Dongwook Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Dongwook Lee has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Adam Reyes
Pacific Fusion Corporation 2 , 6082 Stewart Ave., Fremont, California 94538
6 shared publications - Carlo Graziani
Mathematics and Computer Science Argonne National Laboratory Argonne IL United States
4 shared publications - Petros Tzeferacos
University of Rochester 1 , Rochester, New York 14627
4 shared publications - Youngjun Lee
Argonne National Laboratory
4 shared publications - Rémi Bourgeois
Université Paris-Saclay UVSQ CNRS CEA Maison de la Simulation 91191 Gif-sur-Yvette France
3 shared publications - Andrea Antoni
Flatiron Institute
2 shared publications
Latest Publications
- Open Access
- 9 September 2026
- Published by Elsevier in SSRN Electronic Journal
A hyperbolic-variant successive over-relaxation Poisson solver for divergence cleaning in a cell-centered fifth-order finite volume MHD scheme
We present a fifth-order accurate multidimensional finite volume MHD scheme based on a divergence-cleaning projection method that numerically preserves the solenoidal constraint on the magnetic field. Our projection method utilizes a new hyperbolic-variant successive over-relaxation (SORh) iteration designed specifically to solve the Poisson equation arisingin the projection step. Unlike the conventional FFT-based global approach, SORh is a local method whose overall performance is nonetheless comparable to a highly optimized FFT projection. We integrate the new SORh scheme into an MHD scheme within the projectionframework, introducing two complementary discretizations in a finite-volume form: (i) a second-order baseline algorithm and (ii) a fifth-order accurate algorithm based on the Gaussian-process (GP) approximation. In both, SORh serves as the divergence-cleaningiterative Poisson solver; the first approach uses standard second-order finite differences for the discrete Laplacian, while the second uses a high-order GP-based discretization. In comparison to global FFT-based Poisson solvers, SORh features a matrix-free, single-stencil-pass-per-sweep structure for its local operations, admitting straightforward extension to mixed and non-periodic boundary conditions, variable-coefficient elliptic operators, andadaptive meshes. We also present a von Neumann analysis to characterize SORh's pseudo-time stability limits. The new SORh MHD algorithm is validated across a suite of benchmark test problems, including the smooth circularly polarized Alfvén wave for convergence and several shock-dominated problems, on which the new method is on par with the highly optimized spectral FFT projection in terms of accuracy, stability, and efficiency.
- 9 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessA hyperbolic-variant successive over-relaxation Poisson solver for divergence cleaning in a cell-centered fifth-order finite volume MHD scheme
We present a fifth-order accurate multidimensional finite volume MHD scheme based on a divergence-cleaning projection method that numerically preserves the solenoidal constraint on the magnetic field. Our projection method utilizes a new hyperbolic-variant successive over-relaxation (SORh) iteration designed specifically to solve the Poisson equation arisingin the projection step. Unlike the conventional FFT-based global approach, SORh is a local method whose overall performance is nonetheless comparable to a highly optimized FFT projection. We integrate the new SORh scheme into an MHD scheme within the projectionframework, introducing two complementary discretizations in a finite-volume form: (i) a second-order baseline algorithm and (ii) a fifth-order accurate algorithm based on the Gaussian-process (GP) approximation. In both, SORh serves as the divergence-cleaningiterative Poisson solver; the first approach uses standard second-order finite differences for the discrete Laplacian, while the second uses a high-order GP-based discretization. In comparison to global FFT-based Poisson solvers, SORh features a matrix-free, single-stencil-pass-per-sweep structure for its local operations, admitting straightforward extension to mixed and non-periodic boundary conditions, variable-coefficient elliptic operators, andadaptive meshes. We also present a von Neumann analysis to characterize SORh's pseudo-time stability limits. The new SORh MHD algorithm is validated across a suite of benchmark test problems, including the smooth circularly polarized Alfvén wave for convergence and several shock-dominated problems, on which the new method is on par with the highly optimized spectral FFT projection in terms of accuracy, stability, and efficiency.
