arXiv subjects

Minsung Kim

Publications and source records attributed to Minsung Kim.

52 records · Page 3Linked to original sources

arXiv2023

Blowup Equations for Little Strings

We propose blowup equations for 6d little string theories which generalize Nakajima-Yoshioka's blowup equations for the 4d/5d instanton partition functions on Omega background. We find that unlike the blowup equations for standard SQFTs, we need to sum over auxiliary magnetic fluxes on the blown-up $ \mathbb{P}^1$ for a non-dynamical 2-form gauge field which plays a role in canceling the mixed anomalies of the gauge symmetries. We demonstrate with explicit examples that the blowup equations, when combined with the modular properties, can be solved in order to determine the elliptic genera of little strings.

hep-th↗

arXiv2022

Bag of Tricks for Developing Diabetic Retinopathy Analysis Framework to Overcome Data Scarcity

Recently, diabetic retinopathy (DR) screening utilizing ultra-wide optical coherence tomography angiography (UW-OCTA) has been used in clinical practices to detect signs of early DR. However, developing a deep learning-based DR analysis system using UW-OCTA images is not trivial due to the difficulty of data collection and the absence of public datasets. By realistic constraints, a model trained on small datasets may obtain sub-par performance. Therefore, to help ophthalmologists be less confused about models' incorrect decisions, the models should be robust even in data scarcity settings. To address the above practical challenging, we present a comprehensive empirical study for DR analysis tasks, including lesion segmentation, image quality assessment, and DR grading. For each task, we introduce a robust training scheme by leveraging ensemble learning, data augmentation, and semi-supervised learning. Furthermore, we propose reliable pseudo labeling that excludes uncertain pseudo-labels based on the model's confidence scores to reduce the negative effect of noisy pseudo-labels. By exploiting the proposed approaches, we achieved 1st place in the Diabetic Retinopathy Analysis Challenge.

eess.IV↗

arXiv2022

Predictive Modeling of Charge Levels for Battery Electric Vehicles using CNN EfficientNet and IGTD Algorithm

Convolutional Neural Networks (CNN) have been a good solution for understanding a vast image dataset. As the increased number of battery-equipped electric vehicles is flourishing globally, there has been much research on understanding which charge levels electric vehicle drivers would choose to charge their vehicles to get to their destination without any prevention. We implemented deep learning approaches to analyze the tabular datasets to understand their state of charge and which charge levels they would choose. In addition, we implemented the Image Generator for Tabular Dataset algorithm to utilize tabular datasets as image datasets to train convolutional neural networks. Also, we integrated other CNN architecture such as EfficientNet to prove that CNN is a great learner for reading information from images that were converted from the tabular dataset, and able to predict charge levels for battery-equipped electric vehicles. We also evaluated several optimization methods to enhance the learning rate of the models and examined further analysis on improving the model architecture.

cs.CV↗

arXiv2022

Limit theorems for higher rank actions on Heisenberg nilmanifolds

The main result of this paper is a construction of finitely additive measures for higher rank abelian actions on Heisenberg nilmanifolds. Under a full measure set of Diophantine conditions for the generators of the action, we construct \emph{Bufetov functionals} on rectangles on $(2g+1)$-dimensional Heisenberg manifolds. We prove that deviation of the ergodic integral of higher rank actions is described by the asymptotic of Bufetov functionals for a sufficiently smooth function. As a corollary, the distribution of normalized ergodic integrals which have variance 1, converges along certain subsequences to a non-degenerate compactly supported measure on the real line.

math.DS↗

arXiv2021

New phenomena in deviation of Birkhoff integrals for locally Hamiltonian flows

We consider smooth locally Hamiltonian flows on compact surfaces of genus $g\geq 2$ to prove their deviation of Birkhoff integrals for smooth observables. Our work generalizes results of Forni and Bufetov which prove the existence of deviation spectrum of Birkhoff integrals for observables whose jets vanish at sufficiently high order around fixed points of the flow. They showed that ergodic integrals can display a power spectrum of behaviours with exactly $g$ positive exponents related to the positive Lyapunov exponents of the cocycle so-called Kontsevich-Zorich, a renormalization cocycle over the Teichmüller flow on a stratum of the moduli space of translation surfaces. Our paper extends the study of the spectrum of deviations of ergodic integrals beyond the case of observables whose jets vanish at sufficiently high order around fixed points. We prove the existence of some extra terms in deviation spectrum related to non-vanishing of the derivatives of observables at fixed points. The proof of this new phenomenon is based on tools developed in the recent work of the first author and Ulcigrai for locally Hamiltonian flows having only (simple) non-degenerate saddles. In full generality, due to the occurrence of (multiple) degenerate saddles, we introduce new methods of handling functions with polynomial singularities over a full measure set of IETs.

