arXiv subjects

Bin Wang

Publications and source records attributed to Bin Wang.

At least 541 records · Page 30Linked to original sources

arXiv2022

Dynamic generation or removal of a scalar hair

We study dynamic processes through which the scalar hair of black holes is generated or detached in a theory with a scalar field non-minimally coupled to Gauss-Bonnet and Ricci scalar invariants. We concentrate on the nonlinear temporal evolution of a far-from-equilibrium gravitational system. In our simulations, we choose the initial spacetime to be either a bald Schwarzschild or a scalarized spherically symmetric black hole. Succeeding continuous accretion of the scalar field onto the original black hole, the final fate of the system displays intriguing features, which depend on the initial configurations, strengths of the perturbation, and specific metric parameters. In addition to the scalarization process through which the bald black hole addresses scalar hair, we observe the dynamical descalarization, which removes scalar hair from an original hairy hole after continuous scalar field accretion. We examine the temporal evolution of the scalar field, the metrics, and the Misner-Sharp mass of the spacetime and exhibit rich phase structures through nonlinear dynamical processes.

gr-qc↗

arXiv2022

Optimal Allocation of Virtual Inertia and Droop Control for Renewable Energy in Stochastic Look-Ahead Power Dispatch

To stabilize the frequency of the renewable energy sources (RESs) dominated power system, frequency supports are required by RESs through virtual inertia emulation or droop control in the newly published grid codes. Since the long-term RES prediction involves significant errors, we need online configure the frequency control parameters of RESs in a rolling manner to improve the operation economics under the premise of stabilizing system frequency. To address this concern, this paper proposes a frequency constrained stochastic look-ahead power dispatch (FCS-LAPD) model to formulate the frequency control parameters of RESs and Energy Storage Systems (ESSs) as scheduling variables, which can optimally allocate the virtual inertia and droop coefficient of RESs and ESSs. In this FCS-LAPD model, the uncertainties of RESs are characterized using Gaussian Mixture Model (GMM). The required reserves are determined by frequency control parameters, and the reserve cost coefficients are adjusted properly to allocate the reserves according to the predicted power generation. Due to the nonlinearity of the frequency nadir constraint, a convex hull approximation method is proposed to linearize it with guaranteed feasibility. The proposed FCS-LAPD is ultimately cast as an instance of quadratic programming and can be efficiently solved. Case studies on modified IEEE 24-bus system and a provincial power system in China are conducted to show the effectiveness of the proposed model.

eess.SY↗

arXiv2022

Towards Generalized Open Information Extraction

Open Information Extraction (OpenIE) facilitates the open-domain discovery of textual facts. However, the prevailing solutions evaluate OpenIE models on in-domain test sets aside from the training corpus, which certainly violates the initial task principle of domain-independence. In this paper, we propose to advance OpenIE towards a more realistic scenario: generalizing over unseen target domains with different data distributions from the source training domains, termed Generalized OpenIE. For this purpose, we first introduce GLOBE, a large-scale human-annotated multi-domain OpenIE benchmark, to examine the robustness of recent OpenIE models to domain shifts, and the relative performance degradation of up to 70% implies the challenges of generalized OpenIE. Then, we propose DragonIE, which explores a minimalist graph expression of textual fact: directed acyclic graph, to improve the OpenIE generalization. Extensive experiments demonstrate that DragonIE beats the previous methods in both in-domain and out-of-domain settings by as much as 6.0% in F1 score absolutely, but there is still ample room for improvement.

cs.CL↗

arXiv2022

Prototypical context-aware dynamics generalization for high-dimensional model-based reinforcement learning

The latent world model provides a promising way to learn policies in a compact latent space for tasks with high-dimensional observations, however, its generalization across diverse environments with unseen dynamics remains challenging. Although the recurrent structure utilized in current advances helps to capture local dynamics, modeling only state transitions without an explicit understanding of environmental context limits the generalization ability of the dynamics model. To address this issue, we propose a Prototypical Context-Aware Dynamics (ProtoCAD) model, which captures the local dynamics by time consistent latent context and enables dynamics generalization in high-dimensional control tasks. ProtoCAD extracts useful contextual information with the help of the prototypes clustered over batch and benefits model-based RL in two folds: 1) It utilizes a temporally consistent prototypical regularizer that encourages the prototype assignments produced for different time parts of the same latent trajectory to be temporally consistent instead of comparing the features; 2) A context representation is designed which combines both the projection embedding of latent states and aggregated prototypes and can significantly improve the dynamics generalization ability. Extensive experiments show that ProtoCAD surpasses existing methods in terms of dynamics generalization. Compared with the recurrent-based model RSSM, ProtoCAD delivers 13.2% and 26.7% better mean and median performance across all dynamics generalization tasks.

