arXiv subjects

Gang Li

Publications and source records attributed to Gang Li.

At least 469 records · Page 26Linked to original sources

arXiv2021

A Flexible Joint Model for Multiple Longitudinal Biomarkers and A Time-to-Event Outcome: With Applications to Dynamic Prediction Using Highly Correlated Biomarkers

In biomedical studies it is common to collect data on multiple biomarkers during study follow-up for dynamic prediction of a time-to-event clinical outcome. The biomarkers are typically intermittently measured, missing at some event times, and may be subject to high biological variations, which cannot be readily used as time-dependent covariates in a standard time-to-event model. Moreover, they can be highly correlated if they are from in the same biological pathway. To address these issues, we propose a flexible joint model framework that models the multiple biomarkers with a shared latent reduced rank longitudinal principal component model and correlates the latent process to the event time by the Cox model for dynamic prediction of the event time. The proposed joint model for highly correlated biomarkers is more flexible than some existing methods since the latent trajectory shared by the multiple biomarkers does not require specification of a priori parametric time trend and is determined by data. We derive an Expectation-Maximization (EM) algorithm for parameter estimation, study large sample properties of the estimators, and adapt the developed method to make dynamic prediction of the time-to-event outcome. Bootstrap is used for standard error estimation and inference. The proposed method is evaluated using simulations and illustrated on a lung transplant data to predict chronic lung allograft dysfunction (CLAD) using chemokines measured in bronchoalveolar lavage fluid of the patients.

stat.ME↗

arXiv2021

The axial charges of proton within an extended chiral constituent quark model

We have performed a study of the isovector, octet and singlet axial charges of the proton in an extended chiral constituent quark model, where all the possible $uudq\bar{q}$~($q=u,d,s$) five-quark Fock components in the proton wave function are taken into account. The $^3P_0$ quark-antiquark creation mechanism is assumed to account for the transition coupling between three- and five-quark components in proton, and the corresponding transition coupling strength is fixed by fitting the intrinsic sea flavor asymmetry $\bar{d}-\bar{u}$ data for proton. Accordingly, with all the parameters fixed by empirical values, the probabilities of the intrinsic five-quark Fock components in proton wave function should be $\sim30 - 50\%$, which lead to the numerical results for quark spin $Δu$, $Δd$ and $Δs$, as well the axial charges of proton consistent with the experimental data and predictions by other theoretical approaches.

hep-ph↗

arXiv2021

Block Convolution: Towards Memory-Efficient Inference of Large-Scale CNNs on FPGA

Deep convolutional neural networks have achieved remarkable progress in recent years. However, the large volume of intermediate results generated during inference poses a significant challenge to the accelerator design for resource-constraint FPGA. Due to the limited on-chip storage, partial results of intermediate layers are frequently transferred back and forth between on-chip memory and off-chip DRAM, leading to a non-negligible increase in latency and energy consumption. In this paper, we propose block convolution, a hardware-friendly, simple, yet efficient convolution operation that can completely avoid the off-chip transfer of intermediate feature maps at run-time. The fundamental idea of block convolution is to eliminate the dependency of feature map tiles in the spatial dimension when spatial tiling is used, which is realized by splitting a feature map into independent blocks so that convolution can be performed separately on individual blocks. We conduct extensive experiments to demonstrate the efficacy of the proposed block convolution on both the algorithm side and the hardware side. Specifically, we evaluate block convolution on 1) VGG-16, ResNet-18, ResNet-50, and MobileNet-V1 for ImageNet classification task; 2) SSD, FPN for COCO object detection task, and 3) VDSR for Set5 single image super-resolution task. Experimental results demonstrate that comparable or higher accuracy can be achieved with block convolution. We also showcase two CNN accelerators via algorithm/hardware co-design based on block convolution on memory-limited FPGAs, and evaluation shows that both accelerators substantially outperform the baseline without off-chip transfer of intermediate feature maps.

cs.AR↗

arXiv2021

Decentralized Federated Learning via Mutual Knowledge Transfer

In this paper, we investigate the problem of decentralized federated learning (DFL) in Internet of things (IoT) systems, where a number of IoT clients train models collectively for a common task without sharing their private training data in the absence of a central server. Most of the existing DFL schemes are composed of two alternating steps, i.e., model updating and model averaging. However, averaging model parameters directly to fuse different models at the local clients suffers from client-drift especially when the training data are heterogeneous across different clients. This leads to slow convergence and degraded learning performance. As a possible solution, we propose the decentralized federated earning via mutual knowledge transfer (Def-KT) algorithm where local clients fuse models by transferring their learnt knowledge to each other. Our experiments on the MNIST, Fashion-MNIST, CIFAR-10, and CIFAR-100 datasets reveal that the proposed Def-KT algorithm significantly outperforms the baseline DFL methods with model averaging, i.e., Combo and FullAvg, especially when the training data are not independent and identically distributed (non-IID) across different clients.

