arXiv subjects

Zhihui Li

Publications and source records attributed to Zhihui Li.

At least 55 records · Page 3Linked to original sources

arXiv2024

Metrical theory of power-2-decaying Gauss-like expansion

Each $x\in (0,1]$ can be uniquely expanded as a power-2-decaying Gauss-like expansion, in the form of \begin{equation*} x=\sum_{i=1}^{\infty}2^{-(d_1(x)+d_2(x)+\cdots+d_i(x))},\qquad d_i(x)\in \mathbb{N}. \end{equation*} Let $ϕ:\mathbb{N}\to \mathbb{R}^{+}$ be an arbitrary positive function. We are interested in the size of the set $$F(ϕ)=\{x\in (0,1]:d_n(x)\ge ϕ(n)~~\text{for infinity many}~n\}.$$ We prove a Borel-Bernstein theorem on the zero-one law of the Lebesgue measure of $F(ϕ)$. When the Lebesgue measure of $F(ϕ)$ is zero, we calculate its Hausdorff dimension. Furthermore, we analyse the growth rate of the maximal digit among the first $n$ digits from probability and multifractal perspectives.

math.NT↗

arXiv2023

The Circumgalactic Medium of Extreme Emission Line Galaxies at z ~ 2: Resolved Spectroscopy and Radiative Transfer Modeling of Spatially Extended Lyman-alpha Emission in the KBSS-KCWI Survey

The resonantly scattered Lyman-$α$ line illuminates the extended halos of neutral hydrogen in the circumgalactic medium of galaxies. We present integral field Keck Cosmic Web Imager observations of double-peaked, spatially extended Ly$α$ emission in 12 relatively low-mass ($M_{\star} \sim10^9 \, M_{\odot}$) $z\sim2$ galaxies characterized by extreme nebular emission lines. Using individual spaxels and small bins as well as radially binned profiles of larger regions, we find that for most objects in the sample the Ly$α$ blue-to-red peak ratio increases, the peak separation decreases, and the fraction of flux emerging at line center increases with radius. We use new radiative transfer simulations to model each galaxy with a clumpy, multiphase outflow with radially varying outflow velocity, and self-consistently apply the same velocity model to the low ionization interstellar absorption lines. These models reproduce the trends of peak ratio, peak separation and trough depth with radius, and broadly reconcile outflow velocities inferred from Ly$α$ and absorption lines. The galaxies in our sample are well-described by a model in which neutral, outflowing clumps are embedded in a hotter, more highly ionized inter-clump medium (ICM), whose residual neutral content produces absorption at the systemic redshift. The peak ratio, peak separation and trough flux fraction are primarily governed by the line-of-sight component of the outflow velocity, the HI column density, and the residual neutral density in the ICM respectively. Azimuthal asymmetries in the line profile further suggest non-radial gas motions at large radii and variations in the HI column density in the outer halos.

astro-ph.GA↗

arXiv2023

MARLlib: A Scalable and Efficient Multi-agent Reinforcement Learning Library

A significant challenge facing researchers in the area of multi-agent reinforcement learning (MARL) pertains to the identification of a library that can offer fast and compatible development for multi-agent tasks and algorithm combinations, while obviating the need to consider compatibility issues. In this paper, we present MARLlib, a library designed to address the aforementioned challenge by leveraging three key mechanisms: 1) a standardized multi-agent environment wrapper, 2) an agent-level algorithm implementation, and 3) a flexible policy mapping strategy. By utilizing these mechanisms, MARLlib can effectively disentangle the intertwined nature of the multi-agent task and the learning process of the algorithm, with the ability to automatically alter the training strategy based on the current task's attributes. The MARLlib library's source code is publicly accessible on GitHub: \url{https://github.com/Replicable-MARL/MARLlib}.

cs.LG↗

arXiv2023

Investigation of Compressor Cascade Flow Using Physics- Informed Neural Networks with Adaptive Learning Strategy