cs.LG↗

arXiv2022

Global Adaptive Generative Adjustment

Many traditional signal recovery approaches can behave well basing on the penalized likelihood. However, they have to meet with the difficulty in the selection of hyperparameters or tuning parameters in the penalties. In this article, we propose a global adaptive generative adjustment (GAGA) algorithm for signal recovery, in which multiple hyperpameters are automatically learned and alternatively updated with the signal. We further prove that the output of our algorithm directly guarantees the consistency of model selection and signal estimate. Moreover, we also propose a variant GAGA algorithm for improving the computational efficiency in the high-dimensional data analysis. Finally, in the simulated experiment, we consider the consistency of the outputs of our algorithms, and compare our algorithms to other penalized likelihood methods: the Adaptive LASSO, the SCAD and the MCP. The simulation results support the efficiency of our algorithms for signal recovery, and demonstrate that our algorithms outperform the other algorithms.

stat.ML↗

arXiv2022

Photon region and shadow of a rotating 5D black string

To explore the possible clues for the extra dimension from the Event Horizon Telescope (EHT) observations, we study the shadow of the rotating 5D black string in General Relativity (GR). Instead of investigating the shadow in the effective 4D theory, we concern the motion of photons along the extra dimension $z$ with a conserved momentum $P_z$, which appears as an effective mass in the geodesic equations of photons. The existence of $P_z$ enlarges the photon regions and the shadow of the rotating 5D black string while it has slight impact on the distortion. The EHT observations of M87* and SgrA* can rule out the black string model with an infinite length along the extra dimension, and support the hypothesis that the extra dimension is compact to avoid the Gregory-Laflamme (GL) instability, where the length of the black string/the compact extra dimension can be constrained as $2.03125~\rm{mm} \lesssim \ell \lesssim 2.6~\rm{mm}$ and $2.28070~\rm{mm} \lesssim \ell \lesssim 2.6~\rm{mm}$ respectively.

gr-qc↗

arXiv2022

The length of a compact extra dimension from black hole shadow

Plenty of efforts have been made to explore the black string and its instability, but the fate of the black strings with fewer extra dimensions is still inconclusive. Now starting from the 5D uniform black string, we show that the EHT observations of M87* and SgrA* can not only rule out the black string with an infinite extra dimension, but also constrain the length of a compact extra dimension, which is much smaller than the critical length given from the Gregory-Laflamme (GL) instability. Our findings support the hypothesis that the extra dimension is compact avoiding the GL instability.

gr-qc↗

arXiv2022

Generate, Discriminate and Contrast: A Semi-Supervised Sentence Representation Learning Framework

Most sentence embedding techniques heavily rely on expensive human-annotated sentence pairs as the supervised signals. Despite the use of large-scale unlabeled data, the performance of unsupervised methods typically lags far behind that of the supervised counterparts in most downstream tasks. In this work, we propose a semi-supervised sentence embedding framework, GenSE, that effectively leverages large-scale unlabeled data. Our method include three parts: 1) Generate: A generator/discriminator model is jointly trained to synthesize sentence pairs from open-domain unlabeled corpus; 2) Discriminate: Noisy sentence pairs are filtered out by the discriminator to acquire high-quality positive and negative sentence pairs; 3) Contrast: A prompt-based contrastive approach is presented for sentence representation learning with both annotated and synthesized data. Comprehensive experiments show that GenSE achieves an average correlation score of 85.19 on the STS datasets and consistent performance improvement on four domain adaptation tasks, significantly surpassing the state-of-the-art methods and convincingly corroborating its effectiveness and generalization ability.Code, Synthetic data and Models available at https://github.com/MatthewCYM/GenSE.

cs.CL↗

arXiv2022

A Focused Study on Sequence Length for Dialogue Summarization

Output length is critical to dialogue summarization systems. The dialogue summary length is determined by multiple factors, including dialogue complexity, summary objective, and personal preferences. In this work, we approach dialogue summary length from three perspectives. First, we analyze the length differences between existing models' outputs and the corresponding human references and find that summarization models tend to produce more verbose summaries due to their pretraining objectives. Second, we identify salient features for summary length prediction by comparing different model settings. Third, we experiment with a length-aware summarizer and show notable improvement on existing models if summary length can be well incorporated. Analysis and experiments are conducted on popular DialogSum and SAMSum datasets to validate our findings.

cs.CL↗

arXiv2022

Descalarization by Quenching Charged Hairy Black Hole in asymptotically AdS spacetime