math.DS↗

arXiv2021

5d/6d Wilson loops from blowups

We generalize Nakajima-Yoshioka's blowup formula to calculate the partition functions counting the spectrum of bound states to half-BPS Wilson loop operators in 5d (and 6d) supersymmetric field theories. The partition function in the presence of a Wilson loop operator on the $Ω$-background is factorized when put on the blowup $\hat{\mathbb{C}}^2$ into two Wilson loop partition functions under the localization. This structure provides a set of blowup equations for Wilson loop operators. We explain how to formulate the blowup equations and solve them to compute the partition functions of Wilson loop operators. We test this idea by explicitly calculating the Wilson loop partition functions in various 5d/6d field theories and comparing them against known results and expected dualities.

hep-th↗

arXiv2021

Effective equidistribution for generalized higher step nilflows

The main results of this paper are to prove bounds for ergodic averages for nilflows on general higher step nilmanifolds. Under Diophantine condition on the frequency of a toral projection of the flow, we prove that almost all orbits become equidistributed at the polynomial speed. We analyze the exponent with the speed of decay which is determined by the number of steps and structure of general nilpotent Lie algebras. The main result follows from the technique over controlling scaling operators in irreducible representations and measure estimation on close return orbit on general nilmanifolds.

math.DS↗

arXiv2021

Lattice dynamics of the excitonic insulator Ta$_2$Ni(Se$_{1-x}$S$_x$)$_5$

Recently, we employed electronic polarization-resolved Raman spectroscopy to reveal the strongly correlated excitonic insulator (EI) nature of Ta2NiSe5, Volkov et al. [arXiv:2007.07344], and also showed that for Ta$_2$Ni(Se$_{1-x}$S$_x$)$_5$ alloys the critical excitonic fluctuations diminish with sulfur concentration x exposing a cooperating lattice instability that takes over for large x, Volkov et al. [arXiv:2104.07032]. Here we focus on the lattice dynamics of this EI family. We identify all Raman-active optical phonons of fully symmetric and ac-quadrupole-like symmetries and study their evolution with temperature and sulfur concentration. We demonstrate the change of selection rules at temperatures below the orthorhombic-to-monoclinic transition at Tc(x) that is related to the EI phase. We find that Tc(x) decrease monotonically from 328 K for Ta2NiSe5 to 120 K for Ta2NiS5 and that the magnitude of lattice distortion also decreases with the sulfur concentration x. For x < 0.7, the two lowest-frequency B2g phonon modes show strongly asymmetric lineshapes at high temperatures due to Fano interference with the broad excitonic continuum present in a semimetallic state. Within the framework of extended Fano model, we develop a quantitative description of the interacting exciton-phonon excitation lineshape, enabling us to derive the intrinsic phonon parameters and determine the exciton-phonon interaction strength, that affects the transition temperature Tc(x). We also observe signatures of the acoustic mode scattered assisted by the structural domain walls formed below Tc. Based on our results, we additionally present a consistent interpretation of the origin of oscillations observed in time-resolved pump-probe experiments.