In this study, we utilize the emerging Physics Informed Neural Networks (PINNs) approach for the first time to predict the flow field of a compressor cascade. Different from conventional training methods, a new adaptive learning strategy that mitigates gradient imbalance through incorporating adaptive weights in conjunction with dynamically adjusting learning rate is used during the training process to improve the convergence of PINNs. The performance of PINNs is assessed here by solving both the forward and inverse problems. In the forward problem, by encapsulating the physical relations among relevant variables, PINNs demonstrate their effectiveness in accurately forecasting the compressor's flow field. PINNs also show obvious advantages over the traditional CFD approaches, particularly in scenarios lacking complete boundary conditions, as is often the case in inverse engineering problems. PINNs successfully reconstruct the flow field of the compressor cascade solely based on partial velocity vectors and near-wall pressure information. Furthermore, PINNs show robust performance in the environment of various levels of aleatory uncertainties stemming from labeled data. This research provides evidence that PINNs can offer turbomachinery designers an additional and promising option alongside the current dominant CFD methods.

cs.LG↗

arXiv2023

Surrogate modelling and uncertainty quantification based on multi-fidelity deep neural network

To reduce training costs, several Deep neural networks (DNNs) that can learn from a small set of HF data and a sufficient number of low-fidelity (LF) data have been proposed. In these established neural networks, a parallel structure is commonly proposed to separately approximate the non-linear and linear correlation between the HF- and LF data. In this paper, a new architecture of multi-fidelity deep neural network (MF-DNN) was proposed where one subnetwork was built to approximate both the non-linear and linear correlation simultaneously. Rather than manually allocating the output weights for the paralleled linear and nonlinear correction networks, the proposed MF-DNN can autonomously learn arbitrary correlation. The prediction accuracy of the proposed MF-DNN was firstly demonstrated by approximating the 1-, 32- and 100-dimensional benchmark functions with either the linear or non-linear correlation. The surrogating modelling results revealed that MF-DNN exhibited excellent approximation capabilities for the test functions. Subsequently, the MF DNN was deployed to simulate the 1-, 32- and 100-dimensional aleatory uncertainty propagation progress with the influence of either the uniform or Gaussian distributions of input uncertainties. The uncertainty quantification (UQ) results validated that the MF-DNN efficiently predicted the probability density distributions of quantities of interest (QoI) as well as the statistical moments without significant compromise of accuracy. MF-DNN was also deployed to model the physical flow of turbine vane LS89. The distributions of isentropic Mach number were well-predicted by MF-DNN based on the 2D Euler flow field and few experimental measurement data points. The proposed MF-DNN should be promising in solving UQ and robust optimization problems in practical engineering applications with multi-fidelity data sources.

physics.comp-ph↗

arXiv2023

ALPACA: A New Semi-Analytic Model for Metal Absorption Lines Emerging from Clumpy Galactic Environments

We present a new semi-analytic formalism for modeling metal absorption lines that emerge from a clumpy galactic environment, ALPACA. We predict the ''down-the-barrel'' (DTB) metal absorption line profiles and the EW of absorption at different impact parameters as a function of the properties of the clumps, including the clump kinematics, the clump volume filling factor, the clump number density profile and the clump ion column densities. With ALPACA, we jointly model the stacked DTB CII$λ$1334 spectrum of a sample of $z \sim$ 3 Lyman break galaxies and the EW v.s. $b$ profile of a sample of $z \sim$ 2 star-forming galaxy-galaxy pairs. ALPACA successfully reproduced two datasets simultaneously, and the best-fit prefers a low clump volume filling factor ($\sim 3 \times 10^{-3}$). The radial velocities of the clumps are a superposition of a rapidly accelerated outflow with a maximum velocity of $\sim 400\,\rm km\,s^{-1}$ and a velocity dispersion of $σ_{\rm cl} \sim\,120 \rm km\,s^{-1}$. The joint modeling reveals a physical scenario where the absorption observed at a particular velocity is contributed by the clumps distributed over a fairly broad range of radii. We also find that the commonly adopted Sobolev approximation is at best only applicable within a narrow range of radii where the clumps are undergoing rapid acceleration in a non-volume-filling clumpy medium. Lastly, we find that the clump radial velocity profile may not be fully constrained by the joint modeling and spatially-resolved Ly$α$ emission modeling may help break the degeneracy.

