arXiv subjects

Ping Li

Publications and source records attributed to Ping Li.

At least 19 recordsLinked to original sources

arXiv2026

RGDM: Robust Graph Diffusion Model

Diffusion models represent a powerful class of generative models known for their solid theoretical foundations and remarkable performance across diverse tasks and domains. Existing node level diffusion methods synthesize nodes and edges within a graph, but are not designed to learn a robust generator from adversarially corrupted graphs. In this study, we introduce the Robust Graph Diffusion Model (RGDM), designed to generate labeled synthetic graph structures consisting of nodes and edges that integrate seamlessly into a given graph. The RGDM consists of a Robust Graph Autoencoder (RGAE) and a Latent Diffusion Model (LDM). Leveraging an edge selection mechanism and an innovative low-rank regularization on the latent feature, the RGDM produces clean and high-quality synthetic graph structures, even when trained on graphs subject to adversarial attacks. Comprehensive experimental evaluations reveal that Graph Neural Networks (GNNs) trained on the augmented graph, which is formed by merging the original attacked graph with the synthetic graph structures, exhibit significantly improved robustness against various graph adversarial attacks in the context of semi-supervised node classification. The code of the RGDM is available at https://github.com/Statistical-Deep-Learning/RGDM.

cs.LG↗

arXiv2026

The Survival Paradox of Close-In Planets Orbiting Red Clump Giants

Close-in planets orbiting core-helium-burning red-clump (RC) giants directly challenge tidal-engulfment theory, which predicts that planets within about 1 au of Sun-like stars should be engulfed during red-giant-branch (RGB) expansion. The existence of such survivors would imply missing physical processes in the late stages of stellar and planetary evolution. Yet only one such "forbidden" planet had been identified to date. Here we report three systems-24 Boo b, HD 12648 b, and TYC 4282-605-1 b-whose host stars are confirmed red-clump giants through asteroseismology, showing that each has already undergone dramatic expansion and contraction following helium ignition. Their planets now reside at orbital separations smaller than the maximum stellar radii attained during the RGB phase, implying that they should not have survived. Among them, 24 Boo b orbits at only 0.183 au with a 30-day period, making it the closest planet known around a red-clump giant. These discoveries demonstrate that this paradox is not an isolated anomaly but signals the emergence of a new class of evolved planetary systems. Their survival defies current models of stellar-planetary coevolution and points to unexplored physics shaping planetary architectures around post-RGB stars.

astro-ph.EP↗

arXiv2026

Sharp Generalization for Nonparametric Regression in Interpolation Space by Shallow Over-Parameterized Neural Networks with Channel Attention

We study nonparametric regression using an over-parameterized two-layer ReLU neural network with fixed covariance-based channel attention, trained by gradient descent with both layers updated simultaneously. The spherical covariates follow an arbitrary fixed non-atomic distribution, and the target function lies in an interpolation space with spectral bias widely studied in the statistical learning literature. We prove that early-stopped training achieves regression risk $\cO(\eps_n^2)$, where $\eps_n$ is the critical population rate of the attention kernel $\Kattn$. Its positive eigenvalues are $λ_j^{s'}$, where $λ_j$ are those of the ReLU feature covariance kernel $K$ and $s'=2s+3$ for $s\ge1$. Under polynomial eigenvalue decay $λ_j\asymp j^{-2α}$ with $α>1/2$, the regression risk bound $\cO(\eps_n^2)$ specializes to the minimax optimal rate $\cO(n^{-2αs'/(2αs'+1)})$. Our analysis is based on two new technical results on the optimization and generalization of the attention network. First, the neural network function after every gradient descent iteration admits a decomposition into a function with bounded Reproducing Kernel Hilbert Space (RKHS) norm and a uniformly small residual. Second, local Rademacher complexity converts this decomposition into a regression risk bound. Kernel gradient descent on the same ball attains the same rate, while our theorem gives explicit sufficient width and auxiliary-sample conditions for simultaneous network training. Our analysis is further extended to covariates supported on $\RR^d$ through a novel normalization method. Under a sub-Gaussian covariate distribution, the attention network achieves a nearly optimal regression risk $\cO((\log n)n^{-s'(d+3)/(s'(d+3)+d)})$ without a density assumption.

