arXiv subjects

Kihong Kim

Publications and source records attributed to Kihong Kim.

At least 19 recordsLinked to original sources

arXiv2026

Symmetry classification of temporal reciprocity in time-varying electromagnetic media

Time-varying electromagnetic media exhibit rich nonstationary wave phenomena, but the symmetry governing reversal of arbitrary temporal modulation sequences has remained unclear. We show that, in lossless, spatially homogeneous media with identical initial and final states, the scattering matrices of ordered and reversed sequences are related by inverse--conjugation, independent of the number of stages. This yields a classification of temporal reciprocity in bi-isotropic media: isotropic and chiral media are channel-preserving, whereas Tellegen media are channel-exchanging despite Lorentz nonreciprocity. Deterministic time rewinding follows directly. Our results provide a framework for predicting and designing temporal scattering responses in photonic media.

physics.optics↗

arXiv2025

Deterministic time rewinding of waves in time-varying media

Temporal modulation of material parameters offers unprecedented control over wave dynamics, enabling phenomena beyond the capabilities of static systems. Here we introduce and analyze a robust mechanism for time rewinding, whereby a temporally evolved wave is fully restored to its original state through a carefully engineered sequence of temporal modulations. In electromagnetic systems, time rewinding emerges from impedance-matched or anti-matched hierarchical bilayer structures with matched modulation durations, exploiting total transmission or reflection and reversed phase accumulation. In Dirac systems, it arises via complete interband transition driven by time-dependent vector potentials. Unlike time-reversal holography or quantum time mirrors, which produce wave echoes but only partial waveform recovery, our approach achieves deterministic and complete reconstruction of the entire wave state, including both amplitude and phase. Analytical conditions for robust amplitude and phase restoration are derived and validated through simulations of discrete and continuous modulations, demonstrating resilience to modulation complexity and temporal asymmetry. These findings establish a versatile platform for secure information retrieval, temporal cloaking, programmable metamaterials, and wave-based logic devices.

physics.optics↗

arXiv2025

Disorder-enabled directional delocalization and wave steering in time-modulated Dirac materials

We demonstrate a disorder-enabled yet localization-immune directional transport channel in time-modulated Dirac systems subject to stochastic temporal variations of a vector potential. In a spatially uniform medium, random temporal modulation induces strong Anderson localization for generic propagation directions, whereas waves propagating parallel to the modulation axis remain perfectly delocalized. This behavior originates from the pseudospin structure of the Dirac equation, which enforces exact suppression of interband coupling for specific propagation directions, thereby eliminating disorder-induced backscattering. As a result, temporal disorder acts as a symmetry-selective angular filter, producing highly collimated transport withoutspatial structuring. Unlike conventional impedance matching-based transmission in clean time-varying media, this mechanism arises intrinsically from stochastic temporal modulation and remains robust across a wide range of disorder models. These findings establish temporal disorder as a resource for direction-selective wave control, enabling reconfigurable beam steering, adaptive filtering, and disorder-tolerant nanophotonic components such as temporal beam shapers. More broadly, this phenomenon represents a temporal analogue of disorder-induced delocalization channels known in spatially disordered systems and demonstrates that randomness, typically associated with localization and transport suppression, can instead isolate aperfectly transmitting channel through symmetry-selective dynamics.

physics.optics↗

arXiv2025

Topological edge states and disorder robustness in one-dimensional off-diagonal mosaic lattices

We investigate topological edge states in one-dimensional off-diagonal mosaic lattices, where nearest-neighbor hopping amplitudes are modulated periodically with period $\kappa>1$. Analytically, we demonstrate that discrete edge states emerge at energy levels $E=\epsilon+2t\cos(\pi i/\kappa)$ ($i=1,\cdots,\kappa-1$), extending the Su-Schrieffer-Heeger model to multi-band systems. Numerical simulations show that these edge states are robustly localized and display characteristic nodal structures, with their existence being strongly dictated by the specific edge arrangement of long and short bonds. We further examine their stability under off-diagonal disorder, where the hopping amplitudes $\beta$ fluctuate randomly at intervals of $\kappa$. Using the inverse participation ratio as a localization measure, we show that these topological edge states remain robust over a broad range of disorder strengths. In contrast, additional $\beta$-dependent edge states that appear for $\kappa \ge 4$ are fragile and vanish even under relatively weak disorder. These findings highlight a rich interplay between topology, periodic modulation, and disorder, offering insights for engineering multi-gap topological phases and their realization in synthetic quantum and photonic systems.