astro-ph.GA↗

arXiv2023

SU-Net: Pose estimation network for non-cooperative spacecraft on-orbit

Spacecraft pose estimation plays a vital role in many on-orbit space missions, such as rendezvous and docking, debris removal, and on-orbit maintenance. At present, space images contain widely varying lighting conditions, high contrast and low resolution, pose estimation of space objects is more challenging than that of objects on earth. In this paper, we analyzing the radar image characteristics of spacecraft on-orbit, then propose a new deep learning neural Network structure named Dense Residual U-shaped Network (DR-U-Net) to extract image features. We further introduce a novel neural network based on DR-U-Net, namely Spacecraft U-shaped Network (SU-Net) to achieve end-to-end pose estimation for non-cooperative spacecraft. Specifically, the SU-Net first preprocess the image of non-cooperative spacecraft, then transfer learning was used for pre-training. Subsequently, in order to solve the problem of radar image blur and low ability of spacecraft contour recognition, we add residual connection and dense connection to the backbone network U-Net, and we named it DR-U-Net. In this way, the feature loss and the complexity of the model is reduced, and the degradation of deep neural network during training is avoided. Finally, a layer of feedforward neural network is used for pose estimation of non-cooperative spacecraft on-orbit. Experiments prove that the proposed method does not rely on the hand-made object specific features, and the model has robust robustness, and the calculation accuracy outperforms the state-of-the-art pose estimation methods. The absolute error is 0.1557 to 0.4491 , the mean error is about 0.302 , and the standard deviation is about 0.065 .

cs.CV↗

arXiv2023

Efficient Quantum Secret Sharing Scheme Based On Monotone Span Program

How to efficiently share secrets among multiple participants is a very important problem in key management. In this paper, we propose a multi-secret sharing scheme based on the GHZ state. First, the distributor uses monotone span program to encode the secrets and generate the corresponding secret shares to send to the participants. Then, each participant uses the generalized Pauli operator to embed its own secret share into the transmitted particle. The participant who wants to get the secrets can get multiple secrets at the same time by performing a GHZ-state joint measurement. Futhermore, the scheme is based on a monotone span program, and its access structure is more general than the access structure (t,n) threshold. Compared with other schemes, our proposed scheme is more efficient, less computational cost.

quant-ph↗

arXiv2023

Reentry Risk and Safety Assessment of Spacecraft Debris Based on Machine Learning

Uncontrolled spacecraft will disintegrate and generate a large amount of debris in the reentry process, and ablative debris may cause potential risks to the safety of human life and property on the ground. Therefore, predicting the landing points of spacecraft debris and forecasting the degree of risk of debris to human life and property is very important. In view that it is difficult to predict the process of reentry process and the reentry point in advance, and the debris generated from reentry disintegration may cause ground damage for the uncontrolled space vehicle on expiration of service. In this paper, we adopt the object-oriented approach to consider the spacecraft and its disintegrated components as consisting of simple basic geometric models, and introduce three machine learning models: the support vector regression (SVR), decision tree regression (DTR) and multilayer perceptron (MLP) to predict the velocity, longitude and latitude of spacecraft debris landing points for the first time. Then, we compare the prediction accuracy of the three models. Furthermore, we define the reentry risk and the degree of danger, and we calculate the risk level for each spacecraft debris and make warnings accordingly. The experimental results show that the proposed method can obtain high accuracy prediction results in at least 15 seconds and make safety level warning more real-time.

cs.LG↗

arXiv2022

A physical perturbation based study on the prediction of free-fall disks with chaotic modes in the water

We report a phenomenon that physical perturbations sometimes can benefit the certainty of a free-fall motion with chaotic modes, albeit, as commonly believed, they can ruin it. We statistically compare those factors that may lead to uncertainty, by which we find that the growth of the standard deviation of the landing locations is directly determined by the physical perturbations. A significant yardstick is defined in the meantime. This temporal criterion is of big relevance to the replicability of such problems experimentally, although they are inherently chaotic. Our hypothesis is verified by experiments from other literature. This outcome also provides a practical strategy to evaluate the credible prediction time by estimating the disturbances from physical parameters as a priori.