stat.ML↗

arXiv2026

Eclipse Properties and Superhump Evolution in the SU UMa-Type Dwarf Nova Z Cha

The advent of large-scale time-domain surveys provides both opportunities and challenges for understanding accretion disk evolution in cataclysmic variables (CVs). Using high-cadence photometry from the Transiting Exoplanet Survey Satellite (TESS), we investigate the eclipsing SU UMa-type dwarf nova Z Cha. Leveraging eclipses as a natural probe, we examine the evolution of the accretion disk through variations in eclipse depth, O--C of eclipse minima, and positive superhump (PSH) amplitude. During superoutbursts, all three quantities exhibit quasi-periodic modulations with a common period of $\sim$2 days, consistent with the precession period of an eccentric disk. We interpret these correlated variations as evidence of an eccentric, precessing disk: O--C traces the periodic shift of the system's brightness center, while eclipse depth and PSH amplitude vary with the orientation of the disk bulge relative to the line of sight. In quiescence (Sectors 13 and 93), PSHs with periods of $\sim$0.0762 days show linearly decreasing amplitudes and periods, indicating gradual shrinkage of the eccentric disk and a slowing precession. Remarkably, a coherent signal with a period of $\sim$0.0729~days ($ε^{-}\approx-0.02$) appears in the same quiescent intervals. This signal may represent negative superhumps (NSHs) coexisting with PSHs, although an orbital sideband of the PSH cannot presently be excluded with the available data. If confirmed as NSHs, their coexistence with PSHs would challenge the classical tilted-disk model, and could be explained by retrograde apsidal precession of an eccentric disk, where the inner disk precesses retrogradely (NSHs) and the outer disk progradely (PSHs); this interpretation remains to be tested by further observations.

astro-ph.SR↗

arXiv2026

Sharp Generalization for Nonparametric Regression in Interpolation Space by Over-Parameterized Neural Networks Trained with Preconditioned Gradient Descent and Early Stopping

We study nonparametric regression using an over-parameterized two-layer neural network trained with algorithmic guarantees in this paper. We consider the setting where the training features are drawn uniformly from the unit sphere in $\RR^d$, and the target function lies in an interpolation space commonly studied in statistical learning theory. We demonstrate that training the neural network with a novel Preconditioned Gradient Descent (PGD) algorithm, equipped with early stopping, achieves a sharp regression rate of $\cO(n^{-\frac{2αs'}{2αs'+1}})$, where $α$ parametrizes the polynomial eigenvalue decay of the regular Neural Tangent Kernel (NTK), $λ_j\asymp j^{-2α}$, and $2α=d/(d-1)$ under the spherical uniform design considered here. The target function is assumed to belong to the interpolation space $\bth{\cH_K}^{s'}$ with $s'\ge3$. This rate is even sharper than the currently known nearly-optimal rate of $\cO(n^{-\frac{2αs'}{2αs'+1}})\log^2(1/δ)$~\citep{Li2024-edr-general-domain}, where $n$ is the size of the training data and $δ\in (0,1)$ is a small probability. Our analysis is based on two key technical contributions. First, we present a principled decomposition of the network output at each PGD step into a function in the reproducing kernel Hilbert space (RKHS) of a newly induced integral kernel, and a residual function with small $L^{\infty}$-norm. Second, leveraging this decomposition, we apply local Rademacher complexity theory to tightly control the complexity of the function class comprising all the neural network functions obtained in the PGD iterates. Our results show how PGD changes the effective kernel in the reference training dynamics, inducing an integral kernel with lower complexity than the regular NTK arising from vanilla GD.

