Computer Vision
Semi-automated reconstruction of indoor geometry from 360-degree video for CFD-based airflow analysis in classrooms
Dhruv Gamdha, James Afful, Shambhavi Joshi et al. (+3 more)
Computational Fluid Dynamics (CFD) is widely used to evaluate ventilation and contaminant transport in occupied buildings, but deployment at scale is limited by three bottlenecks: acquiring room geometry without costly scanning hardware or manual CAD modeling, decomposing the scene into individually manipulable objects, and reconfiguring those objects for alternative layouts without re-capturing the room. We present a semi-automated workflow that converts a single 360-degree
video of a room into individually editable, simulation-ready geometry assets. A dense point cloud is reconstructed using Neural Radiance Fields (NeRF), and 2D instance masks from text-prompted SAM 3
segmentation are lifted to 3D using multi-view consensus and depth-band filtering. Points are separated into object instances with an octree, and occlusion gaps are healed with a connectivity graph. Chair templates are fitted by Iterative Closest Point (ICP)
alignment, and table geometry is generated procedurally. A browser-based editor supports quality assurance and rapid construction of alternative layout configurations. A steady Reynolds-averaged OpenFOAM solution then drives transient passive-scalar transport; the setup is verified using a mesh-sensitivity study and validated against an IEA Annex 20
benchmark. We apply the workflow to two university classrooms and a tiered lecture-
hall auditorium. The capture-to-geometry pass takes two to five hours per room on a consumer workstation. In a controlled obstruction sequence in one classroom, the modeled half-clearance time varies non-monotonically as furniture is added, and a cross-room comparison indicates that clearance behavior cannot be reliably extrapolated between rooms, motivating per-room geometry acquisition. By making that acquisition low-cost, the workflow makes geometry-resolved comparative ventilation studies practical for spaces such as classrooms.
Read the full abstract ▾
All authors Dhruv Gamdha, James Afful, Shambhavi Joshi, Ulrike Passe, Adarsh Krishnamurthy, Baskar Ganapathysubramanian
Differential Geometry
Conformal Killing–Yano Ricci solitons: Basic structure and first examples
Mohammadjavad Habibivostakolaei, Abbas M. Sherif and Yen-Kheng Lim
We introduce a conformal Killing-Yano
Ricci soliton, coupling conformal
Ricci soliton geometry to conformal Killing-Yano 2-forms via the divergence of the form. In four-dimensional Lorentzian signature, the principal conformal structure is not preserved under non-trivial conformal deformation, so the associated form on local Kerr-NUT-(A)dS backgrounds is not closed. For Einstein backgrounds in any dimension and signature, integrability forces the form to be Killing for the background and conformal Killing for the necessarily Einstein
soliton metric, with the conformal factor satisfying
eigenvalue and Obata-type Hessian equations; the coupling fixes the Einstein constant. Vanishing coupling reduces the data to Killing data on the fiber; otherwise the
soliton metric is
Ricci-flat. Locally the background is a warped product. In the compact Riemannian case, the
manifold is the round sphere. Explicit static spherically symmetric and BTZ constructions are given, with applications to exact geometries in general relativity.
Read the full abstract ▾
Robotics
RiverVLN: Phase-Grounded Temporal Vision–Language Navigation for Unmanned Surface Vehicles
Jieling Wu, Yuehao Huang, Jiajun Lv et al. (+3 more)
Vision-
language navigation (VLN) has largely been developed for indoor and terrestrial
robots, where
language can often be treated as a static goal and motion is approximated by discrete or near-instantaneous actions. These assumptions break down for unmanned surface vehicles (USVs): river
navigation requires continuous motion under inertia and limited maneuverability, while long-horizon instructions must be executed through sparse and visually ambiguous maritime landmarks. We introduce RiverVLN, to our knowledge the first
benchmark designed for long-horizon USV VLN under continuous riverine motion, and PGT-NAV, a phase-grounded temporal
navigation framework for USVs. Rather than directly mapping an entire instruction to motion, PGT-NAV converts it into an ordered sequence of visually verifiable semantic phases and maintains the active phase online through grounded visual and motion evidence. This explicit semantic progress state is fused with visual-motion history and phase-specific grounding to predict six local SE(2) pose increments. The resulting trajectory is executed in a predict-execute-re-observe loop, where the vessel executes toward W3, updates phase and grounding, and replans through a map-based safety layer. Experiments show that PGT-NAV substantially reduces recursive position and heading drift relative to GNM-style and ViNT-style baselines and achieves an average success rate of 0.79 in Unity-ROS closed-loop
navigation. Unseen bridge-opening trials and real-world USV experiments further demonstrate that the phase-grounded representation transfers from controlled evaluation to physical USV deployment.