cs.LG↗

arXiv2021

A homogeneous spectroscopic analysis of a Kepler legacy sample of dwarfs for gravity-mode asteroseismology

Asteroseismic modelling of the internal structure of main-sequence stars born with a convective core has so far been based on homogeneous analyses of space photometric Kepler light curves of 4 years duration, to which most often incomplete inhomogeneously deduced spectroscopic information was added to break degeneracies. We composed a sample of 111 dwarf gravity-mode pulsators observed by the Kepler space telescope whose light curves allowed for determination of their near-core rotation rates. For this sample we assembled HERMES high-resolution optical spectroscopy at the 1.2-m Mercator telescope. Our spectroscopic information offers additional observational input to also model the envelope layers of these non-radially pulsating dwarfs. We determined stellar parameters and surface abundances in a homogeneous way from atmospheric analysis with spectrum normalisation based on a new machine learning tool. Our results suggest a systematic overestimation of [M/H] in the literature for the studied F-type dwarfs, presumably due to normalisation limitations caused by the dense line spectrum of these rotating stars. CNO-surface abundances were found to be uncorrelated with the rotation properties of the F-type stars. For the B-type stars, we find a hint of deep mixing from C and O abundance ratios; N abundances have too large uncertainties to reveal a correlation with the rotation of the stars. Our spectroscopic stellar parameters and abundance determinations allow for future joint spectroscopic, astrometric (Gaia), and asteroseismic modelling of this legacy sample of gravity-mode pulsators, with the aim to improve our understanding of transport processes in the core-hydrogen burning phase of stellar evolution.

astro-ph.SR↗

arXiv2021

Electron-plasmon interaction induced plasmonic-polaron band replication in epitaxial perovskite SrIrO$_3$ films

Electron-boson interaction is fundamental to a thorough understanding of various exotic properties emerging in many-body physics. In photoemission spectroscopy, photoelectron emission due to photon absorption would trigger diverse collective excitations in solids, including the emergence of phonons, magnons, electron-hole pairs, and plasmons, which naturally provides a reliable pathway to study electron-boson couplings. While fingerprints of electron-phonon/-magnon interactions in this state-of-the-art technique have been well investigated, much less is known about electron-plasmon coupling, and direct observation of the band renormalization solely due to electron-plasmon interactions is extremely challenging. Here by utilizing integrated oxide molecular-beam epitaxy and angle-resolved photoemission spectroscopy, we discover the long sought-after pure electron-plasmon coupling-induced low-lying plasmonic-polaron replica bands in epitaxial semimetallic SrIrO$_3$ films, in which the characteristic low carrier concentration and narrow bandwidth combine to provide a unique platform where the electron-plasmon interaction can be investigated kinematically in photoemission spectroscopy. This finding enriches the forms of electron band normalization on collective modes in solids and demonstrates that, to obtain a complete understanding of the quasiparticle dynamics in 5d electron systems, the electron-plasmon interaction should be considered on equal footing with the acknowledged electron-electron interaction and spin-orbit coupling.

cond-mat.str-el↗

arXiv2021

Gate-Tunable Optical Extinction of Graphene Nanoribbon Nanoclusters

We investigate the optical response of graphene nanoribbons (GNRs) using the broadband nonlinear generation and detection capabilities of nanoscale junctions created at the LaAlO$_3$/SrTiO$_3$ interface. GNR nanoclusters measured to be as small as 1-2 GNRs in size are deposited on the LaAlO$_3$ surface with an atomic force microscope tip. Time-resolved nonlinear optical probes of GNR nanoclusters reveal a strong, gate-tunable second and third harmonic response, as well as strong extinction of visible to near-infrared (VIS-NIR) light at distinct wavelengths, similar to previous reports with graphene.

cond-mat.mes-hall↗

arXiv2021

Insight-HXMT observations of jet-like corona in a black hole X-ray binary MAXI J1820+070

A black hole X-ray binary produces hard X-ray radiation from its corona and disk when the accreting matter heats up. During an outburst, the disk and corona co-evolves with each other. However, such an evolution is still unclear in both its geometry and dynamics. Here we report the unusual decrease of the reflection fraction in MAXI J1820+070, which is the ratio of the coronal intensity illuminating the disk to the coronal intensity reaching the observer, as the corona is observed to contrast during the decay phase. We postulate a jet-like corona model, in which the corona can be understood as a standing shock where the material flowing through. In this dynamical scenario, the decrease of the reflection fraction is a signature of the corona's bulk velocity. Our findings suggest that as the corona is observed to get closer to the black hole, the coronal material might be outflowing faster.

