arXiv subjects

Bo Wang

Publications and source records attributed to Bo Wang.

At least 631 records · Page 35Linked to original sources

arXiv2020

Fast multipole method for 3-D Laplace equation in layered media

In this paper, a fast multipole method (FMM) is proposed for 3-D Laplace equation in layered media. The potential due to charges embedded in layered media is decomposed into a free space component and four types of reaction field components, and the latter can be associated with the potential of a polarization source defined for each type. New multipole expansions (MEs) and local expansions (LEs), as well as the multipole to local (M2L) translation operators are derived for the reaction components, based on which the FMMs for reaction components are then proposed. The resulting FMM for charge interactions in layered media is a combination of using the classic FMM for the free space components and the new FMMs for the reaction field components. With the help of a recurrence formula for the run-time computation of the Sommerfeld-type integrals used in M2L translation operators, pre-computations of a large number of tables are avoided. The new FMMs for the reaction components are found to be much faster than the classic FMM for the free space components due to the separation of equivalent polarization charges and the associated target charges by a material interface. As a result, the FMM for potential in layered media costs almost the same as the classic FMM in the free space case. Numerical results validate the fast convergence of the MEs for the reaction components, and the O(N) complexity of the FMM with a given truncation number p for charge interactions in 3-D layered media.

math.NA↗

arXiv2020

$D^{(\ast)}N$ interaction and the structure of $Σ_c(2800)$ and $Λ_c(2940)$ in chiral effective field theory

We study the $DN$ and $D^\ast N$ interactions to probe the inner structure of $Σ_c(2800)$ and $Λ_c(2940)$ with the chiral effective field theory to the next-to-leading order. We consider the contact term, one-pion-exchange and two-pion-exchange contributions to characterize the short-, long- and mid-range interactions of the $D^{(\ast)}N$ systems. The low energy constants of the $D^{(\ast)}N$ systems are related to those of the $N\bar{N}$ interaction with quark level Lagrangian that inspired by the resonance saturation model. The $Δ(1232)$ degree of freedom is also included in the loop diagrams. The attractive potential in the $[DN]_{J=1/2}^{I=1}$ channel is too weak to form bound state, which indicates the explanation of $Σ_c(2800)$ as the compact charmed baryon is more reasonable. Meanwhile, the potentials of the isoscalar channels are deep enough to yield the molecular states. We obtain the masses of the $[DN]_{J=1/2}^{I=0}$, $[D^\ast N]_{J=1/2}^{I=0}$ and $[D^\ast N]_{J=3/2}^{I=0}$ systems to be $2792.0$ MeV, $2943.6$ MeV and $2938.4$ MeV, respectively. The $Λ_c(2940)$ is probably the isoscalar $D^\ast N$ molecule considering its low mass puzzle. Besides, the $Λ_c(2940)$ signal might contain the spin-$\frac{1}{2}$ and spin-$\frac{3}{2}$ two structures, which can qualitatively explain the significant decay ratio to $D^0p$ and $Σ_cπ$. We also study the $\bar{B}^{(\ast)}N$ systems and predict the possible molecular states in the isoscalar channels. We hope experimentalists could hunt for the open charmed molecular pentaquarks in the $Λ_c^+π^+π^-$ final state.

hep-ph↗

arXiv2020

A Novel CNN-based Method for Accurate Ship Detection in HR Optical Remote Sensing Images via Rotated Bounding Box

Currently, reliable and accurate ship detection in optical remote sensing images is still challenging. Even the state-of-the-art convolutional neural network (CNN) based methods cannot obtain very satisfactory results. To more accurately locate the ships in diverse orientations, some recent methods conduct the detection via the rotated bounding box. However, it further increases the difficulty of detection, because an additional variable of ship orientation must be accurately predicted in the algorithm. In this paper, a novel CNN-based ship detection method is proposed, by overcoming some common deficiencies of current CNN-based methods in ship detection. Specifically, to generate rotated region proposals, current methods have to predefine multi-oriented anchors, and predict all unknown variables together in one regression process, limiting the quality of overall prediction. By contrast, we are able to predict the orientation and other variables independently, and yet more effectively, with a novel dual-branch regression network, based on the observation that the ship targets are nearly rotation-invariant in remote sensing images. Next, a shape-adaptive pooling method is proposed, to overcome the limitation of typical regular ROI-pooling in extracting the features of the ships with various aspect ratios. Furthermore, we propose to incorporate multilevel features via the spatially-variant adaptive pooling. This novel approach, called multilevel adaptive pooling, leads to a compact feature representation more qualified for the simultaneous ship classification and localization. Finally, detailed ablation study performed on the proposed approaches is provided, along with some useful insights. Experimental results demonstrate the great superiority of the proposed method in ship detection.

