arXiv subjects

Cheng Li

Publications and source records attributed to Cheng Li.

At least 415 records · Page 23Linked to original sources

arXiv2019

Automated Droplet Size Distribution Measurements Using Digital Inline Holography

Droplet generation through spray breakup is an unsteady and non-linear process which produces a relatively dense, highly polydisperse aerosol containing non-spherical droplets with sizes spanning several orders of magnitude. Such variability in size and shape can lead to significant sources of error for conventional measurements based on laser scattering. Although direct imaging of droplets can potentially overcome these limitations, imaging suffers from a shallow depth of field as well as occlusions, which prevents the complete spray from being analyzed. In comparison, digital inline holography (DIH), a low-cost coherent imaging technique, can enable high-resolution imaging of the sample over an extended depth of field, typically several orders of magnitude larger than traditional imaging. In this study, we showcase an automated DIH imaging system for characterizing monodisperse and polydisperse aerosol droplet size and shape distributions in the 20 um-3 mm diameter range, over a large sample volume. The high accuracy of the technique is demonstrated by measurements of monodisperse droplets generated by a vibrating orifice droplet generator, achieving a resolution of ~14.2. Measurements of a polydisperse spray from a flat fan nozzle serve to establish the versatility of DIH in extracting a two-dimensional size-eccentricity distribution function, which indicates a strong semilogarithmic scaling between the two parameters that decays as the droplet migrates away from the nozzle. Due to its low cost and compact setup as well as the high density of data obtained, DIH can serve as a promising approach for future aerosol characterization.

physics.app-ph↗

arXiv2019

Plasmonics in Atomically-Thin Crystalline Silver Films

Light-matter interaction at the atomic scale rules fundamental phenomena such as photoemission and lasing, while enabling basic everyday technologies, including photovoltaics and optical communications. In this context, plasmons --the collective electron oscillations in conducting materials-- are important because they allow manipulating optical fields at the nanoscale. The advent of graphene and other two-dimensional crystals has pushed plasmons down to genuinely atomic dimensions, displaying appealing properties such as a large electrical tunability. However, plasmons in these materials are either too broad or lying at low frequencies, well below the technologically relevant near-infrared regime. Here we demonstrate sharp near-infrared plasmons in lithographically-patterned wafer-scale atomically-thin silver crystalline films. Our measured optical spectra reveal narrow plasmons (quality factor $\sim4$), further supported by a low sheet resistance comparable to bulk metal in few-atomic-layer silver films down to seven Ag(111) monolayers. Good crystal quality and plasmon narrowness are obtained despite the addition of a thin passivating dielectric, which renders our samples resilient to ambient conditions. The observation of spectrally sharp and strongly confined plasmons in atomically thin silver holds great potential for electro-optical modulation and optical sensing applications.

cond-mat.mes-hall↗

arXiv2019

A Divide-and-Conquer Bayesian Approach to Large-Scale Kriging

We propose a three-step divide-and-conquer strategy within the Bayesian paradigm that delivers massive scalability for any spatial process model. We partition the data into a large number of subsets, apply a readily available Bayesian spatial process model on every subset, in parallel, and optimally combine the posterior distributions estimated across all the subsets into a pseudo-posterior distribution that conditions on the entire data. The combined pseudo posterior distribution replaces the full data posterior distribution for predicting the responses at arbitrary locations and for inference on the model parameters and spatial surface. Based on distributed Bayesian inference, our approach is called "Distributed Kriging" (DISK) and offers significant advantages in massive data applications where the full data are stored across multiple machines. We show theoretically that the Bayes $L_2$-risk of the DISK posterior distribution achieves the near optimal convergence rate in estimating the true spatial surface with various types of covariance functions, and provide upper bounds for the number of subsets as a function of the full sample size. The model-free feature of DISK is demonstrated by scaling posterior computations in spatial process models with a stationary full-rank and a nonstationary low-rank Gaussian process (GP) prior. A variety of simulations and a geostatistical analysis of the Pacific Ocean sea surface temperature data validate our theoretical results.

stat.ME↗

arXiv2019

Retrieval of Chemical Abundances in Titan's Upper Atmosphere from Cassini UVIS Observations with Pointing Motion