cond-mat.str-el↗

arXiv2021

Orbital-selective Mott phase and non-Fermi liquid in FePS$_3$

The layered metal phosphorous trisulfide FePS$_3$ is reported to be a Mott insulator at ambient conditions and to undergo structural and insulator-metal phase transitions under pressure. However, the character of the resulting metallic states has not been understood clearly so far. Here, we theoretically study the phase transitions of FePS$_3$ using first-principles methods based on density functional theory and embedded dynamical mean field theory. We find that the Mott transition in FePS$_3$ can be orbital-selective, with $t_{2g}$ states undergoing a correlation-induced insulator-to-metal transition while $e_g$ states remain gapped. We show that this orbital-selective Mott phase, which occurs only when non-hydrostatic pressure is used, is a bad metal (or non-Fermi liquid) with large fluctuating moments due to Hund's coupling. Further application of pressure increases the crystal-field splitting and converts the system to a conventional Fermi liquid with low-spin configurations dominant. Our results show that FePS$_3$ is a novel example of a system that realizes an orbital-selective Mott phase, allowing tuning between correlated and uncorrelated metallic properties in an accessible pressure range ($\leq$ 18 GPa).

cond-mat.str-el↗

arXiv2021

Physics-Inspired Heuristics for Soft MIMO Detection in 5G New Radio and Beyond

Overcoming the conventional trade-off between throughput and bit error rate (BER) performance, versus computational complexity is a long-term challenge for uplink Multiple-Input Multiple-Output (MIMO) detection in base station design for the cellular 5G New Radio roadmap, as well as in next generation wireless local area networks. In this work, we present ParaMax, a MIMO detector architecture that for the first time brings to bear physics-inspired parallel tempering algorithmic techniques [28, 50, 67] on this class of problems. ParaMax can achieve near optimal maximum-likelihood (ML) throughput performance in the Large MIMO regime, Massive MIMO systems where the base station has additional RF chains, to approach the number of base station antennas, in order to support even more parallel spatial streams. ParaMax is able to achieve a near ML-BER performance up to 160x160 and 80x80 Large MIMO for low-order modulations such as BPSK and QPSK, respectively, only requiring less than tens of processing elements. With respect to Massive MIMO systems, in 12x24 MIMO with 16-QAM at SNR 16 dB, ParaMax achieves 330 Mbits/s near-optimal system throughput with 4--8 processing elements per subcarrier, which is approximately 1.4x throughput than linear detector-based Massive MIMO systems.

cs.NI↗

arXiv2021

Bootstrapping BPS spectra of 5d/6d field theories

We propose a systematic approach to computing the BPS spectra of any 5d/6d supersymmetric quantum field theory in Coulomb phases, which admits either gauge theory descriptions or geometric descriptions, based on the Nakajima-Yoshioka's blowup equations. We provide a significant generalization of the blowup equation approach in terms of both properly quantized magnetic fluxes on the blowup $\hat{\mathbb{C}}^2$ and the effective prepotential for 5d/6d field theories on the Omega background which is uniquely determined by the Chern-Simons couplings on their Coulomb branches. We employ our method to compute BPS spectra of all rank-1 and rank-2 5d Kaluza-Klein (KK) theories descending from 6d $\mathcal{N}=(1,0)$ superconformal field theories (SCFTs) compactified on a circle with/without twist. We also discuss various 5d SCFTs and KK theories of higher ranks, which include a few exotic cases such as new rank-1 and rank-2 5d SCFTs engineered with frozen singularity as well as the 5d $SU(3)_8$ gauge theory currently having neither a brane web nor a smooth shrinkable geometric description. The results serve as non-trivial checks for a large class of non-trivial dualities among 5d theories and also as independent evidences for the existence of certain exotic theories.

hep-th↗

arXiv2020

Towards Hybrid Classical-Quantum Computation Structures in Wirelessly-Networked Systems

With unprecedented increases in traffic load in today's wireless networks, design challenges shift from the wireless network itself to the computational support behind the wireless network. In this vein, there is new interest in quantum-compute approaches because of their potential to substantially speed up processing, and so improve network throughput. However, quantum hardware that actually exists today is much more susceptible to computational errors than silicon-based hardware, due to the physical phenomena of decoherence and noise. This paper explores the boundary between the two types of computation---classical-quantum hybrid processing for optimization problems in wireless systems---envisioning how wireless can simultaneously leverage the benefit of both approaches. We explore the feasibility of a hybrid system with a real hardware prototype using one of the most advanced experimentally available techniques today, reverse quantum annealing. Preliminary results on a low-latency, large MIMO system envisioned in the 5G New Radio roadmap are encouraging, showing approximately 2--10X better performance in terms of processing time than prior published results.