cs.CV↗

arXiv2020

The formation of single neutron-stars from double white-dwarf mergers via accretion-induced collapse

The merging of double white dwarfs (WDs) may produce the events of accretion-induced collapse (AIC) and form single neutron stars (NSs). Meanwhile, it is also notable that the recently proposed WD+He subgiant scenario has a significant contribution to the production of massive double WDs, in which the primary WD grows in mass by accreting He-rich material from a He subgiant companion. In this work, we aim to study the binary population synthesis (BPS) properties of AIC events from the double WD mergers by considering the classical scenarios and also the contribution of theWD+He subgiant scenario to the formation of double WDs. First, we provided a dense and large model grid of WD+He star systems for producing AIC events through the double WD merger scenario. Secondly, we performed several sets of BPS calculations to obtain the rates and single NS number in our Galaxy. We found that the rates of AIC events from the double WD mergers in the Galaxy are in the range of 1.4-8.9*10^-3 yr^-1 for all ONe/CO WD+ONe/CO WD mergers, and in the range of 0.3-3.8*10^-3 yr^-1 when double COWD mergers are not considered.We also found that the number of single NSs from AIC events in our Galaxy may range from 0.328*10^7 to 1.072*10^8. The chirp mass of double WDs for producing AIC events distribute in the range of 0.55-1.25 Msun. We estimated that more than half of doubleWDs for producing AIC events are capable to be observed by the future space-based gravitational wave detectors.

astro-ph.SR↗

arXiv2020

The formation of type Ia supernovae from carbon-oxygen-silicon white dwarfs

The carbon-oxygen white dwarf (CO WD) + He star channel has been thought to be one of the promising scnarios to produce young type Ia supernovae (SNe Ia). Previous studies found that if the mass-accretion rate is greater than a critical value, the He-accreting CO WD will undergo inwardly propagating (off-centre) carbon ignition when it increases its mass close to the Chandrasekhar limit. The inwardly propagating carbon flame was supposed to reach the centre by previous works, leading to the production of an oxygen-neon (ONe) WD that may collapse into a neutron star but not an SN Ia. However, it is still uncertain how the carbon flame propagates under the effect of mixing mechanisms. In the present work, we aim to investigate the off-centre carbon burning of the He-accreting CO WDs by considering the effect of convective mixing. We found that the temperature of the flame is high enough to burn the carbon into silicon-group elements in the outer part of the CO core even if the convective overshooting is considered, but the flame would quench somewhere inside the WD, resulting in the formation of a C-O-Si WD. Owing to the inefficiency of thermohaline mixing, the C-O-Si WD may explode as an SN Ia if it continues to grow in mass. Our radiation transfer simulations show that the SN ejecta with the silicon-rich outer layer will form high-velocity absorption lines in Si II, leading to some similarities to a class of the high-velocity SNe Ia in the spectral evolution. We estimate that the birthrate of SNe Ia with Si-rich envelope is ~ 10^(-4)/yr in our galaxy.

astro-ph.HE↗

arXiv2020

COVID-19 Chest CT Image Segmentation -- A Deep Convolutional Neural Network Solution