Cassini/UVIS FUV observations of stellar occultations at Titan are well suited for probing its atmospheric composition and structure. However, due to instrument pointing motion, only five out of tens of observations have been analyzed. We present an innovative retrieval method that corrects for the effect of pointing motion by forward modeling the Cassini/UVIS instrument response function with the pointing motion value obtained from the SPICE C-kernel along the spectral dimension. To illustrate the methodology, an occultation observation made during flyby T52 is analyzed, when the Cassini spacecraft had insufficient attitude control. A high-resolution stellar model and an instrument response simulator that includes the position of the point source on the detector are used for the analysis of the pointing motion. The Markov Chain Monte-Carlo method is used to retrieve the line-of-sight abundance profiles of eleven species (CH4, C2H2, C2H4, C2H6, C4H2, C6H6, HCN, C2N2, HC3N, C6N2 and haze particles) in the spectral vector fitting process. We obtain tight constraints on all of the species aside from C2H6, C2N2 and C6N2, for which we only retrieved upper limits. This is the first time that the T52 occultation was used to derive abundances of major hydrocarbon and nitrile species in Titan's upper and middle atmosphere, as pointing motion prohibited prior analysis. With this new method, nearly all of the occultations obtained over the entire Cassini mission could yield reliable profiles of atmospheric composition, allowing exploration of Titan's upper atmosphere over seasons, latitudes, and longitudes.

astro-ph.EP↗

arXiv2019

TF-Ranking: Scalable TensorFlow Library for Learning-to-Rank

Learning-to-Rank deals with maximizing the utility of a list of examples presented to the user, with items of higher relevance being prioritized. It has several practical applications such as large-scale search, recommender systems, document summarization and question answering. While there is widespread support for classification and regression based learning, support for learning-to-rank in deep learning has been limited. We propose TensorFlow Ranking, the first open source library for solving large-scale ranking problems in a deep learning framework. It is highly configurable and provides easy-to-use APIs to support different scoring mechanisms, loss functions and evaluation metrics in the learning-to-rank setting. Our library is developed on top of TensorFlow and can thus fully leverage the advantages of this platform. For example, it is highly scalable, both in training and in inference, and can be used to learn ranking models over massive amounts of user activity data, which can include heterogeneous dense and sparse features. We empirically demonstrate the effectiveness of our library in learning ranking functions for large-scale search and recommendation applications in Gmail and Google Drive. We also show that ranking models built using our model scale well for distributed training, without significant impact on metrics. The proposed library is available to the open source community, with the hope that it facilitates further academic research and industrial applications in the field of learning-to-rank.

cs.IR↗

arXiv2019

Disentangling Content and Style via Unsupervised Geometry Distillation

It is challenging to disentangle an object into two orthogonal spaces of content and style since each can influence the visual observation differently and unpredictably. It is rare for one to have access to a large number of data to help separate the influences. In this paper, we present a novel framework to learn this disentangled representation in a completely unsupervised manner. We address this problem in a two-branch Autoencoder framework. For the structural content branch, we project the latent factor into a soft structured point tensor and constrain it with losses derived from prior knowledge. This constraint encourages the branch to distill geometry information. Another branch learns the complementary style information. The two branches form an effective framework that can disentangle object's content-style representation without any human annotation. We evaluate our approach on four image datasets, on which we demonstrate the superior disentanglement and visual analogy quality both in synthesized and real-world data. We are able to generate photo-realistic images with 256*256 resolution that are clearly disentangled in content and style.

cs.CV↗

arXiv2019

Universal Features of Landau Fans of Twisted Bilayer Graphene with Large Superlattices

Current experiments on different samples of twisted bilayer graphene (TBG) have found different sets of insulating phases. Despite this diversity, many features of these insulating phases appear to be universal. They include the dispersion of Landau fans away from charge neutrality, a reduced Landau fan degeneracy from the expected value at charge neutrality, and the further reduction of this degeneracy when crossing an insulating phase with odd number of electrons in the superlattice unit cell. We point out that all these behaviors as well as the ferromagnetic behavior observed in some of the insulating states suggest an underlying "ideal" pattern, with different part of it realized in the different samples in different experiments. We further show that such pattern can be accounted for by a Hubbard like model for the superlattice augmented with a set of chemical potential dependent mean fields that break the symmetry of the eight internal degrees of freedoms successively. The simultaneous importance of Mott like physics and mean field physics may be a general feature of twisted 2D electronic materials with large superlattices, not necessarily confined to graphene.

cond-mat.str-el↗

arXiv2019

JINGLE, a JCMT legacy survey of dust and gas for galaxy evolution studies: II. SCUBA-2 850 μm data reduction and dust flux density catalogues

