arXiv subjects

Sang-Hoon Lee

Publications and source records attributed to Sang-Hoon Lee.

25 records · Page 2Linked to original sources

arXiv2021

Efficient First-Principles Approach with a Pseudohybrid Density Functional for Extended Hubbard Interactions

For fast and accurate calculations of band gaps of solids, we present an {\it ab initio} method that extends the density functional theory plus on-site Hubbard interaction (DFT+$U$) to include inter-site Hubbard interaction ($V$). This formalism is appropriate for considering various interactions such as a local Coulomb repulsion, covalent hybridizations, and their coexistence in solids. To achieve self-consistent evaluations of $U$ and $V$, we adapt a recently proposed Agapito-Curtarolo-Buongiorno Nardelli pseudohybrid functional for DFT$+U$ to implement a density functional of $V$ and obtain band gaps of diverse bulk materials as accurate as those from $GW$ or hybrid functionals methods with a standard DFT computational cost. Moreover, we also show that computed band gaps of few layers black phosphorous and Si(111)-($2\times1$) surface agree with experiments very well, thus meriting the new method for large-scale as well as high throughput calculations with higher accuracy.

cond-mat.mtrl-sci↗

arXiv2020

Multi-SpectroGAN: High-Diversity and High-Fidelity Spectrogram Generation with Adversarial Style Combination for Speech Synthesis

While generative adversarial networks (GANs) based neural text-to-speech (TTS) systems have shown significant improvement in neural speech synthesis, there is no TTS system to learn to synthesize speech from text sequences with only adversarial feedback. Because adversarial feedback alone is not sufficient to train the generator, current models still require the reconstruction loss compared with the ground-truth and the generated mel-spectrogram directly. In this paper, we present Multi-SpectroGAN (MSG), which can train the multi-speaker model with only the adversarial feedback by conditioning a self-supervised hidden representation of the generator to a conditional discriminator. This leads to better guidance for generator training. Moreover, we also propose adversarial style combination (ASC) for better generalization in the unseen speaking style and transcript, which can learn latent representations of the combined style embedding from multiple mel-spectrograms. Trained with ASC and feature matching, the MSG synthesizes a high-diversity mel-spectrogram by controlling and mixing the individual speaking styles (e.g., duration, pitch, and energy). The result shows that the MSG synthesizes a high-fidelity mel-spectrogram, which has almost the same naturalness MOS score as the ground-truth mel-spectrogram.

eess.AS↗

arXiv2020

First-Principles Study of Two-Dimensional Ferroelectrics Using Self-Consistent Hubbard Parameters

The discovery of two-dimensional (2D) materials possessing switchable spontaneous polarization with atomic thickness opens up exciting opportunities to realize ultrathin, high-density electronic devices with potential applications ranging from memories and sensors to photocatalysis and solar cells. First-principles methods based on density functional theory (DFT) have facilitated the discovery and design of 2D ferroelectrics (FEs). However, DFT calculations employing local and semilocal exchange-correlation functionals failed to predict accurately the band gaps for this family of low-dimensional materials. Here, we present a DFT+$U$+$V$ study on 2D FEs represented by single-layer $α$-In$_2$Se$_3$ and its homologous III$_2$-VI$_3$ compounds with both out-of-plane and in-plane polarization, using Hubbard parameters computed from first-principles. We find that ACBN0, a pseudo-hybrid density functional that allows self-consistent determination of $U$ parameters, improves the prediction of band gaps for all investigated 2D FEs with a computational cost much lower than the hybrid density functional. The inter-site Coulomb interaction $V$ becomes critical for accurate descriptions of the electronic structures of van der Waals heterostructures such as bilayer In$_2$Se$_3$ and In$_2$Se$_3$/InTe. Pertinent to the study of FE-based catalysis, we find that the application of self-consistent $U$ corrections can strongly affect the adsorption energies of open-shell molecules on the polar surfaces of 2D FEs.

cond-mat.mtrl-sci↗

arXiv2020

Audio Dequantization for High Fidelity Audio Generation in Flow-based Neural Vocoder

In recent works, a flow-based neural vocoder has shown significant improvement in real-time speech generation task. The sequence of invertible flow operations allows the model to convert samples from simple distribution to audio samples. However, training a continuous density model on discrete audio data can degrade model performance due to the topological difference between latent and actual distribution. To resolve this problem, we propose audio dequantization methods in flow-based neural vocoder for high fidelity audio generation. Data dequantization is a well-known method in image generation but has not yet been studied in the audio domain. For this reason, we implement various audio dequantization methods in flow-based neural vocoder and investigate the effect on the generated audio. We conduct various objective performance assessments and subjective evaluation to show that audio dequantization can improve audio generation quality. From our experiments, using audio dequantization produces waveform audio with better harmonic structure and fewer digital artifacts.

eess.AS↗

arXiv2019

Prediction of ferroelectricity-driven Berry curvature enabling charge- and spin-controllable photocurrent in tin telluride monolayers

In symmetry-broken crystalline solids, pole structures of Berry curvature (BC) can emerge, and they have been utilized as a versatile tool for controlling transport properties. For example, the monopole component of the BC is induced by the time-reversal symmetry breaking, and the BC dipole arises from a lack of inversion symmetry, leading to the anomalous Hall and nonlinear Hall effects, respectively. Based on first-principles calculations, we show that the ferroelectricity in a tin telluride monolayer produces a unique BC distribution, which offers charge- and spin-controllable photocurrents. Even with the sizable band gap, the ferroelectrically driven BC dipole is comparable to those of small-gap topological materials. By manipulating the photon handedness and the ferroelectric polarization, charge and spin circular photogalvanic currents are generated in a controllable manner. The ferroelectricity in group-IV monochalcogenide monolayers can be a useful tool to control the BC dipole and the nonlinear optoelectronic responses.

cond-mat.mes-hall↗

arXiv2016

Quantum electronic transport of topological surface states in beta-Ag2Se nanowire

Single-crystalline β-Ag2Se nanostructures, a new class of 3D topological insulators (TIs), were synthesized using the chemical vapor transport method. The topological surface states were verified by measuring electronic transport properties including the weak antilocalization effect, Aharonov-Bohm oscillations, and Shubnikov-de Haas oscillations. First-principles band calculations revealed that the band inversion in \b{eta}-Ag2Se is caused by strong spin-orbit coupling and Ag-Se bonding hybridization. These extensive investigations provide new meaningful information about silver-chalcogenide TIs that have anisotropic Dirac cones, which could be useful for spintronics applications.

cond-mat.mes-hall↗

arXiv2015

Proposed searches for candidate sources of gravitational waves in a nearby core-collapse supernova survey

Gravitational wave bursts in the formation of neutron stars and black holes in energetic core-collapse supernovae (CC-SNe) are of potential interest to LIGO-Virgo and KAGRA. Events nearby are readily discovered using moderately sized telescopes. CC-SNe are competitive with mergers of neutron stars and black holes, if the fraction producing an energetic output in gravitational waves exceeds about 1\%. This opportunity motivates the design of a novel Sejong University Core-CollapsE Supernova Survey (SUCCESS), to provide triggers for follow-up searches for gravitational waves. It is based on the 76 cm Sejong University Telescope (SUT) for weekly monitoring of nearby star-forming galaxies, i.e., M51, M81-M82 and Blue Dwarf Galaxies from the Unified Nearby Galaxy Catalog with an expected yield of a few hundred per year. Optical light curves will be resolved for the true time-of-onset for probes of gravitational waves by broadband time-sliced matched filtering.

astro-ph.HE↗