A novel coronavirus disease 2019 (COVID-19) was detected and has spread rapidly across various countries around the world since the end of the year 2019, Computed Tomography (CT) images have been used as a crucial alternative to the time-consuming RT-PCR test. However, pure manual segmentation of CT images faces a serious challenge with the increase of suspected cases, resulting in urgent requirements for accurate and automatic segmentation of COVID-19 infections. Unfortunately, since the imaging characteristics of the COVID-19 infection are diverse and similar to the backgrounds, existing medical image segmentation methods cannot achieve satisfactory performance. In this work, we try to establish a new deep convolutional neural network tailored for segmenting the chest CT images with COVID-19 infections. We firstly maintain a large and new chest CT image dataset consisting of 165,667 annotated chest CT images from 861 patients with confirmed COVID-19. Inspired by the observation that the boundary of the infected lung can be enhanced by adjusting the global intensity, in the proposed deep CNN, we introduce a feature variation block which adaptively adjusts the global properties of the features for segmenting COVID-19 infection. The proposed FV block can enhance the capability of feature representation effectively and adaptively for diverse cases. We fuse features at different scales by proposing Progressive Atrous Spatial Pyramid Pooling to handle the sophisticated infection areas with diverse appearance and shapes. We conducted experiments on the data collected in China and Germany and show that the proposed deep CNN can produce impressive performance effectively.

eess.IV↗

arXiv2020

Forecast for weighing neutrinos in cosmology with SKA

We investigate what role the SKA neutral hydrogen (HI) intensity mapping (IM) sky survey observation will play in weighing neutrinos in cosmology. We use the simulated data of the baryon acoustic oscillation (BAO) measurements from the HI IM survey based on SKA1 and SKA2 to do the analysis. For the current observations, we use the Planck 2015 cosmic microwave background (CMB) anisotropies observation, the optical BAO measurements, the type Ia supernovae (SN) observation (Pantheon compilation), and the latest $H_0$ measurement. We consider three mass ordering cases for massive neutrinos, i.e., the normal hierarchy (NH), inverted hierarchy (IH), and degenerate hierarchy (DH) cases. It is found that the SKA observation can significantly improve the constraints on $Ω_{\rm m}$ and $H_0$. Compared to the current observation, the SKA1 data can improve the constraints on $Ω_{\rm m}$ by about 33%, and on $H_0$ by about 36%; the SKA2 data can improve the constraints on $Ω_{\rm m}$ by about 58%, and on $H_0$ by about 66%. It is also found that the SKA observation can only slightly improve the constraints on $\sum m_ν$. Compared to the current observation, the SKA1 data can improve the constraints on $\sum m_ν$ by about 4%, 3%, and 10%, for the NH, IH, and DH cases, respectively; the SKA2 data can improve the constraints on $\sum m_ν$ by about 7%, 7%, and 16%, for the NH, IH, and DH cases, respectively.

astro-ph.CO↗

arXiv2020

3D Human Pose Estimation using Spatio-Temporal Networks with Explicit Occlusion Training

Estimating 3D poses from a monocular video is still a challenging task, despite the significant progress that has been made in recent years. Generally, the performance of existing methods drops when the target person is too small/large, or the motion is too fast/slow relative to the scale and speed of the training data. Moreover, to our knowledge, many of these methods are not designed or trained under severe occlusion explicitly, making their performance on handling occlusion compromised. Addressing these problems, we introduce a spatio-temporal network for robust 3D human pose estimation. As humans in videos may appear in different scales and have various motion speeds, we apply multi-scale spatial features for 2D joints or keypoints prediction in each individual frame, and multi-stride temporal convolutional net-works (TCNs) to estimate 3D joints or keypoints. Furthermore, we design a spatio-temporal discriminator based on body structures as well as limb motions to assess whether the predicted pose forms a valid pose and a valid movement. During training, we explicitly mask out some keypoints to simulate various occlusion cases, from minor to severe occlusion, so that our network can learn better and becomes robust to various degrees of occlusion. As there are limited 3D ground-truth data, we further utilize 2D video data to inject a semi-supervised learning capability to our network. Experiments on public datasets validate the effectiveness of our method, and our ablation studies show the strengths of our networkś individual submodules.