astro-ph.HE↗

arXiv2021

Hardware Acceleration of Fully Quantized BERT for Efficient Natural Language Processing

BERT is the most recent Transformer-based model that achieves state-of-the-art performance in various NLP tasks. In this paper, we investigate the hardware acceleration of BERT on FPGA for edge computing. To tackle the issue of huge computational complexity and memory footprint, we propose to fully quantize the BERT (FQ-BERT), including weights, activations, softmax, layer normalization, and all the intermediate results. Experiments demonstrate that the FQ-BERT can achieve 7.94x compression for weights with negligible performance loss. We then propose an accelerator tailored for the FQ-BERT and evaluate on Xilinx ZCU102 and ZCU111 FPGA. It can achieve a performance-per-watt of 3.18 fps/W, which is 28.91x and 12.72x over Intel(R) Core(TM) i7-8700 CPU and NVIDIA K80 GPU, respectively.

cs.AR↗

arXiv2021

Modified quantum delayed-choice experiment without quantum control

Wheeler's delayed-choice experiment delays the decision to observe either the wave or particle behavior of a photon until after it has entered the interferometer, and the quantum delayed-choice experiment provides the possibility of observing the wave and particle behavior simultaneously by introducing quantum control device. We here propose a modified quantum delayed-choice experiment without quantum control or entanglement assistance, in which a photon can be prepared in a wave-particle superposition state and the morphing behavior of wave-to-particle transition can be observed easily. It is demonstrated that the presented scheme can allow us to rule out classical hidden variable models in a device-independent manner via violating dimension witness. We also extend the scheme to the situation of two degrees of freedom, first constructing a hybrid quantum delayed-choice experiment which enables simultaneous observation of a photon's wave and particle behaviors in different degrees of freedom, and then proposing a scheme to prepare the single-photon wave-particle entanglement. This study is not only meaningful to explore the wave and particle properties of photons, but also provides potential for the research of the single-particle nonlocality from the perspective of the wave-particle degree of freedom.

quant-ph↗

arXiv2021

High magnetic field induced crossover from the Kondo to Fermi liquid behavior in 1$T$-VTe$_{2}$ single crystals

The magnetic and magnetotransport properties of metallic 1$T$-VTe$_{2}$ single crystals were investigated at temperatures from 1.3 to 300 K and in magnetic fields up to 35 T. Upon applying a high magnetic field, it is found that the electrical resistivity displays a crossover from the logarithmic divergence of the single-impurity Kondo effect to the Fermi liquid behavior at low temperatures. The Brillouin scale of the negative magnetoresistivity above the Kondo temperature $T_{\rm{K}}$ = 12 K indicates that the Kondo features originate from intercalated V ions, with $S$ = 1/2. Both magnetic susceptibility and Hall effect show an anomaly around $T_{\rm{K}}$. By using the modified Hamann expression we successfully describe the temperature-dependent resistivity under various magnetic fields, which shows the characteristic peak below $T_{\rm{K}}$ due to the splitting of the Kondo resonance.

cond-mat.str-el↗

arXiv2021

Pressure-induced Superconductivity in dual-topological semimetal Pt2HgSe3

Recently monolayer jacutingaite (Pt2HgSe3), a naturally occurring exfoliable mineral, discovered in Brazil in 2008, has been theoretically predicted as a candidate quantum spin Hall system with a 0.5 eV band gap, while the bulk form is one of only a few known dual-topological insulators which may host different surface states protected by symmetries. In this work, we systematically investigate both structure and electronic evolution of bulk Pt2HgSe3 under high pressure up to 96 GPa. The nontrivial topology persists up to the structural phase transition observed in the high-pressure regime. Interestingly, we found that this phase transition is accompanied by the appearance of superconductivity at around 55 GPa and the critical transition temperature Tc increases with applied pressure. Our results demonstrate that Pt2HgSe3 with nontrivial topology of electronic states displays new ground states upon compression and raises potentials in application to the next-generation spintronic devices.

cond-mat.supr-con↗

arXiv2021

Spacewalker: Rapid UI Design Exploration Using Lightweight Markup Enhancement and Crowd Genetic Programming

User interface design is a complex task that involves designers examining a wide range of options. We present Spacewalker, a tool that allows designers to rapidly search a large design space for an optimal web UI with integrated support. Designers first annotate each attribute they want to explore in a typical HTML page, using a simple markup extension we designed. Spacewalker then parses the annotated HTML specification, and intelligently generates and distributes various configurations of the web UI to crowd workers for evaluation. We enhanced a genetic algorithm to accommodate crowd worker responses from pairwise comparison of UI designs, which is crucial for obtaining reliable feedback. Based on our experiments, Spacewalker allows designers to effectively search a large design space of a UI, using the language they are familiar with, and improve their design rapidly at a minimal cost.

