Searcharxiv⌕ Search

arXiv subjects

Xi Zhang

Publications and source records attributed to Xi Zhang.

At least 235 records · Page 13Linked to original sources
arXiv2023

Harmonic metrics and semi-simpleness

Given a flat vector bundle over a compact Riemannian manifold, Corlette and Donaldson proved that it admits harmonic metrics if and only if it is semi-simple. In this paper, we extend this equivalence to arbitrary vector bundles without any additional hypothesis, the result can be viewed as a Riemannian Hitchin-Kobayashi correspondence. Furthermore, we also prove an equivalence of categories in Sasakian geometry, relating the category of projective flat complex vector bundles to the category of Higgs bundles.

math.DG↗
arXiv2023

Photochemically-produced SO$_2$ in the atmosphere of WASP-39b

Photochemistry is a fundamental process of planetary atmospheres that regulates the atmospheric composition and stability. However, no unambiguous photochemical products have been detected in exoplanet atmospheres to date. Recent observations from the JWST Transiting Exoplanet Early Release Science Program found a spectral absorption feature at 4.05 $μ$m arising from SO$_2$ in the atmosphere of WASP-39b. WASP-39b is a 1.27-Jupiter-radii, Saturn-mass (0.28 M$_J$) gas giant exoplanet orbiting a Sun-like star with an equilibrium temperature of $\sim$1100 K. The most plausible way of generating SO$_2$ in such an atmosphere is through photochemical processes. Here we show that the SO$_2$ distribution computed by a suite of photochemical models robustly explains the 4.05 $μ$m spectral feature identified by JWST transmission observations with NIRSpec PRISM (2.7$σ$) and G395H (4.5$σ$). SO$_2$ is produced by successive oxidation of sulphur radicals freed when hydrogen sulphide (H$_2$S) is destroyed. The sensitivity of the SO$_2$ feature to the enrichment of the atmosphere by heavy elements (metallicity) suggests that it can be used as a tracer of atmospheric properties, with WASP-39b exhibiting an inferred metallicity of $\sim$10$\times$ solar. We further point out that SO$_2$ also shows observable features at ultraviolet and thermal infrared wavelengths not available from the existing observations.

astro-ph.EP↗
arXiv2023

Material Properties of Organic Liquids, Ices, and Hazes on Titan

Titan has a diverse range of materials in its atmosphere and on its surface: the simple organics that reside in various phases (gas, liquid, ice) and the solid complex refractory organics that form Titan's haze layers. These materials all actively participate in various physical processes on Titan, and many material properties are found to be important in shaping these processes. Future in-situ exploration on Titan would likely encounter a range of materials, and a comprehensive database to archive the material properties of all possible material candidates will be needed. Here we summarize several important material properties of the organic liquids, ices, and the refractory hazes on Titan that are available in the literature and/or that we have computed. These properties include thermodynamic properties (phase change points, sublimation and vaporization saturation vapor pressure, and latent heat), physical property (density), and surface properties (liquid surface tensions and solid surface energies). We have developed a new database to provide a repository for these data and make them available to the science community. These data can be used as inputs for various theoretical models to interpret current and future remote sensing and in-situ atmospheric and surface measurements on Titan. The material properties of the simple organics may also be applicable to giant planets and icy bodies in the outer solar system, interstellar medium, protoplanetary disks, and exoplanets.

astro-ph.EP↗
arXiv2023

HR-NeuS: Recovering High-Frequency Surface Geometry via Neural Implicit Surfaces

Recent advances in neural implicit surfaces for multi-view 3D reconstruction primarily focus on improving large-scale surface reconstruction accuracy, but often produce over-smoothed geometries that lack fine surface details. To address this, we present High-Resolution NeuS (HR-NeuS), a novel neural implicit surface reconstruction method that recovers high-frequency surface geometry while maintaining large-scale reconstruction accuracy. We achieve this by utilizing (i) multi-resolution hash grid encoding rather than positional encoding at high frequencies, which boosts our model's expressiveness of local geometry details; (ii) a coarse-to-fine algorithmic framework that selectively applies surface regularization to coarse geometry without smoothing away fine details; (iii) a coarse-to-fine grid annealing strategy to train the network. We demonstrate through experiments on DTU and BlendedMVS datasets that our approach produces 3D geometries that are qualitatively more detailed and quantitatively of similar accuracy compared to previous approaches.