cs.CV↗

arXiv2020

SAUNet: Shape Attentive U-Net for Interpretable Medical Image Segmentation

Medical image segmentation is a difficult but important task for many clinical operations such as cardiac bi-ventricular volume estimation. More recently, there has been a shift to utilizing deep learning and fully convolutional neural networks (CNNs) to perform image segmentation that has yielded state-of-the-art results in many public benchmark datasets. Despite the progress of deep learning in medical image segmentation, standard CNNs are still not fully adopted in clinical settings as they lack robustness and interpretability. Shapes are generally more meaningful features than solely textures of images, which are features regular CNNs learn, causing a lack of robustness. Likewise, previous works surrounding model interpretability have been focused on post hoc gradient-based saliency methods. However, gradient-based saliency methods typically require additional computations post hoc and have been shown to be unreliable for interpretability. Thus, we present a new architecture called Shape Attentive U-Net (SAUNet) which focuses on model interpretability and robustness. The proposed architecture attempts to address these limitations by the use of a secondary shape stream that captures rich shape-dependent information in parallel with the regular texture stream. Furthermore, we suggest multi-resolution saliency maps can be learned using our dual-attention decoder module which allows for multi-level interpretability and mitigates the need for additional computations post hoc. Our method also achieves state-of-the-art results on the two large public cardiac MRI image segmentation datasets of SUN09 and AC17.

eess.IV↗

arXiv2020

The importance of transparency and reproducibility in artificial intelligence research

In their study, McKinney et al. showed the high potential of artificial intelligence for breast cancer screening. However, the lack of detailed methods and computer code undermines its scientific value. We identify obstacles hindering transparent and reproducible AI research as faced by McKinney et al and provide solutions with implications for the broader field.

stat.AP↗

arXiv2020

Adiabatic regularization and Green's function of a scalar field in de Sitter space: Positive energy spectrum and no trace anomaly

In the conventional adiabatic regularization the vacuum ultraviolet divergences of a quantum field in curved spacetime are removed by subtracting the $k$-mode of the stress tensor to the 4th-order. For a scalar field in de Sitter space, we find that the 4th-order regularized spectral energy density is negative. Moreover, the 2nd-order regularization for minimal coupling ($ξ=0$) and the 0th-order regularization for conformal coupling ($ξ=\frac16$) yield a positive and UV-convergent spectral energy density and power spectrum. The regularized stress tensor in the vacuum is maximally symmetric and can drive inflation, while its $k$-modes representing the primordial fluctuations are nonuniformly distributed. Conventional regularization of a Green's function in position space is generally plagued by a log IR divergence. Only in the massless case with $ξ=0$ or $\frac16$, we can directly regularize the Green's functions and obtain vanishing results that agree with the adiabatic regularization results. In this case, the regularized power spectrum and stress tensor are both zero, and no trace anomaly exists. To overcome the log IR divergence problem in the massive cases with $ξ=0$ and $\frac16$, we perform Fourier transformation of the regularized power spectra and obtain the regularized analytical Green's functions which are IR- and UV-convergent.

gr-qc↗

arXiv2020

Mass transfer of low-mass binaries and chemical anomalies among unevolved stars in globular clusters

While it is well known that mass transfer in binaries can pollute the surfaces of the accretors, it is still unclear whether this mechanism can reproduce the observed chemical inhomogeneities in globular clusters. We study the surface abundances of the accretors in low-mass binaries, as a first step towards understanding whether mass transfer in low-mass binaries is one of the potential origins of the aforementioned abundance anomalies in globular clusters. We use the Modules for Experiments in Stellar Astrophysics code to calculate binary evolutionary models with different initial donor masses between 0.9 and 1.9 $ \rm {M}_\odot$ for an initial metallicity of Z = 0.0034. The results show that in some low-mass binary systems, the accretors exhibit peculiar chemical patterns when they are still unevolved stars, e.g. C and O depletion; Na and N enhancement; constant Mg, Al and C+N+O. The abundance patterns of the accretors are significantly different from their initial abundances (or that of normal single stars), and can match the observed populations. These abundance patterns strongly depend not only on the initial parameters of binaries (donor mass, mass ratio, and orbital period), but also on the assumptions regarding mass-transfer efficiency and angular momentum loss. These results support the hypothesis that mass transfer in low-mass binaries is at least partly responsible for the unevolved anomalous stars in globular clusters. More work on binary evolutionary models and binary population synthesis is required to fully evaluate the contribution of this scenario.