- Open Access
- 1 June 2026
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 3252 (1)
Stability of Coupling Learned Diffusion to Shock Capturing Schemes
We propose a hybrid ML-numerical solver for mixed hyperbolic–parabolic PDEs in which a neural operator replaces only the parabolic (diffusive) contribution while a standard shock-capturing finite-volume method advances the hyperbolic (advective) fluxes. This selective coupling targets the parabolic timestep bottleneck without discarding the robustness of classical advection discretizations. The central challenge is long-horizon stability since small spectral errors in the learned diffusion update can accumulate under repeated timesteps and drive drift or instability in long rollouts. We investigate this in a controlled setting using the one-dimensional viscous Burgers equation, where a U-shaped neural operator (UNO) predicts the cell-centered diffusive contribution and is coupled to a finite-volume hyperbolic update. We find that (i) continuation error is strongly correlated with spectral error, indicating that high-frequency drift is the dominant failure mode, (ii) multi-step (substep) consistency losses provide effective implicit spectral regularization and substantially improve long-time stability compared with single-step MSE and explicit spectral penalties, and (iii) cross-scheme rollouts (MUSCL ↔ WENO5) demonstrate that the learned diffusion surrogate adapts to the training-time hyperbolic discretization, producing rapid instability when numerical diffusion in the hyperbolic flux is reduced and stable but over-diffusive behavior when it is increased. Finally, we show that an eigenvalue-based Jacobian diagnostic can detect unstable regimes.
- 1 June 2026
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 3252 (1)
Open AccessStability of Coupling Learned Diffusion to Shock Capturing Schemes
We propose a hybrid ML-numerical solver for mixed hyperbolic–parabolic PDEs in which a neural operator replaces only the parabolic (diffusive) contribution while a standard shock-capturing finite-volume method advances the hyperbolic (advective) fluxes. This selective coupling targets the parabolic timestep bottleneck without discarding the robustness of classical advection discretizations. The central challenge is long-horizon stability since small spectral errors in the learned diffusion update can accumulate under repeated timesteps and drive drift or instability in long rollouts. We investigate this in a controlled setting using the one-dimensional viscous Burgers equation, where a U-shaped neural operator (UNO) predicts the cell-centered diffusive contribution and is coupled to a finite-volume hyperbolic update. We find that (i) continuation error is strongly correlated with spectral error, indicating that high-frequency drift is the dominant failure mode, (ii) multi-step (substep) consistency losses provide effective implicit spectral regularization and substantially improve long-time stability compared with single-step MSE and explicit spectral penalties, and (iii) cross-scheme rollouts (MUSCL ↔ WENO5) demonstrate that the learned diffusion surrogate adapts to the training-time hyperbolic discretization, producing rapid instability when numerical diffusion in the hyperbolic flux is reduced and stable but over-diffusive behavior when it is increased. Finally, we show that an eigenvalue-based Jacobian diagnostic can detect unstable regimes.
- Open Access
- 22 December 2025
- Published by American Physical Society (APS) in Physical Review Accelerators and Beams
- Vol. 28 (12)
Demonstration of a code coupling framework for modeling beam-collimator impacts in the advanced photon source
The high-brightness beams being produced in current and future accelerators present new machine protection concerns with the potential for high-energy-density (HED) conditions ( > 100 J / mm 3 ) in beam-intercepting components. Simulating HED conditions in accelerators requires utilizing a suite of physics codes for particle dynamics, particle-matter interactions, and hydrodynamics. This paper describes a method of coupling the codes , , and to simulate the effects of a rapid beam loss in the advanced photon source storage ring and the resulting interaction of the beam and collimators. This paper expands previous work [J. Dooling , Collimator irradiation studies at the advanced photon source, in Proceedings of the IBIC-2023 (2023), pp. 245–249] by introducing a definition of the evolving geometry of the collimator surface as well as providing methods for simulating the absorption of synchrotron radiation and tracking shower particles produced during beam strikes. We demonstrate this framework by simulating machine conditions of the APS ring before and after its recent upgrade. Simulation results are compared with observed damage to collimators and test samples taken from the APS ring.