cs.CV↗
arXiv2023

Dual-layer Image Compression via Adaptive Downsampling and Spatially Varying Upconversion

Ultra high resolution (UHR) images are almost always downsampled to fit small displays of mobile end devices and upsampled to its original resolution when exhibited on very high-resolution displays. This observation motivates us on jointly optimizing operation pairs of downsampling and upsampling that are spatially adaptive to image contents for maximal rate-distortion performance. In this paper, we propose an adaptive downsampled dual-layer (ADDL) image compression system. In the ADDL compression system, an image is reduced in resolution by learned content-adaptive downsampling kernels and compressed to form a coded base layer. For decompression the base layer is decoded and upconverted to the original resolution using a deep upsampling neural network, aided by the prior knowledge of the learned adaptive downsampling kernels. We restrict the downsampling kernels to the form of Gabor filters in order to reduce the complexity of filter optimization and also reduce the amount of side information needed by the decoder for adaptive upsampling. Extensive experiments demonstrate that the proposed ADDL compression approach of jointly optimized, spatially adaptive downsampling and upconversion outperforms the state of the art image compression methods.

eess.IV↗
arXiv2023

A broadband thermal emission spectrum of the ultra-hot Jupiter WASP-18b

Close-in giant exoplanets with temperatures greater than 2,000 K (''ultra-hot Jupiters'') have been the subject of extensive efforts to determine their atmospheric properties using thermal emission measurements from the Hubble and Spitzer Space Telescopes. However, previous studies have yielded inconsistent results because the small sizes of the spectral features and the limited information content of the data resulted in high sensitivity to the varying assumptions made in the treatment of instrument systematics and the atmospheric retrieval analysis. Here we present a dayside thermal emission spectrum of the ultra-hot Jupiter WASP-18b obtained with the NIRISS instrument on JWST. The data span 0.85 to 2.85 $μ$m in wavelength at an average resolving power of 400 and exhibit minimal systematics. The spectrum shows three water emission features (at $>$6$σ$ confidence) and evidence for optical opacity, possibly due to H$^-$, TiO, and VO (combined significance of 3.8$σ$). Models that fit the data require a thermal inversion, molecular dissociation as predicted by chemical equilibrium, a solar heavy element abundance (''metallicity'', M/H = 1.03$_{-0.51}^{+1.11}$ $\times$ solar), and a carbon-to-oxygen (C/O) ratio less than unity. The data also yield a dayside brightness temperature map, which shows a peak in temperature near the sub-stellar point that decreases steeply and symmetrically with longitude toward the terminators.

astro-ph.EP↗
arXiv2022

SemanticBEVFusion: Rethink LiDAR-Camera Fusion in Unified Bird's-Eye View Representation for 3D Object Detection

LiDAR and camera are two essential sensors for 3D object detection in autonomous driving. LiDAR provides accurate and reliable 3D geometry information while the camera provides rich texture with color. Despite the increasing popularity of fusing these two complementary sensors, the challenge remains in how to effectively fuse 3D LiDAR point cloud with 2D camera images. Recent methods focus on point-level fusion which paints the LiDAR point cloud with camera features in the perspective view or bird's-eye view (BEV)-level fusion which unifies multi-modality features in the BEV representation. In this paper, we rethink these previous fusion strategies and analyze their information loss and influences on geometric and semantic features. We present SemanticBEVFusion to deeply fuse camera features with LiDAR features in a unified BEV representation while maintaining per-modality strengths for 3D object detection. Our method achieves state-of-the-art performance on the large-scale nuScenes dataset, especially for challenging distant objects. The code will be made publicly available.