stat.ML↗

arXiv2026

Last-Iterate Stationarity of Muon with Finitely Many Newton-Schulz Iterations

Muon has achieved strong empirical performance in large-scale language-model training, motivating a better theoretical understanding of its practical implementation. Unlike the idealized polar-factor update, practical Muon uses a fixed number of quintic Newton--Schulz iterations. Existing analyses mainly establish average- or best-iterate stationarity guarantees, leaving last-iterate convergence less understood. We study last-iterate gradient convergence of practical finite-step Muon on smooth, lower-bounded nonconvex objectives. For deterministic gradients, we show that without input stabilization, Muon can cycle even on a smooth convex quadratic. With positive stabilization, we establish last-iterate gradient convergence for any fixed momentum coefficient below one and a sufficiently small constant learning rate, together with square-summability of gradients. For stochastic gradients, we first show that persistent bounded, unbiased noise can prevent convergence with fixed learning rates and momentum, even with positive stabilization. We then establish almost-sure and mean-square last-iterate gradient convergence by combining decreasing learning rates with increasing momentum under bounded-variance stochastic gradients. Finally, we construct a smooth counterexample showing that neither convergence theorem guarantees a pointwise polynomial rate. We further establish density-one convergence rates, guaranteeing polynomial gradient decay for all but a vanishing fraction of iterations.

cs.LG↗

arXiv2026

An end-to-end function for the photograph of a Schwarzschild black hole

We show that the photograph of a Schwarzschild black hole can be written in closed form. The image plane is embedded as a plane tangent to a sphere centred on the black hole, and the map between a point source and its image point, which we call the photograph function, is treated as the analytic object. Explicit expressions are obtained for sources on the celestial sphere at infinity and for sources in the equatorial plane, without solving for the intermediate null geodesics. The construction is exact, and it makes the multivalued branch structure, the relation to the photon sphere and to the shadow boundary, and the self-similar higher-order images directly readable.

gr-qc↗

arXiv2026

Wan-Alpha++: Animating a Single RGBA Asset into Controllable Transparent Video Effects

RGBA video generation is essential for creating compositable visual assets in filmmaking, game development, and digital content production. However, generating high-quality RGB appearance alongside temporally coherent transparency remains a significant challenge. Existing text-to-RGBA video methods often entangle RGB and alpha distributions and offer limited control over the precise appearance of a target asset. In addition, incorporating image conditioning typically suffers from the emergence-disappearance effects, which fail existing first-frame or first-and-last-frame conditions due to their fully transparent boundary frames. To address these issues, we present Wan-Alpha++, a unified framework that brings controllable RGBA video generation to both text- and image-to-video diffusion backbones through a shared RGBA latent space, extending our conference version Wan-Alpha from the text-conditioned setting to image-conditioned generation with an interior reference frame. Our method first learns shiftable RGB-alpha distributions while preserving the pretrained RGB prior. Specifically, we introduce a transparency-aware bidirectional diffusion objective to produce distinguishable RGBA latent representations, and a transparency-guided mean sampler that shifts the noise distribution to improve transparency generation and enable spatial controllability. Building upon this RGBA generation prior, we introduce temporal phase reparameterization for image-conditioned effect generation. By cyclically transforming a zero-positive-zero alpha pattern into a positive-zero-positive pattern, it relocates visible content to both temporal boundaries, facilitating effective appearance conditioning with limited RGBA data.

cs.CV↗

arXiv2026

Limits of Besov Norms on Quantum Euclidean Spaces

We extend the Bourgain--Brézis--Mironescu (BBM) and Maz'ya--Shaposhnikova (MS) limit formulae to higher-order Besov norms on quantum Euclidean spaces $\R^d_θ$, a fundamental class of noncompact noncommutative manifolds obtained by deformation quantization of $\R^d$. A central ingredient is a Banach realization of the homogeneous Sobolev endpoint that forms a compatible couple with $L_p(\R_θ^d)$, together with the equivalence \[ K\bigl(\x,\varepsilon^k; L_p(\R_θ^d),\dot W_{p,0}^k(\R_θ^d)\bigr) \approx ω_p^k(\x,\varepsilon). \] This framework yields a unified proof of both endpoint formulae through the endpoint continuity of real interpolation scales. The principal new ingredient at the MS endpoint is the sharp large-translation asymptotic \[ \lim_{|h|\to\infty}\|T_h\x-\x\|_p = 2^{1/p}\|\x\|_p, \qquad \x\in L_p(\R_θ^d),\quad 1\leq p<\infty, \] whose proof combines oscillatory decay of the non-pure terms in the tracial expansion with weighted spectral-measure arguments adapted to the semifinite trace.