- 22 December 2025
- Published by American Physical Society (APS) in Physical Review Accelerators and Beams
- Vol. 28 (12)
Open AccessDemonstration of a code coupling framework for modeling beam-collimator impacts in the advanced photon source
The high-brightness beams being produced in current and future accelerators present new machine protection concerns with the potential for high-energy-density (HED) conditions ( > 100 J / mm 3 ) in beam-intercepting components. Simulating HED conditions in accelerators requires utilizing a suite of physics codes for particle dynamics, particle-matter interactions, and hydrodynamics. This paper describes a method of coupling the codes , , and to simulate the effects of a rapid beam loss in the advanced photon source storage ring and the resulting interaction of the beam and collimators. This paper expands previous work [J. Dooling , Collimator irradiation studies at the advanced photon source, in Proceedings of the IBIC-2023 (2023), pp. 245–249] by introducing a definition of the evolving geometry of the collimator surface as well as providing methods for simulating the absorption of synchrotron radiation and tracking shower particles produced during beam strikes. We demonstrate this framework by simulating machine conditions of the APS ring before and after its recent upgrade. Simulation results are compared with observed damage to collimators and test samples taken from the APS ring.
- Open Access
GP-Recipe: Gaussian Process approximation to linear operations in numerical methods
We introduce new Gaussian Process (GP) high-order approximations to linear operations that are frequently used in various numerical methods. Our method employs the kernel-based GP regression modeling, a non-parametric Bayesian approach to regression that operates on the probability distribution over all admissible functions that fit observed data. We begin in the first part with discrete data approximations to various linear operators applied to smooth data using the most popular squared exponential kernel function. In the second part, we discuss data interpolation across discontinuities with sharp gradients, for which we introduce a new GP kernel that fits discontinuous data without oscillations. The current study extends our previous GP work on polynomial-free shock-capturing methods in finite difference and finite volume methods to a suite of linear operator approximations on smooth data. The formulations introduced in this paper can be readily adopted in daily practices in numerical methods, including numerical approximations of finite differences, quadrature rules, interpolations, and reconstructions, which are most frequently used in numerical modeling in modern science and engineering applications. In the test problems, we demonstrate that the GP approximated solutions feature improved solution accuracy compared to the conventional finite-difference counterparts.
Open AccessGP-Recipe: Gaussian Process approximation to linear operations in numerical methods
We introduce new Gaussian Process (GP) high-order approximations to linear operations that are frequently used in various numerical methods. Our method employs the kernel-based GP regression modeling, a non-parametric Bayesian approach to regression that operates on the probability distribution over all admissible functions that fit observed data. We begin in the first part with discrete data approximations to various linear operators applied to smooth data using the most popular squared exponential kernel function. In the second part, we discuss data interpolation across discontinuities with sharp gradients, for which we introduce a new GP kernel that fits discontinuous data without oscillations. The current study extends our previous GP work on polynomial-free shock-capturing methods in finite difference and finite volume methods to a suite of linear operator approximations on smooth data. The formulations introduced in this paper can be readily adopted in daily practices in numerical methods, including numerical approximations of finite differences, quadrature rules, interpolations, and reconstructions, which are most frequently used in numerical modeling in modern science and engineering applications. In the test problems, we demonstrate that the GP approximated solutions feature improved solution accuracy compared to the conventional finite-difference counterparts.