astro-ph.SR↗

arXiv2020

Vector exceptional points with strong superchiral fields

Exceptional points(EPs), branch points of complex energy surfaces at which eigenvalues and eigenvectors coalesce, are ubiquitous in non-Hermitian systems. Many novel properties and applications have been proposed around the EPs. One of the important applications is to enhance the detection sensitivity. However, due to the lack of single-handed superchiral fields, all of the proposed EP-based sensing mechanisms are only useful for the non-chiral discrimination. Here, we propose theoretically and demonstrate experimentally a new type of EP, which is a called radiation vector EP, to fulfill the homogeneous superchiral fields for chiral sensing. This type of EP is realized by suitably tuning the coupling strength and radiation losses for a pair of orthogonal polarization modes in the photonic crystal slab. Based on the unique modal-coupling property at the vector EP, we demonstrate that the uniform superchiral fields can be generated with two beams of lights illuminating on the photonic crystal slab from opposite directions. Thus, the designed photonic crystal slab, which supports the vector EP, can be used to perform surface-enhanced chiral detection. Our findings provide a new strategy for ultrasensitive characterization and quantification of molecular chirality, a key aspect for various bioscience and biomedicine applications.

physics.optics↗

arXiv2020

Constraining Screened Modified Gravity by Space-borne Gravitational-wave Detectors

The screened modified gravity (SMG) is a unified theoretical framework, which describes the scalar-tensor gravity with screening mechanism. Based on the gravitational-wave (GW) waveform derived in our previous work \citep{liu2018waveforms}, in this article we investigate the potential constraints on SMG theory through the GW observation of the future space-borne GW detectors, including LISA, TianQin and Taiji. We find that, for the EMRIs consisting of a massive black hole and a neutron star, if the EMRIs are at Virgo cluster, the GW signals can be detected by the detectors at quite high significant level, and the screened parameter $ε_{\rm NS}$ can be constrained at about $\mathcal{O}(10^{-5})$, which is more than one order of magnitude tighter than the potential constraint given by ground-based Einstein telescope. However, for the EMRIs consisting of a massive black hole and a white dwarf, it is more difficult to be detected than the previous case. For the specific SMG models, including chameleon, symmetron and dilaton, we find these constraints are complementary with that from Cassini experiment, but weaker than those from lunar laser ranging observations and binary pulsars, due to the strong gravitational potentials on the surface of neutron stars. By analyzing the deviation of GW waveform in SMG from that in general relativity, as anticipated, we find the dominant contribution of the SMG constraining comes from the correction terms in the GW phases, rather than the extra polarization modes or the correction terms in the GW amplitudes.

gr-qc↗

arXiv2020

Fast Multipole Method for 3-D Linearized Poisson-Boltzmann Equation in Layered Media

In this paper, we propose a fast multipole method (FMM) for 3-D linearized Poisson-Boltzmann (PB) equation in layered media. The main framework of the algorithm is analogous to the FMM for Helmholtz and Laplace equation in layered media [1,2], using an extension of the Funk-Hecke formula for pure imaginary wave number. Moreover, a recurrence formula is provided for the run-time computation of the Sommerfeld-type integrals used in the FMM algorithm. Due to the similarity between Helmholtz and linearized PB equation, the recurrence formula can also be used for the FMM of Helmholtz equation in layered media with minor changes as mentioned in [1] . Numerical results validate that the FMM for interactions of charges under screen's potentials in layered media has the same accuracy and CPU complexity as the classic FMM for charge interactions in free space.

math.NA↗

arXiv2020

Spectrum of the strange hidden charm molecular pentaquarks in chiral effective field theory