math.FA↗

arXiv2026

Campanato spaces via quantum semigroups

In this paper, we continue to investigate Campanato spaces via semigroups on von Neumann algebras. One of the main results is their coincidence with Lipschitz spaces for {\it all} regularity indices $α> 0$ and for all {\it analytic} semigroups on von Neumann algebras {\it not necessarily being finite}; the desired self-improving property of Campanato spaces has also been verified. As a consequence, we demonstrate that the column Campanato spaces are isomorphic to the row ones for {\it all} $α> 0$. This not only removes several restrictions in the previous work \cite{HJ24} by the first two authors, but also extends the previous results to any analytic semigroup, and thus resolves several problems left open in \cite{HJ24}. Both the results and the proof are new even in the commutative setting.

math.OA↗

arXiv2026

Superoutbursts and Superhumps of Cataclysmic Variables observed with TESS

Superoutbursts and superhumps are characteristic signatures of SU UMa-type cataclysmic variables, yet the full diversity of superhump evolutionary behaviour remains poorly constrained owing to the limitations of ground-based photometric monitoring. Here we report a systematic analysis of 30 SU UMa-type dwarf novae observed with the Transiting Exoplanet Survey Satellite, covering 37 superoutburst events. We detect coherent superhump signals in 29 systems-including five first-time detections-and determine or refine orbital periods for 13 objects. The high-cadence light curves clearly resolve the canonical three-stage (A-B-C) superhump period evolution. Using Stage A periods, we derive dynamical mass ratios for 18 systems, consistent with established cataclysmic variable evolutionary tracks. The measured Stage B period derivatives match the range of values from ground-based campaigns, with prominent positive drifts concentrated in short-period, low-mass-ratio systems. Repeated superoutbursts yield consistent evolutionary patterns, demonstrating that superhump morphology is an intrinsic, repeatable property. Two systems break the standard template: RZ LMi shows an inverted stage sequence inconsistent with classical precession theory, while ASASSN-14kj exhibits no stage evolution but a coherent long-period modulation. Persistent superhumps during normal outbursts in three systems provide direct evidence that eccentric disk structures can survive beyond their parent superoutbursts, revealing a decoupling between thermal accretion and tidal eccentricity cycles. These results demonstrate the power of space-borne photometry for probing accretion disk dynamics and testing precession models across the cataclysmic variable parameter space.

astro-ph.SR↗

arXiv2026

Analysis of Electromagnetic Scattering from Semiconductor Nanostructures by Solving Coupled Volume Integral and Two-fluid Hydrodynamic Equations

Semiconductor-based plasmonic nanostructures support localized surface plasmon modes in the infrared region. Unlike metallic nanostructures, they support both free electrons and holes, requiring a two-fluid hydrodynamic Drude equation (HDE) to accurately capture spatial dispersion effects and low-frequency acoustic plasmon modes that cannot be described by single-fluid models. In this work, a volume integral equation (VIE)-based solver is proposed for the analysis of electromagnetic scattering from semiconductor nanostructures. The proposed approach couples the VIE, formulated in terms of the electric flux density and the free-electron and hole polarization currents, with the two-fluid HDE. The coupled system is discretized using a tetrahedral mesh and solved efficiently using a two-level iterative solver. In contrast to finite-element-based methods, the proposed VIE-based approach does not require domain-wide meshing and inherently satisfies the radiation condition, thereby eliminating artificial absorbing boundaries. Numerical results for InSb-type semiconductor nanostructures demonstrate the accuracy and efficiency of the proposed VIE-based solver and its ability to capture unique optical phenomena, such as acoustic plasmon resonances and the blueshift of localized surface plasmon resonances, that cannot be described by the single-fluid HDE or classical Drude-based models.