physics.optics↗

arXiv2025

Statistical regimes of electromagnetic wave propagation in randomly time-varying media

Wave propagation in time-varying media enables unique control of energy transport by breaking energy conservation through temporal modulation. Among the resulting phenomena, temporal disorder-random fluctuations in material parameters-can suppress propagation and induce localization, analogous to Anderson localization. However, the statistical nature of this process remains incompletely understood. We present a comprehensive analytical and numerical study of electromagnetic wave propagation in spatially uniform media with randomly time-varying permittivity. Using the invariant imbedding method, we derive exact moment equations and identify three distinct statistical regimes for initially unidirectional input: gamma-distributed energy at early times, negative exponential statistics at intermediate times, and a quasi-log-normal distribution at long times, distinct from the true log-normal. In contrast, symmetric bidirectional input yields genuine log-normal statistics across all time scales. These findings are validated using two complementary disorder models--delta-correlated Gaussian noise and piecewise-constant fluctuations--demonstrating that the observed statistics are robust and governed by input symmetry. Momentum conservation constrains the long-time behavior, linking the statistical outcome to the initial conditions. Our results establish a unified framework for understanding statistical wave dynamics in time-modulated systems and offer guiding principles for the design of dynamically tunable photonic and electromagnetic devices.

physics.optics↗

arXiv2025

A Novel Analysis Framework for Microstructural Characterization of Ferroelectric Hafnia: Experimental Validation and Application

Herein, we present a novel analysis framework for grain size profile of ferroelectric hafnia to tackle critical shortcomings inherent in the current microstructural analysis. We vastly enhanced visibility of grains with ion beam treatment and performed accurate grain segmentation using deep neural network (DNN). By leveraging our new method, we discovered unexpected discrepancies that contradict previous results, such as deposition temperature (Tdep) and post-metallization annealing (PMA) dependence of grain size statistics, prompting us to reassess earlier interpretations. Combining microstructural analysis with electrical tests, we found that grain size reduction had both positive and negative outcomes: it caused significant diminishing of die-to-die variation (~68 % decrease in standard deviation) in coercive field (Ec), while triggering an upsurge in leakage current. These uncovered results signify robustness of our method in characterization of ferroelectric hafnia for in-depth examination of both device variability and reliability.

cond-mat.mtrl-sci↗

arXiv2024

Spatial localization and diffusion of Dirac particles and waves induced by random temporal medium variations

Wave propagation in time-varying media has attracted significant attention for its innovative potential to control wave-matter interactions and to develop versatile active materials. While most research has focused on electromagnetic waves, studies on Dirac-type waves remain limited. In this work, we investigate temporal scattering in pseudospin-1/2 Dirac systems with random temporal mass variations. Using the invariant imbedding method, we derive exact expressions for temporal reflectance in both short- and long-time regimes. In the long-time limit, reflectance probabilities become uniformly distributed, and wave group velocities decay to zero, indicating spatial localization. Numerical simulations reveal that narrow wave pulses evolve into Gaussian shapes, with their centers localizing and their widths growing indefinitely due to diffusive behavior. This universal phenomenon is independent of the initial pulse profile and the statistical properties of the random mass. Our findings demonstrate that random temporal variations can induce insulating behavior in Dirac materials, offering potential applications in solid-state physics and optics.

cond-mat.dis-nn↗

arXiv2024

Power-law localization in one-dimensional systems with nonlinear disorder under fixed input conditions

We conduct a numerical investigation into wave propagation and localization in one-dimensional lattices subject to nonlinear disorder, focusing on cases with fixed input conditions. Utilizing a discrete nonlinear Schrödinger equation with Kerr-type nonlinearity and a random coefficient, we compute the averages and variances of the transmittance, $T$, and its logarithm, as functions of the system size $L$, while maintaining constant intensity for the incident wave. In cases of purely nonlinear disorder, we observe power-law localization characterized by $\langle T \rangle \propto L^{-γ_a}$ and $\langle \ln T \rangle \approx -γ_g \ln L$ for sufficiently large $L$. At low input intensities, a transition from exponential to power-law decay in $\langle T \rangle$ occurs as $L$ increases. The exponents $γ_a$ and $γ_g$ are nearly identical, converging to approximately 0.5 as the strength of the nonlinear disorder, $β$, increases. Additionally, the variance of $T$ decays according to a power law with an exponent close to 1, and the variance of $\ln T$ approaches a small constant as $L$ increases. These findings are consistent with an underlying log-normal distribution of $T$ and suggest that wave propagation behavior becomes nearly deterministic as the system size increases. When both linear and nonlinear disorders are present, we observe a transition from power-law to exponential decay in transmittance with increasing $L$ when the strength of linear disorder, $V$, is less than $β$. As $V$ increases, the region exhibiting power-law localization diminishes and eventually disappears when $V$ exceeds $β$, leading to standard Anderson localization.