- Open Access
- 1 May 2024
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 967 (1) , 6
KFVM-WENO: A High-order Accurate Kernel-based Finite Volume Method for Compressible Hydrodynamics
This paper presents a fully multidimensional kernel-based reconstruction scheme for finite volume methods applied to systems of hyperbolic conservation laws, with a particular emphasis on the compressible Euler equations. Nonoscillatory reconstruction is achieved through an adaptive-order weighted essentially nonoscillatory (WENO) method cast into a form suited to multidimensional reconstruction. A kernel-based approach inspired by radial basis functions and Gaussian process modeling, which we call kernel-based finite volume method with WENO, is presented here. This approach allows the creation of a scheme of arbitrary order of accuracy with simply defined multidimensional stencils and substencils. Furthermore, the fully multidimensional nature of the reconstruction allows for a more straightforward extension to higher spatial dimensions and removes the need for complicated boundary conditions on intermediate quantities in modified dimension-by-dimension methods. In addition, a new simple yet effective set of reconstruction variables is introduced, which could be useful in existing schemes with little modification. The proposed scheme is applied to a suite of stringent and informative benchmark problems to demonstrate its efficacy and utility. A highly parallel multi-GPU implementation using Kokkos and the message-passing interface is also provided.
- 1 May 2024
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 967 (1) , 6
Open AccessKFVM-WENO: A High-order Accurate Kernel-based Finite Volume Method for Compressible Hydrodynamics
This paper presents a fully multidimensional kernel-based reconstruction scheme for finite volume methods applied to systems of hyperbolic conservation laws, with a particular emphasis on the compressible Euler equations. Nonoscillatory reconstruction is achieved through an adaptive-order weighted essentially nonoscillatory (WENO) method cast into a form suited to multidimensional reconstruction. A kernel-based approach inspired by radial basis functions and Gaussian process modeling, which we call kernel-based finite volume method with WENO, is presented here. This approach allows the creation of a scheme of arbitrary order of accuracy with simply defined multidimensional stencils and substencils. Furthermore, the fully multidimensional nature of the reconstruction allows for a more straightforward extension to higher spatial dimensions and removes the need for complicated boundary conditions on intermediate quantities in modified dimension-by-dimension methods. In addition, a new simple yet effective set of reconstruction variables is introduced, which could be useful in existing schemes with little modification. The proposed scheme is applied to a suite of stringent and informative benchmark problems to demonstrate its efficacy and utility. A highly parallel multi-GPU implementation using Kokkos and the message-passing interface is also provided.
- Open Access
- 1 September 2023
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 954 (2) , 176
Hydrodynamics and Survivability during Post-main-sequence Planetary Engulfment
The engulfment of substellar bodies (SBs), such as brown dwarfs and planets, by giant stars is a possible explanation for rapidly rotating giants, lithium-rich giants, and the presence of SBs in close orbits around subdwarfs and white dwarfs. We perform three-dimensional hydrodynamical simulations of the flow in the vicinity of an engulfed SB. We model the SB as a rigid body with a reflective surface because it cannot accrete. This reflective boundary changes the flow morphology to resemble that of engulfed compact objects with outflows. We measure the drag coefficients for the ram-pressure and gravitational drag forces acting on the SB, and use them to integrate its trajectory inside the star. We find that engulfment can increase the luminosity of a 1 M ⊙ star by up to a few orders of magnitude. The time for the star to return to its original luminosity is up to a few thousand years when the star has evolved to ≈10 R ⊙ and up to a few decades at the tip of the red giant branch (RGB). No SBs can eject the envelope of a 1 M ⊙ star before it evolves to ≈10 R ⊙ if the orbit of the SB is the only energy source contributing to the ejection. In contrast, SBs as small as ≈10 M Jup can eject the envelope at the tip of the RGB. The numerical framework we introduce here can be used to study planetary engulfment in a simplified setting that captures the physics of the flow at the scale of the SB.