We calculate the effective potentials of the $Ξ_c\bar{D}^{(\ast)}$, $Ξ_c^\prime\bar{D}^{(\ast)}$ and $Ξ_c^\ast\bar{D}^{(\ast)}$ systems with the chiral effective field theory up to the next-to-leading order. We simultaneously consider the short-, intermediate- and long-range interactions. With the newly observed $P_c$ spectra as inputs, we construct the quark-level contact Lagrangians to relate the low energy constants to those of $Σ_c\bar{D}^{(\ast)}$ with the help of quark model. Our calculation indicates there are seven bound states in the $I=0$ strange hidden charm $[Ξ_c^\prime\bar{D}^{(\ast)}]_J~(J=\frac{1}{2},\frac{3}{2})$ and $[Ξ_c^\ast\bar{D}^{(\ast)}]_J~(J=\frac{1}{2},\frac{3}{2},\frac{5}{2})$ systems. Our analyses also disfavor the $Λ_c\bar{D}^{(\ast)}$ bound states. However, we obtain three new hadronic molecules in the isoscalar $[Ξ_c\bar{D}^{(\ast)}]_J~(J=\frac{1}{2},\frac{3}{2})$ systems. The masses of $[Ξ_c\bar{D}]_{1/2}$, $[Ξ_c\bar{D}^{\ast}]_{1/2}$ and $[Ξ_c\bar{D}^{\ast}]_{3/2}$ are predicted to be $4319.4^{+2.8}_{-3.0}$ MeV, $4456.9^{+3.2}_{-3.3}$ MeV and $4463.0^{+2.8}_{-3.0}$ MeV, respectively. We also notice the one-eta-exchange influence is rather feeble. Binding solutions in the $I=1$ channels are nonexistent. We hope the future analyses at LHCb can seek for these new $P_{cs}$s in the $JψΛ$ final states, especially near the thresholds of $Ξ_c\bar{D}^{(\ast)}$.

hep-ph↗

arXiv2020

Niobium doping induced mirror twin boundaries in MBE grown WSe2 monolayers

Mirror twin boundary (MTB) brings unique 1D physics and properties into two-dimensional transition metal dichalcogenides (TMDCs), but they were rarely observed in non-Mo-based TMDCs. Herein, by post-growth Nb doping, high density 4|4E-W and 4|4P-Se MTBs were introduced into molecular beam epitaxy (MBE) grown WSe2 monolayers. Of them, 4|4E-W MTB with a novel structure was discovered experimentally for the first time, while 4|4P-Se MTBs present a random permutations of W and Nb, forming a 1D alloy system. Comparison between the doped and non-doped WSe2 confirmed that Nb dopants are essential for MTB formation. Furthermore, quantitative statistics reveal the areal density of MTBs is directly proportional to the concentration of Nb dopants. To unravel the injection pathway of Nb dopants, first-principles calculations about a set of formation energies for excess Nb atoms with different configurations were conducted, based on which a model explaining the origin of MTBs introduced by excess metal was built. We conclude that the formation of MTBs is mainly driven by the collective evolution of excess Nb atoms introduced into the lattice of host WSe2 crystal and subsequent displacement of metal atoms (W or Nb). This study provides a novel way to tailor the MTBs in 2D TMDC materials via proper metal doping and presents a new opportunities for exploring the intriguing properties.

cond-mat.mtrl-sci↗

arXiv2019

Diversity Transfer Network for Few-Shot Learning

Few-shot learning is a challenging task that aims at training a classifier for unseen classes with only a few training examples. The main difficulty of few-shot learning lies in the lack of intra-class diversity within insufficient training samples. To alleviate this problem, we propose a novel generative framework, Diversity Transfer Network (DTN), that learns to transfer latent diversities from known categories and composite them with support features to generate diverse samples for novel categories in feature space. The learning problem of the sample generation (i.e., diversity transfer) is solved via minimizing an effective meta-classification loss in a single-stage network, instead of the generative loss in previous works. Besides, an organized auxiliary task co-training over known categories is proposed to stabilize the meta-training process of DTN. We perform extensive experiments and ablation studies on three datasets, i.e., \emph{mini}ImageNet, CIFAR100 and CUB. The results show that DTN, with single-stage training and faster convergence speed, obtains the state-of-the-art results among the feature generation based few-shot learning methods. Code and supplementary material are available at: \texttt{https://github.com/Yuxin-CV/DTN}

cs.CV↗