cond-mat.dis-nn↗

arXiv2024

Wavespace: A Highly Explorable Wavetable Generator

Wavetable synthesis generates quasi-periodic waveforms of musical tones by interpolating a list of waveforms called wavetable. As generative models that utilize latent representations offer various methods in waveform generation for musical applications, studies in wavetable generation with invertible architecture have also arisen recently. While they are promising, it is still challenging to generate wavetables with detailed controls in disentangling factors within the latent representation. In response, we present Wavespace, a novel framework for wavetable generation that empowers users with enhanced parameter controls. Our model allows users to apply pre-defined conditions to the output wavetables. We employ a variational autoencoder and completely factorize its latent space to different waveform styles. We also condition the generator with auxiliary timbral and morphological descriptors. This way, users can create unique wavetables by independently manipulating each latent subspace and descriptor parameters. Our framework is efficient enough for practical use; we prototyped an oscillator plug-in as a proof of concept for real-time integration of Wavespace within digital audio workspaces (DAWs).

cs.SD↗

arXiv2024

Scaling behavior of the localization length for TE waves at critical incidence on short-range correlated stratified random media

We theoretically investigate the scaling behavior of the localization length for $s$-polarized electromagnetic waves incident at a critical angle on stratified random media with short-range correlated disorder. By employing the invariant embedding method, extended to waves in correlated random media, and utilizing the Shapiro-Loginov formula of differentiation, we accurately compute the localization length $ξ$ of $s$ waves incident obliquely on stratified random media that exhibit short-range correlated dichotomous randomness in the dielectric permittivity. The random component of the permittivity is characterized by the disorder strength parameter $σ^2$ and the disorder correlation length $l_c$. Away from the critical angle, $ξ$ depends on these parameters independently. However, precisely at the critical angle, we discover that for waves with wavenumber $k$, $kξ$ depends on the single parameter $kl_cσ^2$, satisfying a universal equation $kξ\approx 1.3717\left(kl_cσ^2\right)^{-1/3}$ across the entire range of parameter values. Additionally, we find that $ξ$ scales as $λ^{4/3}$ for the entire range of the wavelength $λ$, regardless of the values of $σ^2$ and $l_c$. We demonstrate that under sufficiently strong disorder, the scaling behavior of the localization length for all other incident angles converges to that for the critical incidence.

physics.optics↗

arXiv2024

Hybrid Video Diffusion Models with 2D Triplane and 3D Wavelet Representation

Generating high-quality videos that synthesize desired realistic content is a challenging task due to their intricate high-dimensionality and complexity of videos. Several recent diffusion-based methods have shown comparable performance by compressing videos to a lower-dimensional latent space, using traditional video autoencoder architecture. However, such method that employ standard frame-wise 2D and 3D convolution fail to fully exploit the spatio-temporal nature of videos. To address this issue, we propose a novel hybrid video diffusion model, called HVDM, which can capture spatio-temporal dependencies more effectively. The HVDM is trained by a hybrid video autoencoder which extracts a disentangled representation of the video including: (i) a global context information captured by a 2D projected latent (ii) a local volume information captured by 3D convolutions with wavelet decomposition (iii) a frequency information for improving the video reconstruction. Based on this disentangled representation, our hybrid autoencoder provide a more comprehensive video latent enriching the generated videos with fine structures and details. Experiments on video generation benchamarks (UCF101, SkyTimelapse, and TaiChi) demonstrate that the proposed approach achieves state-of-the-art video generation quality, showing a wide range of video applications (e.g., long video generation, image-to-video, and video dynamics control).

cs.CV↗

arXiv2024

MoDiTalker: Motion-Disentangled Diffusion Model for High-Fidelity Talking Head Generation