cs.CV↗
arXiv2022

Early Release Science of the exoplanet WASP-39b with JWST NIRCam

Measuring the metallicity and carbon-to-oxygen (C/O) ratio in exoplanet atmospheres is a fundamental step towards constraining the dominant chemical processes at work and, if in equilibrium, revealing planet formation histories. Transmission spectroscopy provides the necessary means by constraining the abundances of oxygen- and carbon-bearing species; however, this requires broad wavelength coverage, moderate spectral resolution, and high precision that, together, are not achievable with previous observatories. Now that JWST has commenced science operations, we are able to observe exoplanets at previously uncharted wavelengths and spectral resolutions. Here we report time-series observations of the transiting exoplanet WASP-39b using JWST's Near InfraRed Camera (NIRCam). The long-wavelength spectroscopic and short-wavelength photometric light curves span 2.0 - 4.0 $μ$m, exhibit minimal systematics, and reveal well-defined molecular absorption features in the planet's spectrum. Specifically, we detect gaseous H$_2$O in the atmosphere and place an upper limit on the abundance of CH$_4$. The otherwise prominent CO$_2$ feature at 2.8 $μ$m is largely masked by H$_2$O. The best-fit chemical equilibrium models favour an atmospheric metallicity of 1-100$\times$ solar (i.e., an enrichment of elements heavier than helium relative to the Sun) and a sub-stellar carbon-to-oxygen (C/O) ratio. The inferred high metallicity and low C/O ratio may indicate significant accretion of solid materials during planet formation or disequilibrium processes in the upper atmosphere.

astro-ph.EP↗
arXiv2022

Early Release Science of the Exoplanet WASP-39b with JWST NIRSpec G395H

Measuring the abundances of carbon and oxygen in exoplanet atmospheres is considered a crucial avenue for unlocking the formation and evolution of exoplanetary systems. Access to an exoplanet's chemical inventory requires high-precision observations, often inferred from individual molecular detections with low-resolution space-based and high-resolution ground-based facilities. Here we report the medium-resolution (R$\sim$600) transmission spectrum of an exoplanet atmosphere between 3-5 $μ$m covering multiple absorption features for the Saturn-mass exoplanet WASP-39b, obtained with JWST NIRSpec G395H. Our observations achieve 1.46x photon precision, providing an average transit depth uncertainty of 221 ppm per spectroscopic bin, and present minimal impacts from systematic effects. We detect significant absorption from CO$_2$ (28.5$σ$) and H$_2$O (21.5$σ$), and identify SO$_2$ as the source of absorption at 4.1 $μ$m (4.8$σ$). Best-fit atmospheric models range between 3 and 10x solar metallicity, with sub-solar to solar C/O ratios. These results, including the detection of SO$_2$, underscore the importance of characterising the chemistry in exoplanet atmospheres, and showcase NIRSpec G395H as an excellent mode for time series observations over this critical wavelength range.

astro-ph.EP↗
arXiv2022

Early Release Science of the exoplanet WASP-39b with JWST NIRISS

Transmission spectroscopy provides insight into the atmospheric properties and consequently the formation history, physics, and chemistry of transiting exoplanets. However, obtaining precise inferences of atmospheric properties from transmission spectra requires simultaneously measuring the strength and shape of multiple spectral absorption features from a wide range of chemical species. This has been challenging given the precision and wavelength coverage of previous observatories. Here, we present the transmission spectrum of the Saturn-mass exoplanet WASP-39b obtained using the SOSS mode of the NIRISS instrument on the JWST. This spectrum spans $0.6 - 2.8 μ$m in wavelength and reveals multiple water absorption bands, the potassium resonance doublet, as well as signatures of clouds. The precision and broad wavelength coverage of NIRISS-SOSS allows us to break model degeneracies between cloud properties and the atmospheric composition of WASP-39b, favoring a heavy element enhancement ("metallicity") of $\sim 10 - 30 \times$ the solar value, a sub-solar carbon-to-oxygen (C/O) ratio, and a solar-to-super-solar potassium-to-oxygen (K/O) ratio. The observations are best explained by wavelength-dependent, non-gray clouds with inhomogeneous coverage of the planet's terminator.

astro-ph.EP↗
arXiv2022

Interpretable and Effective Reinforcement Learning for Attacking against Graph-based Rumor Detection

