Searcharxiv⌕ Search

arXiv subjects

Ke Zhang

Publications and source records attributed to Ke Zhang.

At least 343 records · Page 19Linked to original sources
arXiv2017

On Graphene in the Interstellar Medium

The possible detection of C_{24}, a planar graphene, recently reported in several planetary nebulae by Garciaa-Hernandez et al. (2011, 2012) inspires us to explore whether and how much graphene could exist in the interstellar medium (ISM) and how it would reveal its presence through its ultraviolet (UV) extinction and infrared (IR) emission. In principle, interstellar graphene could arise from the photochemical processing of polycyclic aromatic hydrocarbon (PAH) molecules which are abundant in the ISM through a complete loss of their hydrogen atoms and/or from graphite which is thought to be a major dust species in the ISM through fragmentation caused by grain-grain collisional shattering. Both quantum-chemical computations and laboratory experiments have shown that the exciton-dominated electronic transitions in graphene cause a strong absorption band near 2755 Angstrom. We calculate the UV absorption of graphene and place an upper limit of ~5 ppm of C/H (i.e., ~1.9\% of the total interstellar C) on the interstellar graphene abundance. We also model the stochastic heating of graphene C_{24} excited by single starlight photons of the interstellar radiation field in the ISM and calculate its IR emission spectra. We also derive the abundance of graphene in the ISM to be <5 ppm of C/H by comparing the model emission spectra with that observed in the ISM.

astro-ph.GA↗
arXiv2017

Age Group and Gender Estimation in the Wild with Deep RoR Architecture

Automatically predicting age group and gender from face images acquired in unconstrained conditions is an important and challenging task in many real-world applications. Nevertheless, the conventional methods with manually-designed features on in-the-wild benchmarks are unsatisfactory because of incompetency to tackle large variations in unconstrained images. This difficulty is alleviated to some degree through Convolutional Neural Networks (CNN) for its powerful feature representation. In this paper, we propose a new CNN based method for age group and gender estimation leveraging Residual Networks of Residual Networks (RoR), which exhibits better optimization ability for age group and gender classification than other CNN architectures.Moreover, two modest mechanisms based on observation of the characteristics of age group are presented to further improve the performance of age estimation.In order to further improve the performance and alleviate over-fitting problem, RoR model is pre-trained on ImageNet firstly, and then it is fune-tuned on the IMDB-WIKI-101 data set for further learning the features of face images, finally, it is used to fine-tune on Adience data set. Our experiments illustrate the effectiveness of RoR method for age and gender estimation in the wild, where it achieves better performance than other CNN methods. Finally, the RoR-152+IMDB-WIKI-101 with two mechanisms achieves new state-of-the-art results on Adience benchmark.

cs.CV↗
arXiv2017

Pyramidal RoR for Image Classification

The Residual Networks of Residual Networks (RoR) exhibits excellent performance in the image classification task, but sharply increasing the number of feature map channels makes the characteristic information transmission incoherent, which losses a certain of information related to classification prediction, limiting the classification performance. In this paper, a Pyramidal RoR network model is proposed by analysing the performance characteristics of RoR and combining with the PyramidNet. Firstly, based on RoR, the Pyramidal RoR network model with channels gradually increasing is designed. Secondly, we analysed the effect of different residual block structures on performance, and chosen the residual block structure which best favoured the classification performance. Finally, we add an important principle to further optimize Pyramidal RoR networks, drop-path is used to avoid over-fitting and save training time. In this paper, image classification experiments were performed on CIFAR-10/100 and SVHN datasets, and we achieved the current lowest classification error rates were 2.96%, 16.40% and 1.59%, respectively. Experiments show that the Pyramidal RoR network optimization method can improve the network performance for different data sets and effectively suppress the gradient disappearance problem in DCNN training.

cs.CV↗
arXiv2017

Thermodynamics of the Katok Map

We effect thermodynamical formalism for the non-uniformly hyperbolic $C^\infty$ map of the two dimensional torus known as the Katok map. It is a slowdown of a linear Anosov map near the origin and it is a local (but not small) perturbation. We prove the existence of equilibrium measures for any continuous potential function and obtain uniqueness of equilibrium measures associated to the geometric $t$-potential $φ_t=-t\log |df|_{E^u(x)}|$ for any $t\in(t_0,\infty)$, $t\neq 1$ where $E^u(x)$ denotes the unstable direction. We show that $t_0$ tends to $-\infty$ as the size of the perturbation tends to zero. Finally, we establish exponential decay of correlations as well as the Central Limit Theorem for the equilibrium measures associated to $φ_t$ for all values of $t\in (t_0, 1)$.