We present the SCUBA-2 850 $μm$ component of JINGLE, the new JCMT large survey for dust and gas in nearby galaxies, which with 193 galaxies is the largest targeted survey of nearby galaxies at 850 $μm$. We provide details of our SCUBA-2 data reduction pipeline, optimised for slightly extended sources, and including a calibration model adjusted to match conventions used in other far-infrared data. We measure total integrated fluxes for the entire JINGLE sample in 10 infrared/submillimetre bands, including all WISE, Herschel-PACS, Herschel-SPIRE and SCUBA-2 850 $μm$ maps, statistically accounting for the contamination by CO(J=3-2) in the 850 $μm$ band. Of our initial sample of 193 galaxies, 191 are detected at 250 $μm$ with a $\geq$ 5$σ$ significance. In the SCUBA-2 850 $μm$ band we detect 126 galaxies with $\geq$ 3$σ$ significance. The distribution of the JINGLE galaxies in far-infrared/sub-millimetre colour-colour plots reveals that the sample is not well fit by single modified-blackbody models that assume a single dust-emissivity index $(β)$. Instead, our new 850 $μm$ data suggest either that a large fraction of our objects require $β< 1.5$, or that a model allowing for an excess of sub-mm emission (e.g., a broken dust emissivity law, or a very cold dust component 10 K) is required. We provide relations to convert far-infrared colours to dust temperature and $β$ for JINGLE-like galaxies. For JINGLE the FIR colours correlate more strongly with star-formation rate surface-density rather than the stellar surface-density, suggesting heating of dust is greater due to younger rather than older stellar-populations, consistent with the low proportion of early-type galaxies in the sample.

astro-ph.GA↗

arXiv2019

TransGaGa: Geometry-Aware Unsupervised Image-to-Image Translation

Unsupervised image-to-image translation aims at learning a mapping between two visual domains. However, learning a translation across large geometry variations always ends up with failure. In this work, we present a novel disentangle-and-translate framework to tackle the complex objects image-to-image translation task. Instead of learning the mapping on the image space directly, we disentangle image space into a Cartesian product of the appearance and the geometry latent spaces. Specifically, we first introduce a geometry prior loss and a conditional VAE loss to encourage the network to learn independent but complementary representations. The translation is then built on appearance and geometry space separately. Extensive experiments demonstrate the superior performance of our method to other state-of-the-art approaches, especially in the challenging near-rigid and non-rigid objects translation tasks. In addition, by taking different exemplars as the appearance references, our method also supports multimodal translation. Project page: https://wywu.github.io/projects/TGaGa/TGaGa.html

cs.CV↗

arXiv2019

Adapting RNN Sequence Prediction Model to Multi-label Set Prediction

We present an adaptation of RNN sequence models to the problem of multi-label classification for text, where the target is a set of labels, not a sequence. Previous such RNN models define probabilities for sequences but not for sets; attempts to obtain a set probability are after-thoughts of the network design, including pre-specifying the label order, or relating the sequence probability to the set probability in ad hoc ways. Our formulation is derived from a principled notion of set probability, as the sum of probabilities of corresponding permutation sequences for the set. We provide a new training objective that maximizes this set probability, and a new prediction objective that finds the most probable set on a test document. These new objectives are theoretically appealing because they give the RNN model freedom to discover the best label order, which often is the natural one (but different among documents). We develop efficient procedures to tackle the computation difficulties involved in training and prediction. Experiments on benchmark datasets demonstrate that we outperform state-of-the-art methods for this task.

cs.CL↗

arXiv2019

The clustering of galaxies with pseudo bulge and classical bulge in the local Universe

We investigate the clustering properties and close neighbour counts for galaxies with different types of bulges and stellar masses. We select samples of "classical" and "pseudo" bulges, as well as "bulge-less" disk galaxies, based on the bulge/disk decomposition catalog of SDSS galaxies provided by Simard et al. (2011). For a given galaxy sample we estimate: the projected two-point cross-correlation function with respect to a spectroscopic reference sample, w_p(r_p), and the average background-subtracted neighbour count within a projected separation using a photometric reference sample, N_neighbour(<r_p). We compare the results with the measurements of control samples matched in color, concentration and redshift. We find that, when limited to a certain stellar mass range and matched in color and concentration, all the samples present similar clustering amplitudes and neighbour counts on scales above ~0.1h^{-1}Mpc. This indicates that neither the presence of a central bulge, nor the bulge type is related to intermediate-to-large scale environments. On smaller scales, in contrast, pseudo-bulge and pure-disk galaxies similarly show strong excess in close neighbour count when compared to control galaxies, at all masses probed. For classical bulges, small-scale excess is also observed but only for M_stars < 10^{10} M_sun; at higher masses, their neighbour counts are similar to that of control galaxies at all scales. These results imply strong connections between galactic bulges and galaxy-galaxy interactions in the local Universe, although it is unclear how they are physically linked in the current theory of galaxy formation.

