus equivalent coil misalignment per physical coil on a log axis: without correction the 50% point is at about 1 mm; with three error field correction targets it moves to about 9, 86, and 147 mm — Monte Carlo locking-risk curves for CPD: probability of exceeding the q = 2 locking criterion versus equivalent misalignment per coil, without correction and with three EFC targets. Correction moves the 50% point from about 1 mm of misalignment to roughly 10–150 mm depending on the target.

Programmable stellarator–tokamak hybrid

USTC · summers of 2024 and 2025 · Prof. Caoxiang Zhu's group

Joint work on a hybrid device that uses saddle and dipole coils for large-scale magnetic-configuration discovery. I contributed the tokamak-relevant 3D perturbation analysis and introduced GPEC workflows to the group: localized resonant-field control at single rational surfaces, and a quasi-symmetric perturbation that reduces tokamak NTV torque by about five orders of magnitude.

Scaled resonant field versus normalized flux with the resonant field isolated at a single rational surface — Fig. 2(D): localized resonant-field control at single rational surfaces.

Fig. 2(E): QSMP from the same coils reduces tokamak NTV torque by about five orders of magnitude.

Preprint: arXiv:2605.03599 (third author), submitted to PNAS

Publications

Journal articles and preprints

[P1]
G.-G. Seo, S.-J. Han, J. Kim, Y. Lee, S. M. Yang, J. W. Juhn, J.-K. Park. Nuclear Fusion 66, 126043 (2026). Open access.
DOI PDF
Nearly a decade after the 2015 KSTAR error field study, and after substantial hardware changes to the device, we re-assessed the intrinsic n = 1 error field with compass scans in three independent coil configurations. Using ECE and CES profiles near the q = 2 surface, we separate local field penetration from the later global locking, which gives a cleaner penetration threshold. All three scans point to the same error field direction, about an order of magnitude above the 2015 reference value but still modest in absolute terms, and a rotating-RMP discharge confirms it independently.

Three scans, one answer. Compass scans in the MID, OFFMID, and H-mode top/middle/bottom (T/M/B) coil configurations all converge on a single intrinsic error field direction.

Penetration is not locking. The ECE drop near q = 2 comes first; core sawtooth suppression and rotation collapse follow 30–40 ms later.

Independent check. A slowly rotated n = 1 field penetrates where the compass scan says it should.
BibTeX
```
@article{Seo2026NF,
  author  = {Seo, G.-G. and Han, S.-J. and Kim, J. and Lee, Y. and Yang, S. M. and Juhn, J. W. and Park, J.-K.},
  title   = {Revised error field assessment in {KSTAR} with profile diagnostics of resonant field penetration},
  journal = {Nuclear Fusion},
  volume  = {66},
  number  = {12},
  pages   = {126043},
  year    = {2026},
  doi     = {10.1088/1741-4326/aeac69}
}
```
[P2]

G.-G. Seo, J.-K. Park, S. J. Han, M. Dubrov. arXiv preprint (2026); to be submitted to Nuclear Fusion.
arXiv:2609.29849 PDF
[P3]

G. Yu, X. Nie, G.-G. Seo, D. Huang, H. Liu, J. Liu, J. Cho, H.-S. Kim, J. Xie, G. Zhuang, F. Ding, J.-K. Park, C. Zhu. arXiv preprint (2026); submitted to PNAS.
arXiv:2605.03599 PDF
[P4]
Developing an IMAS-compatible platform for the university-scale tokamak VEST and its application to operating characteristics analysisH.-S. Yun, S. Lee, L. Jung, J.-M. Lee, G. Nam, G.-G. Seo, J. Lee, Y. S. Hwang, J.-K. Park, and the VEST Team. Plasma Physics and Controlled Fusion 67, 115021 (2025).
DOI
[P5]
Demonstration of H-mode error field identification in a single discharge via island healingE. M. Bursch, A. Xie, J. L. Barr, S. K. Kim, N. C. Logan, S. M. Yang, J. G. Bak, W. Choi, Q. Hu, J. W. Juhn, J. Kim, J. Lee, N. Leuthold, J. K. Park, C. Paz-Soldan, L. Piron, G.-G. Seo, J. Seo. arXiv preprint (2026); submitted to Nuclear Fusion.
arXiv:2609.30452

Conference presentations

[C1]
Revised intrinsic error field assessment in KSTAR from local penetration diagnosticsG.-G. Seo, J. Kim, Y. Lee, S.-J. Han, J. W. Juhn, S. M. Yang, H. Han, Y. M. Jeon, G. Y. Park, H.-S. Kim, J.-K. Park. 68th Annual Meeting of the APS Division of Plasma Physics, Chicago, Nov. 2026.
Oral
[C2]
Intrinsic error field diagnosis in KSTAR via cross-geometry compass scans and rotating-RMP validationG.-G. Seo, J. Kim, Y. Lee, S. J. Han, J. W. Juhn, J.-K. Park. 10th Asia-Pacific Conference on Plasma Physics (AAPPS-DPP 2026), Busan, Oct. 2026.
OralPosterPoster Prize
[C3]
Torque-aware error field correction for ITER using quasi-symmetric magnetic perturbationsG.-G. Seo, J.-K. Park, M. Dubrov, S. J. Han. 52nd EPS Conference on Plasma Physics, Edinburgh, Jul. 2026.
PosterAbstract
[C4]
Quasisymmetric correction to non-resonant error fields in ITERJ.-K. Park, G.-G. Seo, S. Han, S. M. Yang, N. C. Logan. 67th Annual Meeting of the APS Division of Plasma Physics, Long Beach, Nov. 2025.
OralAbstract
[C5]
Quasi-symmetric optimization of 3D magnetic perturbationsJ.-K. Park, G.-G. Seo, et al. 29th Workshop on MHD Stability Control, Princeton, Jul. 2025.
Oral
[C6]
Validation of QSMP–RMP transition and prospective QSMP studies on VESTG.-G. Seo, S. Han, G. Nam, Y. Jung, H.-S. Yun, J.-K. Park. 4th International Fusion and Plasma Conference (iFPC 2025), Daejeon, Jun. 2025.
Poster
[C7]
Initial 3D coil design and optimization in VESTG.-G. Seo, H.-S. Yun, Y. Jung, N. C. Logan, W. K. Hong, J.-K. Park. 3rd International Fusion and Plasma Conference (iFPC 2024), Seoul, Jun. 2024.
Oral