nlin.CD↗

arXiv2022

Where Outflows Meet Inflows: Gas Kinematics in SSA22 Lyman-$α$ Blob 2 Decoded by Advanced Radiative Transfer Modelling

We present new spectroscopic observations of Lyman-$α$ (Ly$α$) Blob 2 ($z \sim$ 3.1). We observed extended Ly$α$ emission in three distinct regions, where the highest Ly$α$ surface brightness (SB) center is far away from the known continuum sources. We searched through the MOSFIRE slits that cover the high Ly$α$ SB regions, but were unable to detect any significant nebular emission near the highest SB center. We further mapped the flux ratio of the blue peak to the red peak and found it is anti-correlated with Ly$α$ SB with a power-law index of $\sim$ -0.4. We used radiative transfer models with both multiphase, clumpy and shell geometries and successfully reproduced the diverse Ly$α$ morphologies. We found that most spectra suggest outflow-dominated kinematics, while 4/15 spectra imply inflows. A significant correlation exists between parameter pairs, and the multiphase, clumpy model may alleviate previously reported discrepancies. We also modeled Ly$α$ spectra at different positions simultaneously and found that the variation of the inferred clump outflow velocities can be approximately explained by line-of-sight projection effects. Our results support the `central powering + scattering' scenario, i.e. the Ly$α$ photons are generated by a central powering source and then scatter with outflowing, multiphase HI gas while propagating outwards. The infalling of cool gas near the blob outskirts shapes the observed blue-dominated Ly$α$ profiles, but its energy contribution to the total Ly$α$ luminosity is less than 10%, i.e. minor compared to the photo-ionization by star-forming galaxies and/or AGNs.

astro-ph.GA↗

arXiv2022

A Comprehensive Survey of Scene Graphs: Generation and Application

Scene graph is a structured representation of a scene that can clearly express the objects, attributes, and relationships between objects in the scene. As computer vision technology continues to develop, people are no longer satisfied with simply detecting and recognizing objects in images; instead, people look forward to a higher level of understanding and reasoning about visual scenes. For example, given an image, we want to not only detect and recognize objects in the image, but also know the relationship between objects (visual relationship detection), and generate a text description (image captioning) based on the image content. Alternatively, we might want the machine to tell us what the little girl in the image is doing (Visual Question Answering (VQA)), or even remove the dog from the image and find similar images (image editing and retrieval), etc. These tasks require a higher level of understanding and reasoning for image vision tasks. The scene graph is just such a powerful tool for scene understanding. Therefore, scene graphs have attracted the attention of a large number of researchers, and related research is often cross-modal, complex, and rapidly developing. However, no relatively systematic survey of scene graphs exists at present. To this end, this survey conducts a comprehensive investigation of the current scene graph research. More specifically, we first summarized the general definition of the scene graph, then conducted a comprehensive and systematic discussion on the generation method of the scene graph (SGG) and the SGG with the aid of prior knowledge. We then investigated the main applications of scene graphs and summarized the most commonly used datasets. Finally, we provide some insights into the future development of scene graphs. We believe this will be a very helpful foundation for future research on scene graphs.

cs.CV↗

arXiv2021

A Survey of Deep Active Learning

Active learning (AL) attempts to maximize the performance gain of the model by marking the fewest samples. Deep learning (DL) is greedy for data and requires a large amount of data supply to optimize massive parameters, so that the model learns how to extract high-quality features. In recent years, due to the rapid development of internet technology, we are in an era of information torrents and we have massive amounts of data. In this way, DL has aroused strong interest of researchers and has been rapidly developed. Compared with DL, researchers have relatively low interest in AL. This is mainly because before the rise of DL, traditional machine learning requires relatively few labeled samples. Therefore, early AL is difficult to reflect the value it deserves. Although DL has made breakthroughs in various fields, most of this success is due to the publicity of the large number of existing annotation datasets. However, the acquisition of a large number of high-quality annotated datasets consumes a lot of manpower, which is not allowed in some fields that require high expertise, especially in the fields of speech recognition, information extraction, medical images, etc. Therefore, AL has gradually received due attention. A natural idea is whether AL can be used to reduce the cost of sample annotations, while retaining the powerful learning capabilities of DL. Therefore, deep active learning (DAL) has emerged. Although the related research has been quite abundant, it lacks a comprehensive survey of DAL. This article is to fill this gap, we provide a formal classification method for the existing work, and a comprehensive and systematic overview. In addition, we also analyzed and summarized the development of DAL from the perspective of application. Finally, we discussed the confusion and problems in DAL, and gave some possible development directions for DAL.