Read the full abstract ▾
All authors Jieling Wu, Yuehao Huang, Jiajun Lv, Tao Huang, Yong Liu, Weiwei Liu
Nuclear Theory
Dynamical Coulomb effects on charged-pion HBT-radius splitting in central Au+Au collisions at few-GeV energies
Pengcheng Li, Yongjia Wang and Qingfeng Li
The pair transverse momentum ($k_T$) and collision-energy ($\sqrt{s_{NN}}$) dependent difference between $\pi^{+}\pi^{+}$ and $\pi^{-}\pi^{-}$ femtoscopic correlations was observed by the HADES and
STAR Collaborations. We investigate how much of the observed splitting can be accounted for by Coulomb dynamics in 0-10% central Au+Au collisions at $\sqrt{s_{NN}}=2.42$-5.2 GeV. Using the UrQMD model and the CRAB programme, three transport scenarios that successively include
baryon-
baryon,
baryon-
meson, and
meson-
meson Coulomb interactions are considered. In a separate calculation based on the
baryon-
baryon reference, we reconstruct the time-dependent effective charge $Z_{\rm{eff}}(t)$ and radius $R_{\rm{eff}}(t)$ of the residual charged source and propagate each pion from its individual last strong-interaction point through the evolving field. The calculations reproduce the main $k_{T}$ and $\sqrt{s_{NN}}$ dependences of the radii.
Baryon–
meson Coulomb interactions generate most of the additional charge splitting in the microscopic calculation. The residual-source treatment also enhances the longitudinal and sideward radius ratios relative to the reference, whereas the outward ratio changes little, and the enhancement generally decreases with increasing $\sqrt{s_{NN}}$. Within the present framework, this contribution does not fully account for the data, motivating further investigation of the charged-source geometry and its interplay with strong-interaction dynamics.
Read the full abstract ▾
Computers and Society
Beyond PUE: A Local Impact Audit Framework for Data Center Environmental Accountability
Sharifa Sultana and Syed Ishtiaque Ahmed
Standard data center sustainability metrics, including Power Usage Effectiveness (PUE), Water Usage Effectiveness (WUE), and Carbon Usage Effectiveness (CUE), measure a facility's resource use and emissions intensity, normalized to IT energy use, without directly representing local resource scarcity, infrastructure capacity, or social footprint. This gap has become politically consequential. In the first quarter of 2026 alone, local opposition delayed or canceled roughly $130 billion in projects across the United States, driven overwhelmingly by recurring concerns over water use, power demand, infrastructure capacity, and transparency rather than internal efficiency, matching the total for all of 2025 [11]. We propose a five-category local impact audit framework covering efficiency, water stewardship, carbon and renewables, regulatory compliance, and local disclosure. The framework is designed for recurring quarterly assessment and independent verification against public records. We illustrate its application using publicly available data from three Illinois facilities that are currently at the center of local policy disputes, and we examine the data-access barriers that constrain independent verification. We position this framework as both a research contribution and a practical instrument for county-level policymakers evaluating data center permitting and moratorium decisions.