physics.optics↗

arXiv2026

TDFNet: Tri-projection Deformable Fusion Network for Panoramic Salient Object Detection

Recent years have witnessed the growing potential of panoramic salient object detection in robotic vision, virtual reality, and related applications. However, projecting spherical scenes onto 2D planes inevitably introduces geometric distortions, which fundamentally limit the effectiveness of existing projection-based methods. Specifically, Equirectangular Projection (ERP) suffers from severe polar stretching distortions, while cube map projection introduces discontinuities across cube-face boundaries, resulting in degraded feature discriminability and compromised geometric consistency. To address these limitations, we propose TDFNet, the first Tri-projection Deformable Fusion Network for panoramic salient object detection, exploiting complementary projection representations to alleviate geometric distortions and improve detection performance.Specifically, we design a cross-projection deformable attention (CDA) module that leverages spatial correspondences between different projections to construct geometry-aware sampling locations, guiding deformable attention for cross-projection contextual aggregation and enhancing robustness against projection-induced deformations. Furthermore, we introduce a latitude-guided fusion module, which utilizes spherical latitude priors to construct geometric confidence weights for adaptively balancing ERP and CMP features. Meanwhile, LGF incorporates distortion-reduced semantic references from Tangent Projection to achieve cross-projection feature refinement and spatial alignment.By constructing a three-branch encoding architecture based on ERP, CMP, and Tangent Projection, TDFNet simultaneously preserves global spatial continuity, local geometric details, and fine-grained boundary information.

cs.CV↗

arXiv2026

Topological obstructions to geometric positivity and negativity on Calabi-Yau manifolds

We study whether the topology underlying a Calabi-Yau manifold can support natural geometric positivity or negativity structures. In even complex dimension $n \geq 4$ (assuming $b_2=1$ when $n \geq 6$), we strengthen a theorem of Oguiso-Peternell by proving that a Calabi-Yau manifold is not homeomorphic to a weak Fano $n$-fold. The same obstruction applies to Kähler manifolds with quasi-positive holomorphic sectional curvature. A transformation-group analogue excludes, in particular, symplectic manifolds admitting Hamiltonian circle actions with isolated fixed points. On the negative side, we show that the fundamental group of a Calabi-Yau manifold is not isomorphic to that of a Kähler hyperbolic manifold. Taken together, these results exhibit a common topological rigidity separating Calabi-Yau manifolds from several fundamental classes governed by geometric positivity or negativity.

math.AG↗

arXiv2026

Channel-wise Dynamic Knowledge Distillation via Adaptive Sample Generation for Action Recognition

Knowledge Distillation (KD) offers a promising yet underexplored path for compressing large action recognition models. However, existing KD methods suffer from two key limitations: 1) reliance on fixed input samples leads to suboptimal feature alignment between the frozen teacher (larger model) and the learnable student (smaller model), and 2) applying a uniform distillation strength for all channels fails to account for their varying importance in capturing distinct knowledge (e.g., motion tempo or magnitude) across training epochs. This motivates us to develop an Adaptive Sample-aware Channel-wise Dynamic (ASCD) KD approach, which operates in two stages. First, we use an adaptive sample generation module to create updated samples by incorporating semantics from sample gradients, which are derived by minimizing a feature loss weighted by channel centroid frequency differences at each layer. Meanwhile, crucial motion-related details are preserved by applying a Gaussian mask to frequency features. Second, we employ a channel-wise dynamic distillation module to train student on these generated samples, guided by sample gradients and feature frequencies. For efficiency, samples are updated periodically rather than per epoch. Extensive experiments on three video benchmarks (UCF101, Kinetics-400, Something-Something-v2) and two image datasets (CIFAR-100, ImageNet) demonstrate the state-of-the-art performance of our method. Code is available at https://github.com/mlvccn/ASCD_KD_Action.

cs.CV↗

arXiv2026

Coordinate Independence of the Schwarzschild Black Hole Accretion Vlasov Gas Model