cs.HC↗

arXiv2021

Rotation of the convective core in $γ$ Dor stars measured by dips in period spacings of g modes coupled with inertial modes

The relation of period spacing ($ΔP$) versus period ($P$) of dipole prograde g modes is known to be useful to measure rotation rates in the g-mode cavity of rapidly rotating $γ$ Dor and slowly pulsating B (SPB) stars. In a rapidly rotating star, an inertial mode in the convective core can resonantly couple with g modes propagative in the surrounding radiative region. The resonant coupling causes a dip in the $P$-$ΔP$ relation, distinct from the modulations due to the chemical composition gradient. Such a resonance dip in $ΔP$ of prograde dipole g modes appears around a frequency corresponding to a spin parameter $2f_{\rm rot}{\rm(cc)}/ν_{\rm co-rot} \sim 8-11$ with $f_{\rm rot}$(cc) being the rotation frequency of the convective core and $ν_{\rm co-rot}$ the pulsation frequency in the co-rotating frame. The spin parameter at the resonance depends somewhat on the extent of core overshooting, central hydrogen abundance, and other stellar parameters. We can fit the period at the observed dip with the prediction from prograde dipole g modes of a main-sequence model, allowing the convective core to rotate differentially from the surrounding g-mode cavity. We have performed such fittings for 16 selected $γ$ Dor stars having well defined dips, and found that the majority of $γ$ Dor stars we studied rotate nearly uniformly, while convective cores tend to rotate slightly faster than the g-mode cavity in less evolved stars.

astro-ph.SR↗

arXiv2021

From high-$T_c$ to low-$T_c$: Multi-orbital effects in transition metal oxides

Despite the structural resemblance of certain cuprate and nickelate parent compounds there is a striking spread of $T_c$ among such transition metal oxide superconductors. We adopt a minimal two-orbital $e_g$ model which covers cuprates and nickelate heterostructures in different parametric limits, and analyse its superconducting instabilities. The joint consideration of interactions, doping, Fermiology, and in particular the $e_g$ orbital splitting allows us to explain the strongly differing pairing propensities in cuprate and nickelate superconductors.

cond-mat.supr-con↗

arXiv2021

A flow approach to the generalized Loewner-Nirenberg problem of the $σ_k$-Ricci equation

We introduce a flow approach to the generalized Loewner-Nirenberg problem $(1.5)-(1.7)$ of the $σ_k$-Ricci equation on a compact manifold $(M^n,g)$ with boundary. We prove that for initial data $u_0\in C^{4,α}(M)$ which is a subsolution to the $σ_k$-Ricci equation $(1.5)$, the Cauchy-Dirichlet problem $(3.1)-(3.3)$ has a unique solution $u$ which converges in $C^4_{loc}(M^{\circ})$ to the solution $u_{\infty}$ of the problem $(1.5)-(1.7)$, as $t\to\infty$.

math.AP↗

arXiv2021

Probe CP violation in $H\to γZ$ through forward-backward asymmetry

We suggest that the forward-backward asymmetry $(A_{FB})$ of the charged leptons in $gg\to H\toγZ\toγ\ell^-\ell^+$ process could be used to probe the CP violating $HγZ$ coupling when the interference from $gg\toγZ\toγ\ell^-\ell^+$ process is included. With CP violation in $HγZ$ coupling, the interference effect leads to a non-vanishing $A_{FB}$, which is also sensitive to the strong phase differences. The resonant and non-resonant strong phases together make $A_{FB}(\hat{s})$ change sign around Higgs mass $M_H$. For phenomenology study, we suggest the integral over one-side mass region below $M_H$ to magnify the $A_{FB}$ strength.

hep-ph↗

arXiv2020

CP-violating Higgs Di-tau Decays: Baryogenesis and Higgs Factories

We demonstrate how probes of CP-violating observables in Higgs di-tau decays at prospective future lepton colliders could provide a test of weak scale baryogenesis with significant discovery potential. Measurements at the Circular Electron Positron Collider, for example, could exclude a CP phase larger than $2.9^\circ$ ($5.6^\circ$) at 68% (95%) C.L. assuming the Standard Model value for magnitude of the tau lepton Yukawa coupling. Conversely, this sensitivity would allow for a $5\,σ$ discovery for 82% of the CP phase range $[0,2π)$. The reaches of the Future Circular Collider - ee and International Linear Collider are comparable. As a consequence, future lepton colliders could establish the presence of CP violation required by lepton flavored electroweak baryogenesis with at least $3\,σ$ sensitivity. Our results illustrate that Higgs factories are not just precision machines but can also make $\mathcal O(1)$ measurement of the new physics beyond the Standard Model.

hep-ph↗