Read the full abstract ▾
Quantum Physics
Feedback-Induced Dynamical Phases in a Self-Adaptive Quantum Kicked Rotor
Pan Gao, Zheng-Wei Zhou, Guang-Can Guo et al. (+1 more)
We introduce a self-adaptive Floquet system based on a
quantum kicked rotor, in which the kicking strength itself becomes a dynamical variable generated self-consistently through cavity-mediated feedback. A superradiant transition gives rise to cavity-mediated kicking and two competing instability channels, symmetric and antisymmetric, which provide a unified organizing principle for the nonequilibrium Floquet phases. For resonant kicking, their competition produces double-kick dynamics that support resonant ballistic transport and an emergent antiresonance with period-quadrupled rotor
evolution, arising from a balance between the two instability channels. Remarkably, for incommensurate kicking, the antisymmetric instability stabilizes a robust period-doubled localized phase with persistent subharmonic dynamics despite the underlying incommensurate driving, revealing localized temporal order absent in conventional kicked rotors. As the feedback strength increases, correlated temporal fluctuations progressively suppress
quantum interference, driving crossovers from period-doubled localization to irregular localization and eventually to subdiffusive transport. Our results establish a general framework for self-adaptive
quantum-chaotic dynamics and demonstrate how dynamical feedback can fundamentally reshape transport, localization, and temporal order in driven
quantum systems.
Read the full abstract ▾
All authors Pan Gao, Zheng-Wei Zhou, Guang-Can Guo, Xi-Wang Luo
Number Theory
Higher Reciprocity, Cassels Pairings, and Selmer Towers for the 3/5 Congruent Number Problem
Kaisheng Lei and Shisong Xu
We study the arithmetic of the
elliptic curves \[ A_m:y^2=x(x-m)(x+4m) \] attached to the $3/5$ congruent number problem. A difference of ternary representation numbers controls the relevant central $L$-values. For $p\equiv11\pmod{40}$ the ordinary Cassels pairing degenerates; we construct an explicit $4$-cover and show that the next Cassels–Tate pairing is governed by the normalized representation defect, equivalently by a factorial character, a Pell symbol, and a class number congruence. For composite parameters we compute the ordinary Cassels
matrices of four twists and, in the two
prime case, a degree $1024$ governing field for their joint distribution. The same higher descent extends to larger radicals: a second rational pushout determines the full $\Lambda'$ row of the next pairing. We also determine two explicit Selmer towers with four dimensional ordinary radical, and all finite $2$-power Selmer groups when the ordinary radical is one dimensional.
Read the full abstract ▾
Robotics
HEARTH: An Object-Centric RGB-Thermal-3D Dataset for Temperature-Aware Robot Manipulation
Yuning Su, Borui Li, Yonghao Shi et al. (+2 more)
Language-guided manipulation can depend on physical properties that visible appearance does not reveal. Temperature is one such property, but object
datasets for
robot learning rarely associate measured temperatures with object appearance and geometry. We present HEARTH, an object-centric RGB-thermal-3D
dataset of 90 physical objects from 18 everyday categories, comprising 145 captured object states. Our pipeline maps apparent surface temperatures onto reconstructed meshes through camera calibration and pose transfer. The
dataset includes raw temperature measurements, camera parameters, RGB-textured meshes, and thermal textures for simulation. We use these assets to construct three LIBERO-derived tasks and collect 1,200 demonstrations for fine-tuning a pretrained
vision-
language-action (VLA) model, $\pi_{0.5}$. In an ablation study, adding thermal observations to the VLA increases success on temperature-dependent object-selection tasks from 35.0% for the RGB-only baseline to 75.0%. These results demonstrate the utility of HEARTH for training
robot policies to follow temperature-related instructions.
Read the full abstract ▾
All authors Yuning Su, Borui Li, Yonghao Shi, Bofei Liu, Xing-Dong Yang
Software Engineering
Omni2Web: Benchmarking Audiovisual Website Development
Minghao Han, Zhenghao Xing, Xize Cheng et al. (+7 more)
Screen-recorded web editing requests contain weak deictic expressions such as “this” and “there,” whose referents depend on
speech, cursor trajectories, page state, and edit history. Such requests require intent recovery beyond the explicit specifications assumed by many existing web-editing
benchmarks. We introduce Omni2Web, a bilingual
benchmark of 918 instances spanning 13,907 edit steps. It defines three complementary tracks: Direct Editing evaluates webpage editing from recordings, Instruction Recovery measures explicit intent recovery, and Instruction Utility tests whether recovered instructions can drive a fixed code executor. We evaluate 17 open- and closed-source models. The best models attain 51.17 on the Edit Fidelity Score (EFS) for Direct Editing and 49.14 on the Instruction Recovery Score (IRS); under the fixed executor, the strongest recovered instructions reach 51.08 EFS, still far below the 89.69 EFS obtained with oracle instructions. Step-level analyses show that correct grounding does not guarantee successful edits, while some Omni models recover instructions that the fixed coding model executes substantially better than their direct edits. Controlled ablations further demonstrate the value of temporally aligned audiovisual evidence, while alternative judges preserve the leader and broad ordering. Together, these findings reveal substantial headroom in
multimodal intent recovery and code execution and highlight the promise of pairing Omni rewriters with coding models.