Social networks are frequently polluted by rumors, which can be detected by advanced models such as graph neural networks. However, the models are vulnerable to attacks and understanding the vulnerabilities is critical to rumor detection in practice. To discover subtle vulnerabilities, we design a powerful attacking algorithm to camouflage rumors in social networks based on reinforcement learning that can interact with and attack any black-box detectors. The environment has exponentially large state spaces, high-order graph dependencies, and delayed noisy rewards, making the state-of-the-art end-to-end approaches difficult to learn features as large learning costs and expressive limitation of graph deep models. Instead, we design domain-specific features to avoid learning features and produce interpretable attack policies. To further speed up policy optimization, we devise: (i) a credit assignment method that decomposes delayed rewards to atomic attacking actions proportional to the their camouflage effects on target rumors; (ii) a time-dependent control variate to reduce reward variance due to large graphs and many attacking steps, supported by the reward variance analysis and a Bayesian analysis of the prediction distribution. On three real world datasets of rumor detection tasks, we demonstrate: (i) the effectiveness of the learned attacking policy compared to rule-based attacks and current end-to-end approaches; (ii) the usefulness of the proposed credit assignment strategy and variance reduction components; (iii) the interpretability of the policy when generating strong attacks via the case study.

cs.LG↗
arXiv2022

Unsupervised Scene Sketch to Photo Synthesis

Sketches make an intuitive and powerful visual expression as they are fast executed freehand drawings. We present a method for synthesizing realistic photos from scene sketches. Without the need for sketch and photo pairs, our framework directly learns from readily available large-scale photo datasets in an unsupervised manner. To this end, we introduce a standardization module that provides pseudo sketch-photo pairs during training by converting photos and sketches to a standardized domain, i.e. the edge map. The reduced domain gap between sketch and photo also allows us to disentangle them into two components: holistic scene structures and low-level visual styles such as color and texture. Taking this advantage, we synthesize a photo-realistic image by combining the structure of a sketch and the visual style of a reference photo. Extensive experimental results on perceptual similarity metrics and human perceptual studies show the proposed method could generate realistic photos with high fidelity from scene sketches and outperform state-of-the-art photo synthesis baselines. We also demonstrate that our framework facilitates a controllable manipulation of photo synthesis by editing strokes of corresponding sketches, delivering more fine-grained details than previous approaches that rely on region-level editing.

cs.CV↗
arXiv2022

Depletion of gaseous CO in protoplanetary disks by surface-energy-regulated ice formation

Empirical constraints of fundamental properties of protoplanetary disks are essential for understanding planet formation and planetary properties (1,2). Carbon monoxide (CO) gas is often used to constrain disk properties (3). However, estimates show that the CO gas abundance in disks is depleted relative to expected values (4,5,6,7), and models of various disk processes impacting the CO abundance could not explain this depletion on observed 1Myr timescales (8,9,10,11,12,13,14). Here we demonstrate that surface energy effects on particles in disks, such as the Kelvin effect, that arise when ice heterogeneously nucleates onto an existing particle can efficiently trap CO in its ice phase. In previous ice formation models, CO gas was released when small ice-coated particles were lofted to warmed disk layers. Our model can reproduce the observed abundance, distribution and time evolution of gaseous CO in the four most studied protoplanetary disks (7). We constrain the solid and gaseous CO inventory at the midplane and disk diffusivities and resolve inconsistencies in estimates of the disk mass -- three crucial parameters that control planetary formation.

astro-ph.EP↗
arXiv2022

Strain tunability of perpendicular magnetic anisotropy in van der Waals ferromagnets VI3

Layered ferromagnets with high coercivity have special applications in nanoscale memory elements in electronic circuits, such as data storage. Therefore, searching for new hard ferromagnets and effectively tuning or enhancing the coercivity are the hottest topics in layered magnets today. Here, we report a strain tunability of perpendicular magnetic anisotropy in van der Waals (vdW) ferromagnets VI3 using magnetic circular dichroism measurements. For an unstrained flake, the M-H curve shows a rectangular-shaped hysteresis loop with perpendicular magnetic anisotropy and a large coercivity (up to 1.775 T at 10 K). Furthermore, the coercivity can be enhanced to a maximum of 2.6 T at 10 K under a 2.9% in-plane tensile strain. Our DFT calculations show that the magnetic anisotropy energy (MAE) can be dramatically increased after applying an in-plain tensile strain, which contributes to the enhancement of coercivity in the VI3 flake. Meanwhile, the strain tunability on the coercivity of CrI3, with a similar crystal structure, is limited. The main reason is the strong spin-orbital coupling in V3+ in VI6 octahedra in comparison with that in Cr3+. The strain tunability of coercivity in VI3 flakes highlights its potential for integration into vdW heterostructures, paving the way toward nanoscale spintronic devices and applications in the future.