- 1 September 2023
- Published by American Astronomical Society in The Astrophysical Journal
- Vol. 954 (2) , 176
Open AccessHydrodynamics and Survivability during Post-main-sequence Planetary Engulfment
The engulfment of substellar bodies (SBs), such as brown dwarfs and planets, by giant stars is a possible explanation for rapidly rotating giants, lithium-rich giants, and the presence of SBs in close orbits around subdwarfs and white dwarfs. We perform three-dimensional hydrodynamical simulations of the flow in the vicinity of an engulfed SB. We model the SB as a rigid body with a reflective surface because it cannot accrete. This reflective boundary changes the flow morphology to resemble that of engulfed compact objects with outflows. We measure the drag coefficients for the ram-pressure and gravitational drag forces acting on the SB, and use them to integrate its trajectory inside the star. We find that engulfment can increase the luminosity of a 1 M ⊙ star by up to a few orders of magnitude. The time for the star to return to its original luminosity is up to a few thousand years when the star has evolved to ≈10 R ⊙ and up to a few decades at the tip of the red giant branch (RGB). No SBs can eject the envelope of a 1 M ⊙ star before it evolves to ≈10 R ⊙ if the orbit of the SB is the only energy source contributing to the ejection. In contrast, SBs as small as ≈10 M Jup can eject the envelope at the tip of the RGB. The numerical framework we introduce here can be used to study planetary engulfment in a simplified setting that captures the physics of the flow at the scale of the SB.
- Open Access
Hydrodynamics and survivability during post-main-sequence planetary engulfment
The engulfment of substellar bodies (SBs, such as brown dwarfs and planets) by giant stars is a possible explanation for rapidly rotating giants, lithium-rich giants, and the presence of SBs in close orbits around subdwarfs and white dwarfs. We simulate the flow in the vicinity of an engulfed SB in three-dimensional hydrodynamics. We model the SB as a rigid body with a reflective surface because it cannot accrete. This reflective boundary changes the flow morphology to resemble that of engulfed compact objects with outflows. We measure the drag coefficients for the ram pressure and gravitational drag forces acting on the SB, and use them to integrate its trajectory inside the star. We find that engulfment can increase the luminosity of a $1M_\odot$ star by up to a few orders of magnitude. The time for the star to return to its original luminosity is up to a few thousand years when the star has evolved to $\approx10R_\odot$ and up to a few decades at the tip of the red giant branch. No SBs can eject the envelope of a $1M_\odot$ star before it evolves to $\approx10R_\odot$, if the orbit of the SB is the only energy source contributing to the ejection. In contrast, SBs as small as $\approx10M_\text{Jup}$ can eject the envelope at the tip of the red giant branch. The numerical framework we introduce here can be used to study planetary engulfment in a simplified setting that captures the physics of the flow at the scale of the SB.Open AccessHydrodynamics and survivability during post-main-sequence planetary engulfment
The engulfment of substellar bodies (SBs, such as brown dwarfs and planets) by giant stars is a possible explanation for rapidly rotating giants, lithium-rich giants, and the presence of SBs in close orbits around subdwarfs and white dwarfs. We simulate the flow in the vicinity of an engulfed SB in three-dimensional hydrodynamics. We model the SB as a rigid body with a reflective surface because it cannot accrete. This reflective boundary changes the flow morphology to resemble that of engulfed compact objects with outflows. We measure the drag coefficients for the ram pressure and gravitational drag forces acting on the SB, and use them to integrate its trajectory inside the star. We find that engulfment can increase the luminosity of a $1M_\odot$ star by up to a few orders of magnitude. The time for the star to return to its original luminosity is up to a few thousand years when the star has evolved to $\approx10R_\odot$ and up to a few decades at the tip of the red giant branch. No SBs can eject the envelope of a $1M_\odot$ star before it evolves to $\approx10R_\odot$, if the orbit of the SB is the only energy source contributing to the ejection. In contrast, SBs as small as $\approx10M_\text{Jup}$ can eject the envelope at the tip of the red giant branch. The numerical framework we introduce here can be used to study planetary engulfment in a simplified setting that captures the physics of the flow at the scale of the SB. - 13 September 2022