cs.NI↗

arXiv2020

Leveraging Quantum Annealing for Large MIMO Processing in Centralized Radio Access Networks

User demand for increasing amounts of wireless capacity continues to outpace supply, and so to meet this demand, significant progress has been made in new MIMO wireless physical layer techniques. Higher-performance systems now remain impractical largely only because their algorithms are extremely computationally demanding. For optimal performance, an amount of computation that increases at an exponential rate both with the number of users and with the data rate of each user is often required. The base station's computational capacity is thus becoming one of the key limiting factors on wireless capacity. QuAMax is the first large MIMO centralized radio access network design to address this issue by leveraging quantum annealing on the problem. We have implemented QuAMax on the 2,031 qubit D-Wave 2000Q quantum annealer, the state-of-the-art in the field. Our experimental results evaluate that implementation on real and synthetic MIMO channel traces, showing that 10~$μ$s of compute time on the 2000Q can enable 48 user, 48 AP antenna BPSK communication at 20 dB SNR with a bit error rate of $10^{-6}$ and a 1,500 byte frame error rate of $10^{-4}$.

cs.NI↗

arXiv2019

Topological band gap in intercalated epitaxial graphene

Functional manipulation of graphene is an important topic in view of both fundamental researches and practical applications. In this study, we show that intercalation of 5$d$ transition metals in epitaxial graphene on SiC is a promising approach to realize topologically nontrivial phases with a finite band gap in graphene. Using first-principles calculations based on density functional theory, we show that the Re- and Ta-intercalated graphene become two-dimensional topological insulators which exhibit linear Dirac cones and quadratic bands with topological band gaps, respectively. The appearance of the topological states is attributed to the strong spin-orbit coupling strength of the intercalants. We find that topological edge states exist within the finite bulk band gap in accordance with the bulk-boundary correspondence. We also discuss the spin splitting of the band structure due to the inversion symmetry breaking and the spin-orbit coupling. Our results demonstrate that intercalation of graphene is an effective and viable method to manipulate the band gap and the topological character of graphene. Such intercalated graphene systems are potentially useful for spintronics and quantum computing applications.

cond-mat.mes-hall↗

arXiv2018

Topological states in A15 superconductors

Superconductors with the A15 structure are prototypical type-II s-wave superconductors which have generated considerable interest in early superconducting material history. However, the topological nature of the electronic structure remains unnoticed so far. Here, using first-principles band structure calculations based on density functional theory, we show that the A15 superconductors (Ta$_3$Sb, Ta$_3$Sn, and Ta$_3$Pb) have nontrivial band topology in the bulk electronic band structures, leading to the formation of topological surface states near the Fermi energy. Due to the bulk superconductivity, the proximity effect in the topological surface states would induce topological superconductivity even without heterostructure of a topological insulator and an s-wave superconductor. Our results indicate that the A15 superconductors are promising candidates for the realization of the topological superconductivity and the Majorana fermion.

cond-mat.mtrl-sci↗

arXiv2016

Strongly enhanced Rashba splittings in oxide heterostructure: a tantalite monolayer on BaHfO$_3$

In the two-dimensional electron gas (2DEG) emerging at the transition metal oxide surface and interface, it has been pointed out that the Rashba spin-orbit interaction, the momentum-dependent spin splitting due to broken inversion symmetry and atomic spin-orbit coupling, can have profound effects on electronic ordering in the spin, orbit, and charge channels, and may help give rise to exotic phenomena such as ferromagnetism-superconductivity coexistence and topological superconductivity. Although a large Rashba splitting is expected to improve experimental accessibility of such phenomena, it has not been understood how we can maximally enhance this splitting. Here, we present a promising route to realize significant Rashba-type band splitting using a thin film heterostructure. Based on first-principles methods and analytic model analyses, a tantalate monolayer on BaHfO$_3$ is shown to host two-dimensional bands originating from Ta $t_{2g}$ states with strong Rashba spin splittings - up to nearly 10% of the bandwidth - at both the band minima and saddle points due to the maximal breaking of the inversion symmetry. Such 2DEG band structure makes this oxide heterostructure a promising platform for realizing both a topological superconductor which hosts Majorana fermions and the electron correlation physics with strong spin-orbit coupling.

cond-mat.str-el↗