This paper presents a detailed study of the coordinate dependence of Vlasov gas accretion onto a Schwarzschild black hole. Asymptotic results at infinity and near horizon are obtained via Taylor expansions for three different statistical distributions within the framework of the most general stationary spherically symmetric spacetime. Our findings demonstrate that the particle number density, energy density, radial and tangential pressures, and accretion rates are independent of the coordinate choice, even though individual components such as the particle current density and the stress-energy tensor explicitly depend on the coordinate system. Consequently, the accretion theory can be formulated without reference to any particular coordinate system. We also show that the mean energy of the accreted particles is $m_0+k_BT$, lower than the mean energy $m_0+\frac{3}{2}k_BT$ of the Maxwell-Boltzmann system in the classical limit. And the specific entropy of the accreted particles is lower than the global average by $\frac{3}{2}k_B$. This is because particles of lower energy are more easily accreted, while particles of higher energy are more readily scattered. We also present numerical results at finite radii for the relevant physical quantities.

gr-qc↗

arXiv2026

KIC 6464285: A Solar-type Eclipsing Binary in a Hierarchical Triple System with Quasi-periodic Out-of-eclipse Modulations

We present the first detailed analysis of the solar-type triple system KIC 6464285. Combining long-term, high-precision photometry from Kepler, TESS, and ZTF with low-resolution spectra from LAMOST and near-infrared high-resolution spectra from SDSS/APOGEE, we performed a joint analysis of the light curves, eclipse timing variations (ETVs), and radial velocities. Spectroscopic analysis reveals the system to be triple-lined, with the inner binary's primary being a G-type main-sequence star and a mass ratio of $0.627(7)$. Light curve modeling indicates that the inner binary is detached, with filling factors of approximately 26\% and 11\% for the primary and secondary, respectively, and a tertiary light contribution of about 27\%. ETV analysis shows a significant light-travel-time effect (LTTE), consistent with the presence of the tertiary companion, whose minimum mass is estimated as $M_{3,\rm min}=0.74(1)~M_\odot$. The light curve exhibits a pronounced O'Connell effect and quasi-periodic variations, indicative of starspot activity modulating the photometry on a $\sim$131-day timescale. Kepler observations further reveal 30 superflares, each with total energies exceeding $10^{34}$ erg. This study presents detailed observational constraints on the orbital configuration, stellar properties, and magnetic activity of KIC 6464285, providing a benchmark for studies of hierarchical triple systems.

astro-ph.SR↗

arXiv2026

Heterogeneous and Adept Snapshot Distillation for 3D Semantic Segmentation

Multi-modal fusion and multi-model ensembling are prevalent in enhancing the performance of 3D semantic segmentation. Despite the impressive performance, these methods either rely on auxiliary input signals or suffer from costly computational expense. To efficaciously enhance the segmentation performance without introducing intolerable costs, we propose to transfer the rich knowledge from the multi-modal model (i.e., point clouds and images) and multiple model experts to the point-cloudbased network through knowledge distillation. Specifically, we present Information-oriented Heterogeneous Distillation (IHD) to help the uni-modal model absorb the complementary knowledge from the multi-modal teacher. We design the Information-Oriented Filtering (IOF) strategy to select informative images from the continuous image sequence for multi-modal fusion. This practice can boost the performance of the multi-modal teacher, thus benefiting the learning of the student. Besides, as opposed to vanilla model ensembling that requires the separate training of each expert, we propose Adept Snapshot Distillation (ASD). ASD treats the freely available model snapshots generated during the training phase as multiple experts, which significantly reduces the training cost for model ensembling. For each expert teacher, it only provides supervision to the student in the class where it is adept. The resulting Heterogeneous and Adept Snapshot Knowledge Distillation, dubbed HAS-KD, attains state-of-the-art results on ScanNetV2 and S3DIS datasets. HAS-KD can be seamlessly integrated into contemporary 3D segmentation algorithms and bring considerable gains without introducing extra inference burdens. The code will be made publicly available upon publication.

cs.CV↗