Read the full abstract ▾
All authors Minghao Han, Zhenghao Xing, Xize Cheng, Yuxuan Wang, Junming Lin, Ling Wang, Yinsong Yan, Yunfei Chu, Qize Yang, Jin Xu
Sound
MuLA-Bench: A Multilingual Long-Form Audio Understanding Benchmark via Multi-Tier Auditing
Zeyu Yang, Xinyu Zhang, Zibo Bi et al. (+5 more)
Long-form audio performance is often summarized by context length and aggregate accuracy, obscuring how
language, evidence, and task jointly shape difficulty. We introduce MuLA-Bench: 5,038 open-ended questions over 1,769 in-the-wild recordings totaling 1,377.9 hours, covering 16
languages and eight domains. A balanced
Language x Domain semantic track supports controlled comparisons, while a complementary acoustic track preserves naturally occurring non-
speech evidence. Evidence-grounded generation, shortcut checks, and
language-expert review provide auditable questions without translating a shared source set or injecting target sounds. We evaluate ten audio-
language models and conduct pooled diagnostics on a fixed eight-model cohort.
Language rankings change across domains and tasks; acoustic-semantic performance gaps vary with the requested operation; and temporal errors can persist after the correct event is identified. Long-range
retrieval is comparatively strong, while precise clock
alignment and factual grounding of natural acoustic events remain fragile. MuLA-Bench thus exposes conditional failure patterns that a single long-context score does not capture.
Read the full abstract ▾
All authors Zeyu Yang, Xinyu Zhang, Zibo Bi, Pei Zhang, Xize Cheng, Jin Xu, Baosong Yang, Satoshi Nakamura
Analysis of PDEs
Continuous sonic-supersonic flows in two-dimensional infinitely long divergent nozzles
Lili Qian, Chunpeng Wang and Zihao Zhang
This paper concerns continuous sonic-supersonic flows in a two-dimensional infinitely long divergent nozzle with straight solid walls. For a given inlet that is a small perturbation of an arc centered at the vertex of the nozzle, we establish the global existence and uniqueness of a continuous sonic-supersonic potential flow which is close to the radially symmetric sonic-supersonic flow. The flow is sonic at the inlet, with its velocity along the normal direction, and becomes supersonic in the nozzle. Such a sonic-supersonic model is governed by a quasilinear nonstrictly
hyperbolic equation with degeneracy at the inlet, and the degeneracy is weak in the sense that there exist exactly two distinct characteristics from each sonic point, except for those at the walls, pointing into the supersonic region. One of the crucial ingredients of the analysis is to obtain the precise asymptotic behavior near the sonic inlet by the method of characteristics, which is essential for establishing the local existence of continuous sonic-supersonic flows. Another one is to extend the local solution globally and derive uniform estimates in the supersonic region.