In this work, we study the real-time dynamics of the charged hairy black hole with the time-dependent source of scalar field in asymptotically anti-de Sitter (AdS) spacetime. The numerical results reveal a novel descalarization mechanism. In order to obtain the hairy black hole as the initial data for the quench process, we first analyze the quasi-normal modes of the massive complex scalar field on the Reissner-Nordström anti-de Sitter (RN-AdS) black hole background. We find the dominant unstable modes for large and small RN-AdS black holes come from the zero-damped modes and AdS modes, respectively. Then, the unstable RN-AdS black holes are perturbed to trigger the transition to hairy black holes. With the hairy black hole in hand, we specify a time dependent scalar source for the system. As the source is turned on, the electric charge, energy and scalar condensation of the system start to oscillate with the entropy increasing monotonically. Finally, with the decay of the scalar source, the system gradually settles down to a new state. Interestingly, the final state of the evolution could be a hairy black hole with less scalar condensation, a RN-AdS black hole or a Schwarzschild-AdS black hole, which depends on the quench strength. However, as long as the quench strength is large enough, the system always loses all the electric charge and converges to the Schwarzschild-AdS black hole.

hep-th↗

arXiv2022

Structure-Unified M-Tree Coding Solver for MathWord Problem

As one of the challenging NLP tasks, designing math word problem (MWP) solvers has attracted increasing research attention for the past few years. In previous work, models designed by taking into account the properties of the binary tree structure of mathematical expressions at the output side have achieved better performance. However, the expressions corresponding to a MWP are often diverse (e.g., $n_1+n_2 \times n_3-n_4$, $n_3\times n_2-n_4+n_1$, etc.), and so are the corresponding binary trees, which creates difficulties in model learning due to the non-deterministic output space. In this paper, we propose the Structure-Unified M-Tree Coding Solver (SUMC-Solver), which applies a tree with any M branches (M-tree) to unify the output structures. To learn the M-tree, we use a mapping to convert the M-tree into the M-tree codes, where codes store the information of the paths from tree root to leaf nodes and the information of leaf nodes themselves, and then devise a Sequence-to-Code (seq2code) model to generate the codes. Experimental results on the widely used MAWPS and Math23K datasets have demonstrated that SUMC-Solver not only outperforms several state-of-the-art models under similar experimental settings but also performs much better under low-resource conditions.

cs.CL↗

arXiv2022

Computing Medial Axis Transform with Feature Preservation via Restricted Power Diagram

We propose a novel framework for computing the medial axis transform of 3D shapes while preserving their medial features via restricted power diagram (RPD). Medial features, including external features such as the sharp edges and corners of the input mesh surface and internal features such as the seams and junctions of medial axis, are important shape descriptors both topologically and geometrically. However, existing medial axis approximation methods fail to capture and preserve them due to the fundamentally under-sampling in the vicinity of medial features, and the difficulty to build their correct connections. In this paper we use the RPD of medial spheres and its affiliated structures to help solve these challenges. The dual structure of RPD provides the connectivity of medial spheres. The surface restricted power cell (RPC) of each medial sphere provides the tangential surface regions that these spheres have contact with. The connected components (CC) of surface RPC give us the classification of each sphere, to be on a medial sheet, a seam, or a junction. They allow us to detect insufficient sphere sampling around medial features and develop necessary conditions to preserve them. Using this RPD-based framework, we are able to construct high quality medial meshes with features preserved. Compared with existing sampling-based or voxel-based methods, our method is the first one that can preserve not only external features but also internal features of medial axes.

cs.GR↗

arXiv2022

Analyzing and Evaluating Faithfulness in Dialogue Summarization

Dialogue summarization is abstractive in nature, making it suffer from factual errors. The factual correctness of summaries has the highest priority before practical applications. Many efforts have been made to improve faithfulness in text summarization. However, there is a lack of systematic study on dialogue summarization systems. In this work, we first perform the fine-grained human analysis on the faithfulness of dialogue summaries and observe that over 35% of generated summaries are faithfully inconsistent respective the source dialogues. Furthermore, we present a new model-level faithfulness evaluation method. It examines generation models with multi-choice questions created by rule-based transformations. Experimental results show that our evaluation schema is a strong proxy for the factual correctness of summarization models. The human-annotated faithfulness samples and the evaluation toolkit are released to facilitate future research toward faithful dialogue summarization.

cs.CL↗

arXiv2022

Energy Circuit-based Integrated Energy Management System: Theory, Implementation, and Application

Integrated energy systems (IESs), in which various energy flows are interconnected and coordinated to release potential flexibility for more efficient and secure operation, have drawn increasing attention in recent years. In this article, an integrated energy management system (IEMS) that performs online analysis and optimization on coupling energy flows in an IES is comprehensively introduced. From the theory perspective, an energy circuit method (ECM) that models natural gas networks and heating networks in the frequency domain is discussed. This method extends the electric circuit modeling of power systems to IESs and enables the IEMS to manage large-scale IESs. From the implementation perspective, the architecture design and function development of the IEMS are presented. Tutorial examples with illustrative case studies are provided to demonstrate its functions of dynamic state estimation, energy flow analysis, security assessment and control, and optimal energy flow. From the application perspective, real-world engineering demonstrations that apply IEMSs in managing building-scale, park-scale, and city-scale IESs are reported. The economic and environmental benefits obtained in these demonstration projects indicate that the IEMS has broad application prospects for a low/zero-carbon future energy system.