cs.LG↗

arXiv2021

Projective robustness for quantum channels and measurements and their operational significance

Recently, the projective robustness of quantum states has been introduced in [arXiv:2109.04481(2021)]. It shows that the projective robustness is a useful resource monotone and can comprehensively characterize capabilities and limitations of probabilistic protocols manipulating quantum resources deterministically. In this paper, we will extend the projective robustness to any convex resource theories of quantum channels and measurements. First, We introduce the projective robustness of quantum channels and prove that it satisfies some good properties, especially sub- or supermultiplicativity under any free quantum process. Moreover, we use the projective robustness of channels to give lower bounds on the errors and overheads in any channel resource distillation. Meanwhile, we show that the projective robustness of channels quantifies the maximal advantage that a given channel outperforms all free channels in simultaneous discrimination and exclusion of a fixed state ensemble. Second, we define the projective robustness of quantum measurements and prove that it exactly quantifies the maximal advantage that a given measurement provides over all free measurements in simultaneous discrimination and exclusion of two fixed state ensembles. Finally, within a specific channel resource setting based on measurement incompatibility, we show that the projective robustness of quantum channels coincides with the projective robustness of measurement incompatibility.

quant-ph↗

arXiv2021

Deciphering the Lyman-$α$ Emission Line: Towards the Understanding of Galactic Properties Extracted from Ly$α$ Spectra via Radiative Transfer Modeling

Existing ubiquitously in the Universe with the highest luminosity, the Lyman-$α$ emission line encodes abundant physical information about the gaseous medium it interacts with. Nevertheless, the resonant nature of Ly$α$ complicates the radiative transfer (RT) modeling of the line profile, making the extraction of physical properties of the surrounding gaseous medium notoriously difficult. In this paper, we revisit the problem of deciphering the Ly$α$ emission line with RT modeling. We reveal intrinsic parameter degeneracies in the widely-used shell model in the optically thick regime for both static and outflowing cases, which suggest the limitations of the model. We have also explored the connection between the more physically realistic multiphase, clumpy model and the shell model. We find that the parameters of a ``very clumpy'' slab model and the shell model have the following correspondences: (1) the total column density of the clumpy slab model is equal to the HI column density of the shell model; (2) the effective temperature of the clumpy slab model, which incorporates the clump velocity dispersion, is equal to the effective temperature of the shell model; (3) the average radial clump outflow velocity is equal to the shell expansion velocity; (4) large intrinsic line widths are required in the shell model to reproduce the wings of the clumpy slab models; (5) adding another phase of hot inter-clump medium will increase peak separation, and the fitted shell expansion velocity lies between the outflow velocities of two phases of gas. Our results provide a viable solution to the major discrepancies associated with Ly$α$ fitting reported in previous literature, and emphasize the importance of utilizing information from additional observations to break the intrinsic degeneracies as well as interpreting the model parameters in a more physically realistic context.

astro-ph.GA↗

arXiv2021

DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers

Dynamic networks have shown their promising capability in reducing theoretical computation complexity by adapting their architectures to the input during inference. However, their practical runtime usually lags behind the theoretical acceleration due to inefficient sparsity. Here, we explore a hardware-efficient dynamic inference regime, named dynamic weight slicing, which adaptively slice a part of network parameters for inputs with diverse difficulty levels, while keeping parameters stored statically and contiguously in hardware to prevent the extra burden of sparse computation. Based on this scheme, we present dynamic slimmable network (DS-Net) and dynamic slice-able network (DS-Net++) by input-dependently adjusting filter numbers of CNNs and multiple dimensions in both CNNs and transformers, respectively. To ensure sub-network generality and routing fairness, we propose a disentangled two-stage optimization scheme with training techniques such as in-place bootstrapping (IB), multi-view consistency (MvCo) and sandwich gate sparsification (SGS) to train supernet and gate separately. Extensive experiments on 4 datasets and 3 different network architectures demonstrate our method consistently outperforms state-of-the-art static and dynamic model compression methods by a large margin (up to 6.6%). Typically, DS-Net++ achieves 2-4x computation reduction and 1.62x real-world acceleration over MobileNet, ResNet-50 and Vision Transformer, with minimal accuracy drops (0.1-0.3%) on ImageNet. Code release: https://github.com/changlin31/DS-Net