astro-ph.GA↗

arXiv2019

SDSS-IV MaNGA: Stellar initial mass function variation inferred from Bayesian analysis of the integral field spectroscopy of early type galaxies

We analyze the stellar initial mass functions (IMF) of a large sample of early type galaxies (ETGs) provided by MaNGA. The large number of IFU spectra of individual galaxies provide high signal-to-noise composite spectra that are essential for constraining IMF and to investigate possible radial gradients of the IMF within individual galaxies. The large sample of ETGs also make it possible to study how the IMF shape depends on various properties of galaxies. We adopt a novel approach to study IMF variations in ETGs, use Bayesian inferences based on full spectrum fitting. The Bayesian method provides a statistically rigorous way to explore potential degeneracy in spectrum fitting, and to distinguish different IMF models with Bayesian evidence. We find that the IMF slope depends systematically on galaxy velocity dispersion, in that galaxies of higher velocity dispersion prefer a more bottom-heavy IMF, but the dependence is almost entirely due to the change of metallicity, $Z$, with velocity dispersion. The IMF shape also depends on stellar age, $A$, but the dependence is completely degenerate with that on metallicity through a combination $AZ^{-1.42}$. Using independent age and metallicity estimates we find that the IMF variation is produced by metallicity instead of age. The IMF near the centers of massive ETGs appears more bottom-heavy than that in the outer parts, while a weak opposite trend is seen for low-mass ETGs. Uncertainties produced by star formation history, dust extinction, $α$-element abundance enhancement and noise in the spectra are tested.

astro-ph.GA↗

arXiv2019

SDSS-IV MaNGA: Environmental dependence of the Mgb/ -sigma_* relation for nearby galaxies

We use a sample of ~3000 galaxies from the MaNGA MPL-7 internal data release to study the alpha abundance distribution within low-redshift galaxies. We use the Lick index ratio Mgb/ as an alpha abundance indicator to study relationships between the alpha abundance distribution and galaxy properties such as effective stellar velocity dispersion within 0.3 effective radii (sigma_*), galaxy environment, and dark matter halo formation time (z_f). We find that (1) all galaxies show a tight correlation between Mgb/ and sigma_*; (2) `old' (H_beta < 3) low-sigma_* galaxies in high local density environment and inner regions within galaxy groups are enhanced in Mgb/ , while `young' (H_beta>3) galaxies and high-mass galaxies show no or less environmental dependence; (3) `old' galaxies with high-z_f show enhanced Mgb/ over low- and medium-z_f; (4) Mgb/ gradients are close to zero and show dependence on sigma_* but no obvious dependence on the environment or z_f. Our study indicates that stellar velocity dispersion or galaxy mass is the main parameter driving the Mgb/ enhancement, although environments appear to have modest effects, particularly for low- and medium-mass galaxies.

astro-ph.GA↗

arXiv2019

Delay Outage Probability of Multi-relay Selection for Mobile Relay Edge Computing Systems

In this paper, we deal with the problem of relay selection in mobile edge computing networks, where a source node transmits a computation-intensive task to a destination via the aid of multiple relay nodes. It differs from the traditional relay nodes in the way that each relay node is equipped with an edge computing server, thus each relay node can execute the received task and forwards the computed result to the destination. Accordingly, we define a delay outage probability to evaluate the impact of the relay computing ability on the link communications, and then propose a latency-best relay selection (LBRS) scheme that not only consider the communication capability, but also consider the computing ability. The performance of the proposed relay selection scheme with the traditional communication-only relay selection (CORS) and computing-only selection (CPORS) schemes in terms of the delay outage probability and the diversity order is analyzed, compared with other relay selection schemes. We show that the proposed LBRS scheme reduces to the traditional CORS scheme under the high signal-to-noise ratio (SNR) region. We further reveal that the diversity orders of both the proposed LBRS and CORS schemes are dependent on the computing ability of relay nodes.

cs.IT↗

arXiv2019

Advantages of the multinucleon transfer reactions based on 238U target for producing neutron-rich isotopes around N = 126