math.DS↗
arXiv2017

Mass inventory of the giant-planet formation zone in a solar nebula analog

The initial mass distribution in the solar nebula is a critical input to planet formation models that seek to reproduce today's Solar System. Traditionally, constraints on the gas mass distribution are derived from observations of the dust emission from disks, but this approach suffers from large uncertainties in grain growth and gas-to-dust ratio. On the other hand, previous observations of gas tracers only probe surface layers above the bulk mass reservoir. Here we present the first partially spatially resolved observations of the $^{13}$C$^{18}$O J=3-2 line emission in the closest protoplanetary disk, TW Hya, a gas tracer that probes the bulk mass distribution. Combining it with the C$^{18}$O J=3-2 emission and the previously detected HD J=1-0 flux, we directly constrain the mid-plane temperature and optical depths of gas and dust emission. We report a gas mass distribution of 13$^{+8}_{-5}\times$(R/20.5AU)$^{-0.9^{+0.4}_{-0.3}}$ g cm$^{-2}$ in the expected formation zone of gas and ice giants (5-21AU). We find the total gas/millimeter-sized dust mass ratio is 140 in this region, suggesting that at least 2.4M_earth of dust aggregates have grown to >centimeter sizes (and perhaps much larger). The radial distribution of gas mass is consistent with a self-similar viscous disk profile but much flatter than the posterior extrapolation of mass distribution in our own and extrasolar planetary systems.

astro-ph.EP↗
arXiv2017

Extensible 3D architecture for superconducting quantum computing

Using a multi-layered printed circuit board, we propose a 3D architecture suitable for packaging supercon- ducting chips, especially chips that contain two-dimensional qubit arrays. In our proposed architecture, the center strips of the buried coplanar waveguides protrude from the surface of a dielectric layer as contacts. Since the contacts extend beyond the surface of the dielectric layer, chips can simply be flip-chip packaged with on-chip receptacles clinging to the contacts. Using this scheme, we packaged a multi-qubit chip and per- formed single-qubit and two-qubit quantum gate operations. The results indicate that this 3D architecture provides a promising scheme for scalable quantum computing.

quant-ph↗
arXiv2017

Hyperbolicity of minimizers and regularity of viscosity solutions for random Hamilton-Jacobi equations

We show that for a family of randomly kicked Hamilton-Jacobi equations, the unique global minimizer is hyperbolic, almost surely. Furthermore, we prove the unique forward and backward viscosity solutions, though in general only Lipshitz, are smooth in a neighbourhood of the global minimizer. Our result generalizes the result of E, Khanin, Mazel and Sinai (\cite{EKMS00}) to dimension $d\ge 2$, and extends the result of Iturriaga and Khanin in \cite{IK03}.

math.DS↗
arXiv2017

Exponential convergence of solutions for random Hamilton-Jacobi equations

We show that for a family of randomly kicked Hamiton-Jacobi equations on the torus, almost surely, the solution of an initial value problem converges exponentially fast to the unique stationary solution. Combined with the results in \cite{IK03} and \cite{KZ12}, this completes the program started in \cite{EKMS00} for the multi-dimensional setting.

math.DS↗
arXiv2017

Residual Networks of Residual Networks: Multilevel Residual Networks

A residual-networks family with hundreds or even thousands of layers dominates major image recognition tasks, but building a network by simply stacking residual blocks inevitably limits its optimization ability. This paper proposes a novel residual-network architecture, Residual networks of Residual networks (RoR), to dig the optimization ability of residual networks. RoR substitutes optimizing residual mapping of residual mapping for optimizing original residual mapping. In particular, RoR adds level-wise shortcut connections upon original residual networks to promote the learning capability of residual networks. More importantly, RoR can be applied to various kinds of residual networks (ResNets, Pre-ResNets and WRN) and significantly boost their performance. Our experiments demonstrate the effectiveness and versatility of RoR, where it achieves the best performance in all residual-network-like structures. Our RoR-3-WRN58-4+SD models achieve new state-of-the-art results on CIFAR-10, CIFAR-100 and SVHN, with test errors 3.77%, 19.73% and 1.59%, respectively. RoR-3 models also achieve state-of-the-art results compared to ResNets on ImageNet data set.

