Research interest
Developing and applying first-principle and numerical methods, along with machine learning models, to solve complex scientific problems using computer-based simulations.
Machine learning (ML) and Optimization
- Developing methodology and applying interatomic machine learning potentials for predicting atomic interactions, structures, dynamics, and reactions.
- On-the-fly ML+MC for accelaing MC sampling
- Supervised and unsupervised learning for physics problems
Condensed matter physics
- First-principles calculations; DFT, DFT+U, DMFT, GW
- linear response theory MFT; (magnetic force theory), Jx the open source MFT software, https://kaist-elst.github.io/Jx.jl/
Machine learning-based numerical simulations, integration of machine learning into scientific modeling, empirical modeling for large-scale simulations, development of methodology based on first-principles physics, numerical optimization methods for understanding experimental phenomena, and software development for experimental orchestration are among the research scopes in this field. These areas can potentially contribute to solving complex scientific problems in materials science, chemistry, and physics.
Experiences
Combining machine learning and physics to accelerate atomic simulations. Developing first-principles methods for studying both physical properties and chemical reactions.
First-principles method developments and develop machine learning algorithms for physics.
Method development for accelerating e-Beam Monte-Carlo simulation.
Hubbard U calculation through machine learning.
Compensate e-Beam scattering proximity effect by inverse function design with Monte Carlo.
Presentation & Awards
Outstanding presentation awards
Duk Joo Kim Young Scientist Award,
- Korean Physical Society, (2020).
Jx; An open-source software to calculate magnetic coupling constant and matrix,
- KIAS, Poster (2019)
Analytic continuation via “domain knowledge free” machine learning,
- APCTP-KIAS:Quantum Materials Symposium 2018, Poster (2018).
Analytic continuation via “domain knowledge free” machine learning,
- KPS2018 Spring, Contributed talk (2018)
9 th Summer school on ‘Scientific Computing and Machine Learning’,
- KIAS-CAC (2018)
Reliability and applicability of magnetic force linear response theory; Numerical parameters, predictability, and orbital resolution,
- 10th BK21 Young Physicists Workshop, Poster (2018)
Reliability and applicability of magnetic force linear response theory; Numerical parameters, predictability, and orbital resolution,
- KPS2017 Fall, Contributed talk (2017)
Orbital resolved exchange interactions combined with QSGW self energy.
- KIAS, Poster (2017)
Invited talks
On-the-fly accelerating Monte Carlo simulations with XAI
- Seoul National University of Science and Technology, Seoul (2021.08)
Reliably accelerating Monte Carlo simulations with XAI
- Machine Learning on Condensed Matter physics APCTP, (2021.08)
Analysis and debugging of material informatics using tools
- 2021 Summer Material Informatics Convergence Education Program, (2021.08)
Reliably accelerating Monte Carlo simulations with XAI
- PCS-IBS, Deajeon (2021.08)
Explainable AI (XAI) for scientific computing.
- Korea Institute for Advanced Study (KIAS), Seoul (2021.06.16)
Computational method developments for material science (Machine learning, first-principles method development & more).
- Korea Institute of Science and Technology (KIST), Seoul (2020.11.25)
Method development for numerical condensed matter physics; Jx, the magnetic exchange interactions and accelerated analytic continuation via machine learning,
- Kyungpook National University., Daegu (2020.01.30)
Machine learning for numerical condensed matter physics; analytic continuation and Monte Carlo sampling,
- Kangwon National Univ., Chuncheon (2020.01.21)
Machine learning for numerical condensed matter physics; analytic continuation and Monte Carlo sampling,
- POSTECH, Pohang (2019.12)
First principles based methodologies for correlated materials magnetic force theory and analytic continuation,
- 5th Workshop on Supercomputing for Computational Bio/Nano/Materials Science, Daejeon (2019.7)
Analytic continuation via “domain-knowledge free” machine learning,
- PCS-IBS, Daejeon (2019.3)
Analytic continuation via domain-knowledge free machine learning,
- KPS 2018Fall. Focus sessions AI (artificial intelligence) techniques for condensed matter physics and material science II (2018.10)
Analytic continuation via “domain-knowledge free” machine learning,
- KIAS-Comp.Sciences, Seoul (2018.6)
Reliability and applicability of magnetic force linear response theory; Numerical parameters, predictability, and orbital resolution with OpenMX,
- ISSP-The University of Tokyo (2018.7.10)
Recent presentations
On-the-fly machine learning algorithm for accelerating Monte Carlo sampling; Application to the stochastic analytical continuation,
- APS march meeting, Boston (2020.03), Contributed talk
On-the-fly machine-learning algorithm for accelerating Monte Carlo sampling; Application to the stochastic analytical continuation,
- KPS, (2019.10), Contributed talk
Jx; An open-source software to calculate magnetic coupling constant and matrix
- Asian 22, Osaka (2019.10), Poster
Analytic continuation via domain-knowledge free machine learning,
- APS march meeting, Boston (2019.03), Contributed talk
Analytic continuation via domain-knowledge free machine learning,
- Kavli Institute for Theoretical Physics, (2019.02), Poster
First-principles calculation of Heisenberg exchange coupling and branching ratio, International Workshop on Superconductivity and Related Functional Materials
- (NIMS), Japan (2016.12), Poster
Analytic continuation via domain-knowledge free machine learning
- Materials Research Society, Boston (2018.11), Poster
Publications
Google scholar linkElectrical and Electronic Materials (submitted), 2020.
Comput. Phys. Commun., 247:106927, February 2020.
Phys. Rev. B, 101:064513, February 2020, *these authors are equally contributed.
Journal of Physics Condensed Matter, 31(40):405503, October 2019.
Phys. Rev. B, 97(12):125132, March 2018.
Phys. Rev. B, 98(24):245101, December 2018.
Scientific Reports, 8(1):7016, May 2018
arXiv:1904.04510 [cond-mat], April 2019 (under review).
Phys. Rev. B, 97(21):214431, June 2018.
Physical Review B, 94(11), September 2016.
JPCM accepted, February 2020.
Phys. Rev. Materials, 3(3):031001, March 2019.
Phys. Rev. B, 98(12):125126, September 2018.
EPL, 125(4):47005, February 2019.
Materials Today Physics, 8:18–24, March 2019.
J. Phys. Soc. Jpn., 84(3):034602, February 2015.
Solar Energy Materials and Solar Cells, 132:56–66, January 2015.
In Preparation
On-the-fly machine learning algorithm for accelerating Monte Carlo sampling; Application to the stochastic analytical continuation.
in Preparation.