The mechanism of multinucleon transfer (MNT) reactions for producing neutron-rich heavy nuclei around N = 126 is investigated within two different theoretical frameworks: dinuclear system (DNS) model and isospin-dependent quantum molecular dynamics (IQMD) model. The effects of mass asymmetry relaxation, N=Z equilibration, and shell closures on production cross sections of neutron-rich heavy nuclei are investigated. For the first time, the advantages for producing neutron-rich heavy nuclei around N = 126 is found in MNT reactions based on 238U target. We propose the reactions with 238U target for producing unknown neutron-rich heavy nuclei around N = 126 in the future.

nucl-th↗

arXiv2019

Linking bar- and interaction-driven molecular gas concentration with centrally-enhanced star formation in EDGE-CALIFA galaxies

We study the spatially resolved star formation history and molecular gas distribution of 58 nearby galaxies, using integral field spectroscopy from the CALIFA survey and CO $J=1\rightarrow 0$ intensity mapping from the CARMA EDGE survey. We use the 4000 Å break (D$_n$4000), the equivalent width of the H$δ$ absorption line (EW H$δ_A$), and the equivalent width of the H$α$ emission line (EW H$α$) to measure the recent star formation history (SFH) of these galaxies. We measure radial profiles of the three SFH indicators and molecular gas mass surface density, from which we measure the level of centrally enhanced star formation and the molecular gas concentration. When we separate our galaxies into categories of barred (17 galaxies), unbarred (24 galaxies), and merging/paired (17 galaxies) we find that the galaxies which have centrally-enhanced star formation (19/58) are either barred (13/19) or in mergers/pairs (6/19) with relatively high molecular gas concentrations. A comparison between our barred galaxies and a snapshot of a hydrodynamic $N$-body simulation of a barred galaxy shows that the current theory of bar formation and evolution can qualitatively reproduce the main features of the observed galaxies in our sample, including both the sharp decrease of stellar age in the galactic center and the gradual decrease of age with increasing distance from center. These findings provide substantial evidence for a picture in which cold gas is transported inward by a bar or tidal interaction, which leads to the growth and rejuvenation of star formation in the central region.

astro-ph.GA↗

arXiv2019

Simulating Non-hydrostatic atmospheres on Planets (SNAP): formulation, validation and application to the Jovian atmosphere

A new non-hydrostatic and cloud-resolving atmospheric model is developed for studying moist convection and cloud formation in planetary atmospheres. It is built on top of the Athena++ framework, utilizing its static/adaptive mesh-refinement, parallelization, curvilinear geometry, and dynamic task scheduling. We extend the original hydrodynamic solver to vapors, clouds, and precipitation. Microphysics is formulated generically so that it can be applied to both Earth and Jovian planets. We implemented the Low Mach number Approximate Riemann Solver (LMARS) for simulating low speed atmospheric flows in addition to the usual Roe and HLLC Riemann solvers. Coupled with a fifth-order Weighted Essentially Nonoscillatory (WENO) subgrid-reconstruction method, the sharpness of critical fields such as clouds is well-preserved, and no extra hyperviscosity or spatial filter is needed to stabilize the model. Unlike many atmospheric models, total energy is used as the prognostic variable of the thermodynamic equation. One significant advantage of using total energy as a prognostic variable is that the entropy production due to irreversible mixing process can be properly captured. The model is designed to provide a unified framework for exploring planetary atmospheres across various conditions, both terrestrial and Jovian. First, a series of standard numerical tests for Earth's atmosphere is carried out to demonstrate the performance and robustness of the new model. Second, simulation of an idealized Jovian atmosphere in radiative-convective equilibrium shows that 1) the temperature gradient is superadiabatic near the water condensation level because of the changing of the mean molecular weight, and 2) the mean profile of ammonia gas shows a depletion in the subcloud layer down to nearly 10 bars. Relevance to the recent Juno observations is discussed.

astro-ph.EP↗

arXiv2018

Anti-drift in electronic nose via dimensionality reduction: a discriminative subspace projection approach

Sensor drift is a well-known issue in the field of sensors and measurement and has plagued the sensor community for many years. In this paper, we propose a sensor drift correction method to deal with the sensor drift problem. Specifically, we propose a discriminative subspace projection approach for sensor drift reduction in electronic noses. The proposed method inherits the merits of the subspace projection method called domain regularized component analysis. Moreover, the proposed method takes the source data label information into consideration, which minimizes the within-class variance of the projected source samples and at the same time maximizes the between-class variance. The label information is exploited to avoid overlapping of samples with different labels in the subspace. Experiments on two sensor drift datasets have shown the effectiveness of the proposed approach.

cs.LG↗