Other co-authored work

Manuscript under review

[O1]
Long-pulse sustainment of edge localized mode suppression via integrated resonant magnetic perturbation control in KSTARJ.-K. Park, G. Shin, S. Kim, R. Shousha, S. M. Yang, N. C. Logan, Q. Hu, M. Kim, S. Han, S. Jo, G.-G. Seo, et al. Submitted to Nuclear Fusion (2026).

Conference abstracts

[O2]
Robustness and effects of quasi-symmetric magnetic perturbation in KSTARS. Han, G.-G. Seo, et al. 10th Asia-Pacific Conference on Plasma Physics (AAPPS-DPP 2026), Busan, Oct. 2026.
[O3]
Quasi-symmetric magnetic perturbation for robust tokamak operationS. Han, G.-G. Seo, J. Kang, S. Jo, J.-K. Park. 67th Annual Meeting of the APS Division of Plasma Physics, Long Beach, Nov. 2025.
PosterAbstract

Curriculum Vitae

Open CV as PDF Updated October 2026

Education

Seoul National UniversityMar. 2025 – Feb. 2027 (expected) M.S., Nuclear Engineering (Fusion Plasma) · Advisor: Prof. Jong-Kyu Park

Supported by the BK21 FOUR graduate fellowship, Ministry of Education (2025 – present)
M.S. thesis in preparation

Seoul National UniversityMar. 2019 – Feb. 2025 B.S., Nuclear Engineering

Undergraduate thesis: “Initial 3D coil design and optimization in VEST”
Mandatory military service, Republic of Korea Army, Mar. 2021 – Sep. 2022

Honors and awards

Poster Prize, 10th Asia-Pacific Conference on Plasma Physics (AAPPS-DPP 2026)Oct. 2026

Best Paper Award for Undergraduate Research, Seoul National UniversityFeb. 2025

Research experience

Researcher, 3DST Lab, Seoul National UniversityJun. 2023 – present Undergraduate researcher Jun. 2023 – Feb. 2025; graduate researcher Mar. 2025 – present

KSTAR error field experiment (2024 – present): experimental proposal and session leadership, with KFE, PPPL, and Columbia; first-author Nuclear Fusion paper
ITER quasi-symmetric error field correction (2024 – 2026), with the ITER Organization
Physics-guided surrogate modeling toward real-time KSTAR RMP optimization (2026 – present)
CPD error field correction and RMP coil optimization (2025 – present)
VEST 3D coil optimization and installation (2023 – 2025)

Research Student, SNU–PPPL Summer CollaborationJul. – Aug. 2025 Princeton Plasma Physics Laboratory and Columbia University

GPEC Hackathon at Columbia (3D MHD stability and perturbed-equilibrium implementation); consulted with the HBT-EP group on non-axisymmetric coil design

Collaboration with Prof. Caoxiang Zhu's group, USTCJul. – Aug. 2024; Jul. 2025

Programmable stellarator–tokamak hybrid; third author on the resulting preprint

Other collaborations

Columbia University (Dr. Nikolas Logan, Evan Bursch): ongoing collaboration on KSTAR error field physics
PPPL: GPEC deployment on the institutional cluster · Aditya-U: GPEC advisory support

Skills

Plasma modeling: GPEC, EFIT, KineticEFIT, DCON, CHEASE, FOCUS, OMFIT
Experiment and hardware: Coil and power-system design, switching-circuit design and installation, ECE/CES diagnostic analysis
Programming and ML: Python, Fortran, C/C++, MATLAB, Julia, PyTorch, scikit-learn, NumPy/SciPy
Computing: Linux, HPC cluster workflows, parallel CPU computing, MDS+, Git, LaTeX
Languages: Korean (native), English (fluent), Korean Sign Language (intermediate), Chinese (basic)

Leadership, service, and activities

KFAS Study Abroad Program, Korea Foundation for Advanced Studies2026 – present

Deputy Laboratory Leader, 3DST Lab (Laboratory Leader in 2024)Jan. 2026 – present

Education Volunteer, Samsung Sori-Saem Welfare Center2025 – present

Member, Hanam Dreamers (Hanam City university-student leadership group)2020

CBRN Smoke Equipment Instructor, Republic of Korea ArmyMar. 2021 – Sep. 2022 Technical instruction to groups of 30–100 personnel, from enlisted soldiers to officers and civilian firefighters