cond-mat.mtrl-sci↗
arXiv2022

Structured Graph Variational Autoencoders for Indoor Furniture layout Generation

We present a structured graph variational autoencoder for generating the layout of indoor 3D scenes. Given the room type (e.g., living room or library) and the room layout (e.g., room elements such as floor and walls), our architecture generates a collection of objects (e.g., furniture items such as sofa, table and chairs) that is consistent with the room type and layout. This is a challenging problem because the generated scene should satisfy multiple constrains, e.g., each object must lie inside the room and two objects cannot occupy the same volume. To address these challenges, we propose a deep generative model that encodes these relationships as soft constraints on an attributed graph (e.g., the nodes capture attributes of room and furniture elements, such as class, pose and size, and the edges capture geometric relationships such as relative orientation). The architecture consists of a graph encoder that maps the input graph to a structured latent space, and a graph decoder that generates a furniture graph, given a latent code and the room graph. The latent space is modeled with auto-regressive priors, which facilitates the generation of highly structured scenes. We also propose an efficient training procedure that combines matching and constrained learning. Experiments on the 3D-FRONT dataset show that our method produces scenes that are diverse and are adapted to the room layout.

cs.CV↗
arXiv2022

A Simple Condensation Model for the H2SO4-H2O Gas-cloud System on Venus

The current Venus climate is largely regulated by globally-covered concentrated sulfuric acid clouds from binary condensation of sulfuric acid (H2SO4) and water (H2O). To understand this complicated H2SO4-H2O gas-cloud system, previous theoretical studies either adopted complicated microphysical calculations or assumed that both H2SO4 and H2O vapor follow their saturation vapor pressure. In this study, we developed a simple one-dimensional cloud condensation model including condensation, diffusion and sedimentation of H2SO4 and H2O but without detailed microphysics. Our model is able to explain the observed vertical structure of cloud and upper haze mass loading, cloud acidity, H2SO4, and H2O vapor, and the mode-2 particle size on Venus. We found that most H2SO4 is stored in the condensed phase above 48 km, while the partitioning of H2O between the vapor and clouds is complicated. The cloud cycle is mostly driven by evaporation and condensation of H2SO4 rather than H2O and is about seven times stronger than the H2SO4 photochemical cycle. Most of the condensed H2O in the upper clouds is evaporated before the falling particles reach the middle clouds. The cloud acidity is affected by the temperature and the condensation-evaporation cycles of both H2SO4 and H2O. Because of the large chemical production of H2SO4 vapor and relatively inefficient cloud condensation, the simulated H2SO4 vapor above 60 km is largely supersaturated by more than two orders of magnitude, which could be tested by future observations.

astro-ph.EP↗
arXiv2022

A Fast, Semi-analytical Model for the Venusian Binary Cloud System

The Venusian clouds originate from the binary condensation of H$_{2}$SO$_{4}$ and H$_{2}$O. The two components strongly interact with each other via chemistry and cloud formation. Previous works adopted sophisticated microphysical approaches to understand the clouds. Here we show that the observed vapor and cloud distributions on Venus can be well explained by a semi-analytical model. Our model assumes local thermodynamical equilibrium for water vapor but not for sulfuric acid vapor, and includes the feedback of cloud condensation and acidity to vapor distributions. The model predicts strong supersaturation of the H$_{2}$SO$_{4}$ vapor above 60 km, consistent with our recent cloud condensation model. The semi-analytical model is 100 times faster than the condensation model and 1000 times faster than the microphysical models. This allows us to quickly explore a large parameter space of the sulfuric acid gas-cloud system. We found that the cloud mass loading in the upper clouds has an opposite response of that in the lower clouds to the vapor mixing ratios in the lower atmosphere. The transport of water vapor influences the cloud acidity in all cloud layers while the transport of sulfuric acid vapor only dominates in the lower clouds. This cloud model is fast enough to be coupled with the climate models and chemistry models to understand the cloudy atmospheres of Venus and Venus-like extra-solar planets.

astro-ph.EP↗