cs.CV↗

arXiv2021

The KBSS-KCWI Survey: The connection between extended Ly$α$ halos and galaxy azimuthal angle at $z\sim 2-3$

We present the first statistical analysis of kinematically-resolved, spatially-extended Ly$α$ emission around $z = 2-3$ galaxies in the Keck Baryonic Structure Survey (KBSS) using the Keck Cosmic Web Imager (KCWI). Our sample of 59 star-forming galaxies ($z_\mathrm{med} = 2.29$) comprises the subset with typical KCWI integration times of ~5 hours and with existing imaging data from the Hubble Space Telescope and/or adaptive optics-assisted integral field spectroscopy. The high resolution images were used to evaluate the azimuthal dependence of the diffuse Ly$α$ emission with respect to the stellar continuum within projected galactocentric distances of $\lesssim 30$ proper kpc. We introduce cylindrically-projected 2D spectra (CP2D) that map the averaged Ly$α$ spectral profile over a specified range of azimuthal angle, as a function of impact parameter around galaxies. The averaged CP2D spectrum of all galaxies shows clear signatures of Ly$α$ resonant scattering by outflowing gas. We stacked the CP2D spectra of individual galaxies over ranges of azimuthal angle with respect to their major axes. The extended Ly$α$ emission along the galaxy principal axes are statistically indistinguishable, with residual asymmetry of $\le$ 2% ($\sim 2 σ$) of the integrated Ly$α$ emission. The symmetry implies that the Ly$α$ scattering medium is dominated by outflows in all directions within 30 kpc. Meanwhile, we find that the blueshifted component of Ly$α$ emission is marginally stronger along galaxy minor axes for galaxies with relatively weak Ly$α$ emission. We speculate that this weak directional dependence of Ly$α$ emission becomes discernible only when the Ly$α$ escape fraction is low. These discoveries highlight the need for similar analyses in simulations with Ly$α$ radiative transfer modeling.

astro-ph.GA↗

arXiv2021

Dynamic Slimmable Network

Current dynamic networks and dynamic pruning methods have shown their promising capability in reducing theoretical computation complexity. However, dynamic sparse patterns on convolutional filters fail to achieve actual acceleration in real-world implementation, due to the extra burden of indexing, weight-copying, or zero-masking. Here, we explore a dynamic network slimming regime, named Dynamic Slimmable Network (DS-Net), which aims to achieve good hardware-efficiency via dynamically adjusting filter numbers of networks at test time with respect to different inputs, while keeping filters stored statically and contiguously in hardware to prevent the extra burden. Our DS-Net is empowered with the ability of dynamic inference by the proposed double-headed dynamic gate that comprises an attention head and a slimming head to predictively adjust network width with negligible extra computation cost. To ensure generality of each candidate architecture and the fairness of gate, we propose a disentangled two-stage training scheme inspired by one-shot NAS. In the first stage, a novel training technique for weight-sharing networks named In-place Ensemble Bootstrapping is proposed to improve the supernet training efficacy. In the second stage, Sandwich Gate Sparsification is proposed to assist the gate training by identifying easy and hard samples in an online way. Extensive experiments demonstrate our DS-Net consistently outperforms its static counterparts as well as state-of-the-art static and dynamic model compression methods by a large margin (up to 5.9%). Typically, DS-Net achieves 2-4x computation reduction and 1.62x real-world acceleration over ResNet-50 and MobileNet with minimal accuracy drops on ImageNet. Code release: https://github.com/changlin31/DS-Net .

cs.CV↗