cs.CV↗
arXiv2017

Discovery and Physical Characterization of a Large Scattered Disk Object at 92 AU

We report the observation and physical characterization of the possible dwarf planet \UZ\ ("DeeDee"), a dynamically detached trans-Neptunian object discovered at 92 AU. This object is currently the second-most distant known trans-Neptunian object with reported orbital elements, surpassed in distance only by the dwarf planet Eris. The object was discovered with an $r$-band magnitude of 23.0 in data collected by the Dark Energy Survey between 2014 and 2016. Its 1140-year orbit has $(a,e,i) = (109~\mathrm{AU}, 0.65, 26.8^{\circ})$. It will reach its perihelion distance of 38 AU in the year 2142. Integrations of its orbit show it to be dynamically stable on Gyr timescales, with only weak interactions with Neptune. We have performed followup observations with ALMA, using 3 hours of on-source integration time to measure the object's thermal emission in the Rayleigh-Jeans tail. The signal is detected at 7$σ$ significance, from which we determine a $V$-band albedo of $13.1^{+3.3}_{-2.4}\mathrm{(stat)}^{+2.0}_{-1.4}\mathrm{(sys)}$ percent and a diameter of $635^{+57}_{-61}\mathrm{(stat)}^{+32}_{-39}\mathrm{(sys)}$~km, assuming a spherical body with uniform surface properties.

astro-ph.EP↗
arXiv2016

Hydrocarbon emission rings in protoplanetary disks induced by dust evolution

We report observations of resolved C2H emission rings within the gas-rich protoplanetary disks of TW Hya and DM Tau using the Atacama Large Millimeter Array (ALMA). In each case the emission ring is found to arise at the edge of the observable disk of mm-sized grains (pebbles) traced by (sub)mm-wave continuum emission. In addition, we detect a C3H2 emission ring with an identical spatial distribution to C2H in the TW Hya disk. This suggests that these are hydrocarbon rings (i.e. not limited to C2H). Using a detailed thermo-chemical model we show that reproducing the emission from C2H requires a strong UV field and C/O > 1 in the upper disk atmosphere and outer disk, beyond the edge of the pebble disk. This naturally arises in a disk where the ice-coated dust mass is spatially stratified due to the combined effects of coagulation, gravitational settling and drift. This stratification causes the disk surface and outer disk to have a greater permeability to UV photons. Furthermore the concentration of ices that transport key volatile carriers of oxygen and carbon in the midplane, along with photochemical erosion of CO, leads to an elemental C/O ratio that exceeds unity in the UV-dominated disk. Thus the motions of the grains, and not the gas, lead to a rich hydrocarbon chemistry in disk surface layers and in the outer disk midplane.

astro-ph.EP↗
arXiv2016

Video Summarization with Long Short-term Memory

We propose a novel supervised learning technique for summarizing videos by automatically selecting keyframes or key subshots. Casting the problem as a structured prediction problem on sequential data, our main idea is to use Long Short-Term Memory (LSTM), a special type of recurrent neural networks to model the variable-range dependencies entailed in the task of video summarization. Our learning models attain the state-of-the-art results on two benchmark video datasets. Detailed analysis justifies the design of the models. In particular, we show that it is crucial to take into consideration the sequential structures in videos and model them. Besides advances in modeling techniques, we introduce techniques to address the need of a large number of annotated data for training complex learning models. There, our main idea is to exploit the existence of auxiliary annotated video datasets, albeit heterogeneous in visual styles and contents. Specifically, we show domain adaptation techniques can improve summarization by reducing the discrepancies in statistical properties across those datasets.

cs.CV↗
arXiv2016

Summary Transfer: Exemplar-based Subset Selection for Video Summarization

Video summarization has unprecedented importance to help us digest, browse, and search today's ever-growing video collections. We propose a novel subset selection technique that leverages supervision in the form of human-created summaries to perform automatic keyframe-based video summarization. The main idea is to nonparametrically transfer summary structures from annotated videos to unseen test videos. We show how to extend our method to exploit semantic side information about the video's category/genre to guide the transfer process by those training videos semantically consistent with the test input. We also show how to generalize our method to subshot-based summarization, which not only reduces computational costs but also provides more flexible ways of defining visual similarity across subshots spanning several frames. We conduct extensive evaluation on several benchmarks and demonstrate promising results, outperforming existing methods in several settings.