- Published by Elsevier in Journal of Computational Physics
- Vol. 471
GP-MOOD: A positivity-preserving high-order finite volume method for hyperbolic conservation laws
- 13 September 2022
- Published by Elsevier in Journal of Computational Physics
- Vol. 471
GP-MOOD: A positivity-preserving high-order finite volume method for hyperbolic conservation laws
- 26 February 2022
- Published by Mathematical Sciences Publishers in Communications in Applied Mathematics and Computational Science
- Vol. 17 (1) , 1-41
An application of Gaussian process modeling for high-order accurate adaptive mesh refinement prolongation
- 26 February 2022
- Published by Mathematical Sciences Publishers in Communications in Applied Mathematics and Computational Science
- Vol. 17 (1) , 1-41
An application of Gaussian process modeling for high-order accurate adaptive mesh refinement prolongation
- Open Access
- 1 January 2022
- Published by Elsevier in SSRN Electronic Journal
Gp-Mood: A Positive-Preserving High-Order Finite Volume Method for Hyperbolic Conservation Laws
We present an a posteriori shock-capturing finite volume method algorithm called GP-MOOD that solves a compressible hyperbolic conservative system at high-order solution accuracy (e.g., third-, fifth-, and seventh-order) in multiple spatial dimensions. The GP-MOOD method combines two methodologies, the polynomial-free spatial reconstruction methods of GP (Gaussian Process) and the a posteriori detection algorithms of MOOD (Multidimensional Optimal Order Detection). The spatial approximation of our GP-MOOD method uses GP’s unlimited spatial reconstruction that builds upon our previous studies on GP reported in Reyes et al., Journal of Scientific Computing, 76 (2017) and Journal of Computational Physics, 381 (2019). This paper focuses on ex- tending GP’s flexible variability of spatial accuracy to an a posteriori detection formalism based on the MOOD approach. We show that GP’s polynomial-free reconstruction provides a seamless pathway to the MOOD’s order cascading formalism by utilizing GP’s novel property of variable (2R + 1)th-order spatial accuracy on a multidimensional GP stencil defined by the GP radius R, whose size is smaller than that of the standard polynomial MOOD methods. The resulting GP-MOOD method is a positivity-preserving method. We examine the numerical stability and accuracy of GP-MOOD on smooth and discontinuous flows in multiple spatial dimensions without resorting to any conventional, computationally expensive a priori nonlinear limiting mechanism to maintain numerical stability.
- 1 January 2022
- Published by Elsevier in SSRN Electronic Journal
Open AccessGp-Mood: A Positive-Preserving High-Order Finite Volume Method for Hyperbolic Conservation Laws
We present an a posteriori shock-capturing finite volume method algorithm called GP-MOOD that solves a compressible hyperbolic conservative system at high-order solution accuracy (e.g., third-, fifth-, and seventh-order) in multiple spatial dimensions. The GP-MOOD method combines two methodologies, the polynomial-free spatial reconstruction methods of GP (Gaussian Process) and the a posteriori detection algorithms of MOOD (Multidimensional Optimal Order Detection). The spatial approximation of our GP-MOOD method uses GP’s unlimited spatial reconstruction that builds upon our previous studies on GP reported in Reyes et al., Journal of Scientific Computing, 76 (2017) and Journal of Computational Physics, 381 (2019). This paper focuses on ex- tending GP’s flexible variability of spatial accuracy to an a posteriori detection formalism based on the MOOD approach. We show that GP’s polynomial-free reconstruction provides a seamless pathway to the MOOD’s order cascading formalism by utilizing GP’s novel property of variable (2R + 1)th-order spatial accuracy on a multidimensional GP stencil defined by the GP radius R, whose size is smaller than that of the standard polynomial MOOD methods. The resulting GP-MOOD method is a positivity-preserving method. We examine the numerical stability and accuracy of GP-MOOD on smooth and discontinuous flows in multiple spatial dimensions without resorting to any conventional, computationally expensive a priori nonlinear limiting mechanism to maintain numerical stability.