Conventional GAN-based models for talking head generation often suffer from limited quality and unstable training. Recent approaches based on diffusion models aimed to address these limitations and improve fidelity. However, they still face challenges, including extensive sampling times and difficulties in maintaining temporal consistency due to the high stochasticity of diffusion models. To overcome these challenges, we propose a novel motion-disentangled diffusion model for high-quality talking head generation, dubbed MoDiTalker. We introduce the two modules: audio-to-motion (AToM), designed to generate a synchronized lip motion from audio, and motion-to-video (MToV), designed to produce high-quality head video following the generated motion. AToM excels in capturing subtle lip movements by leveraging an audio attention mechanism. In addition, MToV enhances temporal consistency by leveraging an efficient tri-plane representation. Our experiments conducted on standard benchmarks demonstrate that our model achieves superior performance compared to existing models. We also provide comprehensive ablation studies and user study results.

cs.CV↗

arXiv2024

ExeGPT: Constraint-Aware Resource Scheduling for LLM Inference

This paper presents ExeGPT, a distributed system designed for constraint-aware LLM inference. ExeGPT finds and runs with an optimal execution schedule to maximize inference throughput while satisfying a given latency constraint. By leveraging the distribution of input and output sequences, it effectively allocates resources and determines optimal execution configurations, including batch sizes and partial tensor parallelism. We also introduce two scheduling strategies based on Round-Robin Allocation and Workload-Aware Allocation policies, suitable for different NLP workloads. We evaluate ExeGPT on six LLM instances of T5, OPT, and GPT-3 and five NLP tasks, each with four distinct latency constraints. Compared to FasterTransformer, ExeGPT achieves up to 15.2x improvements in throughput and 6x improvements in latency. Overall, ExeGPT achieves an average throughput gain of 2.9x across twenty evaluation scenarios. Moreover, when adapting to changing sequence distributions, the cost of adjusting the schedule in ExeGPT is reasonably modest. ExeGPT proves to be an effective solution for optimizing and executing LLM inference for diverse NLP workload and serving conditions.

cs.DC↗

arXiv2023

Transport and localization properties of excitations in one-dimensional lattices with diagonal disordered mosaic modulations

We present a numerical study of the transport and localization properties of excitations in one-dimensional lattices with diagonal disordered mosaic modulations. The model is characterized by the modulation period $κ$ and the disorder strength $W$. We calculate the disorder averages $\langle T\rangle$, $\langle \ln T\rangle$, and $\langle P\rangle$, where $T$ is the transmittance and $P$ is the participation ratio, as a function of energy $E$ and system size $L$, for different values of $κ$ and $W$. For excitations at quasiresonance energies determined by $κ$, we find power-law scaling behaviors of the form $\langle T \rangle \propto L^{-γ_{a}}$, $\langle \ln T \rangle \approx -γ_g \ln L$, and $\langle P \rangle \propto L^β$, as $L$ increases to a large value. This behavior is in contrast to the exponential localization behavior occurring at all other energies. The appearance of sharp peaks in the participation ratio spectrum at quasiresonance energies provides additional evidence for the existence of an anomalous power-law localization phenomenon. The corresponding eigenstates demonstrate multifractal behavior and exhibit unique node structures. In addition, we investigate the time-dependent wave packet dynamics and calculate the mean square displacement $\langle m^2(t) \rangle$, spatial probability distribution, participation number, and return probability. When the wave packet's initial momentum satisfies the quasiresonance condition, we observe a subdiffusive spreading of the wave packet, characterized by $\langle m^2(t) \rangle\propto t^η$ where $η$ is always less than 1. We also note the occurrence of partial localization at quasiresonance energies, as indicated by the saturation of the participation number and a nonzero value for the return probability at long times.

cond-mat.dis-nn↗

arXiv2023

Propagation of Dirac waves through various temporal interfaces, slabs, and crystals

We investigate the influence of the temporal variations of various medium parameters on the propagation of Dirac-type waves in materials where the quasiparticles are described by a generalized version of the pseudospin-1/2 Dirac equation. Our considerations also include the propagation of electromagnetic waves in metamaterials with the Dirac-type dispersion. We focus on the variations of the scalar and vector potentials, mass, Fermi velocity, and tilt velocity describing the Dirac cone tilt. We derive the scattering coefficients associated with the temporal interfaces and slabs analytically and find that the temporal scattering is caused by the changes of the mass, Fermi velocity, and vector potential, but does not arise from the changes of the scalar potential and tilt velocity. We also explore the conditions under which the temporal Brewster effect and total interband transition occur and calculate the change in total wave energy. We examine bilayer Dirac temporal crystals where parameters switch between two different sets of values periodically and prove that these systems do not have momentum gaps. Finally, we assess the potential for observing these temporal scattering effects in experiments.