cs.CV↗
arXiv2016

The Radial Distribution of H2 and CO in TW Hya as Revealed by Resolved ALMA Observations of CO Isotopologues

CO is widely used as a tracer of molecular gas. However, there is now mounting evidence that gas phase carbon is depleted in the disk around TW Hya. Previous efforts to quantify this depletion have been hampered by uncertainties regarding the radial thermal structure in the disk. Here we present resolved ALMA observations of 13CO 3-2, C18O 3-2, 13CO 6-5, and C18O 6-5 emission in TW Hya, which allow us to derive radial gas temperature and gas surface density profiles, as well as map the CO abundance as a function of radius. These observations provide a measurement of the surface CO snowline at ~30 AU and show evidence for an outer ring of CO emission centered at 53 AU, a feature previously seen only in less abundant species. Further, the derived CO gas temperature profile constrains the freeze-out temperature of CO in the warm molecular layer to < 21 K. Combined with the previous detection of HD 1-0, these data constrain the surface density of the warm H2 gas in the inner ~30 AU. We find that CO is depleted by two orders of magnitude from R=10-60 AU, with the small amount of CO returning to the gas phase inside the surface CO snowline insufficient to explain the overall depletion. Finally, this new data is used in conjunction with previous modeling of the TW Hya disk to constrain the midplane CO snowline to 17-23 AU.

astro-ph.SR↗
arXiv2016

On the Commonality of 10-30AU Sized Axisymmetric Dust Structures in Protoplanetary Disks

An unsolved problem in step-wise core-accretion planet formation is that rapid radial drift in gas-rich protoplanetary disks should drive mm-/meter-sized particles inward to the central star before large bodies can form. One promising solution is to confine solids within small scale structures. Here we investigate dust structures in the (sub)mm continuum emission of four disks (TW Hya, HL Tau, HD 163296 and DM Tau), a sample of disks with the highest spatial resolution ALMA observations to date. We retrieve the surface brightness distributions using synthesized images and fitting visibilities with analytical functions. We find that the continuum emission of the four disks is ~axi-symmetric but rich in 10-30AU-sized radial structures, possibly due to physical gaps, surface density enhancements or localized dust opacity variations within the disks. These results suggest that small scale axi-symmetric dust structures are likely to be common, as a result of ubiquitous processes in disk evolution and planet formation. Compared with recent spatially resolved observations of CO snowlines in these same disks, all four systems show enhanced continuum emission from regions just beyond the CO condensation fronts, potentially suggesting a causal relationship between dust growth/trapping and snowlines.

astro-ph.EP↗
arXiv2015

Measurements of water surface snow lines in classical protoplanetary disks

We present deep Herschel-PACS spectroscopy of far-infrared water lines from a sample of four protoplanetary disks around solar-mass stars, selected to have strong water emission at mid-infrared wavelengths. By combining the new Herschel spectra with archival Spitzer-IRS spectroscopy, we retrieve a parameterized radial surface water vapor distribution from 0.1-100 AU using two-dimensional dust and line radiative transfer modeling. The surface water distribution is modeled with a step model comprising of a constant inner and outer relative water abundance and a critical radius at which the surface water abundance is allowed to change. We find that the four disks have critical radii of $\sim 3-11$ AU, at which the surface water abundance decreases by at least 5 orders of magnitude. The measured values for the critical radius are consistently smaller than the location of the surface snow line, as predicted by the observed spectral energy distribution. This suggests that the sharp drop-off of the surface water abundance is not solely due to the local gas-solid balance, but may also be driven by the de-activation of gas-phase chemical pathways to water below 300 K. Assuming a canonical gas-to-dust ratio of 100, as well as coupled gas and dust temperatures $T_{\rm gas}=T_{\rm dust}$, the best-fit inner water abundances become implausibly high (0.01-1.0 ${\rm H_{2}}^{-1}$). Conversely, a model in which the gas and dust temperatures are decoupled leads to canonical inner disk water abundances of $\sim 10^{-4} \rm H_{2}^{-1}$, while retaining gas-to-dust ratios of 100. That is, the evidence for gas-dust decoupling in disk surfaces is stronger than for enhanced gas-to-dust ratios.

astro-ph.SR↗