eess.SY↗

arXiv2022

Dirac observables in the 4-dimensional phase space of Ashtekar's variables and spherically symmetric loop quantum black holes

In this paper, we study a proposal put forward recently by Bodendorfer, Mele and Münch and Garc\'ıa-Quismondo and Marugán, in which the two polymerization parameters of spherically symmetric black hole spacetimes are the Dirac observables of the four-dimensional Ashtekar's variables. In this model, black and white hole horizons in general exist and naturally divide the spacetime into the external and internal regions. In the external region, the spacetime can be made asymptotically flat by properly choosing the dependence of the two polymerization parameters on the Ashtekar variables. Then, we find that the asymptotical behavior of the spacetime is universal, and, to the leading order, the curvature invariants are independent of the mass parameter $m$. For example, the Kretschmann scalar approaches zero as $K \simeq A_0r^{-4}$ asymptotically, where $A_0$ is generally a non-zero constant and independent of $m$, and $r$ the geometric radius of the two-spheres. In the internal region, all the physical quantities are finite, and the Schwarzschild black hole singularity is replaced by a transition surface whose radius is always finite and non-zero. The quantum gravitational effects are negligible near the black hole horizon for very massive black holes. However, the behavior of the spacetime across the transition surface is significantly different from all loop quantum black holes studied so far. In particular, the location of the maximum amplitude of the curvature scalars is displaced from the transition surface and depends on $m$, so does the maximum amplitude. In addition, the radius of the white hole is much smaller than that of the black hole, and its exact value sensitively depends on $m$, too.

gr-qc↗

arXiv2022

On the late-time tails of massive perturbations in spherically symmetric black holes

It was first pointed out by Koyama and Tomimatsu that, under reasonable assumptions, the asymptotic late-time tails of massive scalar perturbations in the far zone of spherically symmetric black hole spacetimes decays universally as $t^{-5/6}$. The late-time tail is furnished by the contribution from the branch cut of the frequency-domain Green's function, which is constructed in terms of two appropriate solutions of the corresponding homogeneous equation. The present study is focused on some particular forms of the in-going wave that were not explicitly considered in the original derivations but nonetheless have been taken into account in the literature by other authors. In this regard, we reassess the authors' arguments and provide a detailed complimentary analysis that covers a few specific aspects. For some particular cases, the tail is found to possess the form $t^{-1}$. We also discuss the possible implications of the present findings.

gr-qc↗

arXiv2022

Scene Text Image Super-Resolution via Content Perceptual Loss and Criss-Cross Transformer Blocks

Text image super-resolution is a unique and important task to enhance readability of text images to humans. It is widely used as pre-processing in scene text recognition. However, due to the complex degradation in natural scenes, recovering high-resolution texts from the low-resolution inputs is ambiguous and challenging. Existing methods mainly leverage deep neural networks trained with pixel-wise losses designed for natural image reconstruction, which ignore the unique character characteristics of texts. A few works proposed content-based losses. However, they only focus on text recognizers' accuracy, while the reconstructed images may still be ambiguous to humans. Further, they often have weak generalizability to handle cross languages. To this end, we present TATSR, a Text-Aware Text Super-Resolution framework, which effectively learns the unique text characteristics using Criss-Cross Transformer Blocks (CCTBs) and a novel Content Perceptual (CP) Loss. The CCTB extracts vertical and horizontal content information from text images by two orthogonal transformers, respectively. The CP Loss supervises the text reconstruction with content semantics by multi-scale text recognition features, which effectively incorporates content awareness into the framework. Extensive experiments on various language datasets demonstrate that TATSR outperforms state-of-the-art methods in terms of both recognition accuracy and human perception.

cs.CV↗

arXiv2022

Time exponential integrator Fourier pseudospectral methods with high accuracy and multiple conservation laws for three-dimensional Maxwell's equations

Maxwell equations describe the propagation of electromagnetic waves and are therefore fundamental to understanding many problems encountered in the study of antennas and electromagnetics. The aim of this paper is to propose and analyse an efficient fully discrete scheme for solving three-dimensional Maxwell's equations. This is accomplished by combining time exponential integrator and Fourier pseudospectral methods. Fast computation is implemented in the scheme by using the Fast Fourier Transform algorithm which is well known in scientific computations. An optimal error estimate which is not encumbered by the CFL condition is established and the resulting scheme is proved to be of spectral accuracy in space and infinite-order accuracy in time. Furthermore, the scheme is shown to have multiple conservation laws including discrete energy, helicity, momentum, symplecticity, and divergence-free field conservations. All the theoretical results of the accuracy and conservations are numerically illustrated by two numerical tests.

math.NA↗