cond-mat.mes-hall↗

arXiv2022

DiffFace: Diffusion-based Face Swapping with Facial Guidance

In this paper, we propose a diffusion-based face swapping framework for the first time, called DiffFace, composed of training ID conditional DDPM, sampling with facial guidance, and a target-preserving blending. In specific, in the training process, the ID conditional DDPM is trained to generate face images with the desired identity. In the sampling process, we use the off-the-shelf facial expert models to make the model transfer source identity while preserving target attributes faithfully. During this process, to preserve the background of the target image and obtain the desired face swapping result, we additionally propose a target-preserving blending strategy. It helps our model to keep the attributes of the target face from noise while transferring the source facial identity. In addition, without any re-training, our model can flexibly apply additional facial guidance and adaptively control the ID-attributes trade-off to achieve the desired results. To the best of our knowledge, this is the first approach that applies the diffusion model in face swapping task. Compared with previous GAN-based approaches, by taking advantage of the diffusion model for the face swapping task, DiffFace achieves better benefits such as training stability, high fidelity, diversity of the samples, and controllability. Extensive experiments show that our DiffFace is comparable or superior to the state-of-the-art methods on several standard face swapping benchmarks.

cs.CV↗

arXiv2022

Delocalization and re-entrant localization of flat-band states in non-Hermitian disordered lattice models with flat bands

We present a numerical study of Anderson localization in disordered non-Hermitian lattice models with flat bands. Specifically we consider one-dimensional stub and two-dimensional kagome lattices that have a random scalar potential and a uniform imaginary vector potential and calculate the spectra of the complex energy, the participation ratio, and the winding number as a function of the strength of the imaginary vector potential, $h$. The flat-band states are found to show a double transition from localized to delocalized and back to localized states with $h$, in contrast to the dispersive-band states going through a single delocalization transition. When $h$ is sufficiently small, all flat-band states are localized. As $h$ increases above a certain critical value $h_1$, some pair of flat-band states become delocalized. The participation ratio associated with them increases substantially and their winding numbers become nonzero. As $h$ increases further, more and more flat-band states get delocalized until the fraction of the delocalized states reaches a maximum. For larger $h$ values, a re-entrant localization takes place and, at another critical value $h_2$, all flat-band states return to compact localized states with very small participation ratios and zero winding numbers. This re-entrant localization transition, which is due to the interplay among disorder, non-Hermiticity, and flat band, is a phenomenon occurring in many models having an imaginary vector potential and a flat band simultaneously. We explore the spatial characteristics of the flat-band states by calculating the local density distribution.

cond-mat.dis-nn↗

arXiv2022

Giant overreflection of magnetohydrodynamic waves from inhomogeneous plasmas with nonuniform shear flows

We study theoretically mode conversion and resonant overreflection of magnetohydrodynamic waves in an inhomogeneous plane-stratified plasma in the presence of a nonuniform shear flow, using precise numerical calculations of the reflection and transmission coefficients and the field distributions based on the invariant imbedding method. The cases where the flow velocity and the external magnetic field are directed perpendicularly to the inhomogeneity direction and both the flow velocity and the plasma density vary arbitrarily along it are considered. When there is a shear flow, the wave frequency is modulated locally by the Doppler shift and resonant amplification and overreflection occur where the modulated frequency is negative and its absolute value matches the local Alfvén or slow frequency. For many different types of the density and flow velocity profiles, we find that, especially when the parameters are such that the incident waves are totally reflected, there arises a giant overreflection where the reflectance is much larger than 10 in a fairly broad range of the incident angles, the frequency, and the plasma $β$ and its maximum attains values larger than $10^5$. In a finite $β$ plasma, both incident fast and slow magnetosonic waves are found to cause strong overreflection and there appear multiple positions exhibiting both Alfvén and slow resonances inside the plasma. We explain the mechanism of overreflection in terms of the formation of inhomogeneous and open cavities close to the resonances and the strong enhancement of the wave energy due to the occurrence of semi-bound states there. We give discussions of the observational consequences in magnetized terrestrial and solar plasmas.

physics.plasm-ph↗