Read the full abstract ▾
Robotics
Daojie Peng, Bingtao Wang, Fulong Ma et al. (+3 more)
Socially competent
humanoid robots must communicate affect and intent through gesture as well as
speech, yet open-ended interaction must become motion that is both expressive and executable on a specific body. This demands semantic flexibility for contextual social intent while preserving deterministic, embodiment-aware
robot control. We present EmoPose, a
vision-
language model (VLM)-guided framework that bridges this gap through an executable semantic interface. Given
language, dialogue history, and optional visual context, the VLM selects an ordered gesture plan containing a communicative class, library variant, intensity, and
speech anchor. A scalable
robot-owned motion library defines the available expressive vocabulary and the source of 14-DoF joint targets. Pose Studio supports automatic trajectory generation, MuJoCo preview, and automatic synchronization of new library entries with the VLM guide; deterministic
robot-side modules validate plans, construct trajectories, schedule gestures, and manage queueing and interruption. This division lets the interaction repertoire grow for new social contexts without changing the control interface or delegating raw joint commands to the foundation model. On the EmoPose-Bench, structured GPT-5.5 planning reaches $98.25\pm0.52\%$ on the Easy tier and $76.50\pm0.54\%$ overall, exceeding same-model direct-label prompting. Further tests validate dialogue-context use and ordered multi-action composition. The system completes the nominal MuJoCo suite and realizes all 29 authored variants on the physical Unitree G1. A four-stop laboratory tour demonstrates expressive narration with interruption, camera-grounded dialogue, and
navigation.
Read the full abstract ▾
All authors Daojie Peng, Bingtao Wang, Fulong Ma, Wenjun Yue, Liang Zhang, Jun Ma
Number Theory
Lower-order terms in mean-square formulas for the Euler–Zagier double zeta-function and regularized multiple zeta values
Tomokazu Onozuka
We study the mean square of the Euler–Zagier double
zeta-function when the second variable moves vertically. Mean-square formulas in and beyond the region of absolute convergence were previously obtained by Matsumoto and Tsumura and by Ikeda, Matsuoka and Nagata. In particular, the latter authors determined the leading terms in the following three boundary cases: \[ \zeta_2\left(1+a,\frac12+it\right),\qquad \zeta_2\left(1+ib,\frac12+it\right),\qquad \zeta_2\left(1-a,\frac12+a+it\right), \] where $a\in\mathbb C$ with $\Re a>0$ and $b\in\mathbb R$ are fixed. The mean squares in the first and third cases, as well as in the second case when $b\neq0$, have leading terms of order $T\log T$. When $b=0$, the second case reduces to the corner $(1,1/2)$, where the leading term is of order $T(\log T)^3$. In the present paper, we refine these results by obtaining asymptotic formulas with remainder $o(T)$. In the case $b\neq0$, an additional oscillatory term of order $T$ occurs. We also show that, at the parameter points $(s_1,\sigma_2)=(k,\ell/2)$ with integers $k\geq1$ and $\ell\geq2$, the constant multiplying $T$ in the mean-square formula is a linear combination of multiple
zeta values. At the boundary points $(k,1/2)$ with integers $k\geq1$, the harmonic finite parts of the corresponding divergent sums are expressed in terms of harmonic regularized multiple
zeta values.
Read the full abstract ▾
Chemical Physics
Quantitative Classical Relaxation Dynamics of Penning-Ionized He/Ne–H$_2^+$/HD$^+$ Complexes
Meenu Upadhyay, Baruch Margulis, Octavio Roncero et al. (+3 more)
Penning ionization allows in situ generation of ionic complexes to launch vibrational relaxation and dissociation dynamics for rare gas–H$_2^+$/HD$^+$ complexes. The ensuing translational kinetic energy release (TKER) spectra characterizing the diatomic cation are a particularly sensitive way to probe the long range part of the intermolecular interactions. TKER distributions from quasi-classical trajectory (QCT) simulations for the Ne–H$_2^+$ and Ne–HD$^+$ complexes quantitatively match measured positions and relative intensities of most experimentally observed features and are also on par with time-dependent
quantum simulations. Specifically, the QCT-simulations allow to separate signatures arising from H$_2^+$/HD$^+$ relaxation into different final vibrational states, such as for $(v=2) \rightarrow (v' = 0/1 )$. Trajectory-resolved QCT analysis also provides mechanistic interpretations of the spatial and time-dependent dynamics underlying the TKER distributions. Rare gas–H$_2^+$/HD$^+$ complex lifetimes directly correlate with the final rotational
quantum number $j'$ of the diatomic. Low-$j'$ products are associated primarily with an initial axial approach of the rare gas atom, whereas formation of higher-$j'$ states predominantly results from an initial T-shaped approach geometry.
Read the full abstract ▾
All authors Meenu Upadhyay, Baruch Margulis, Octavio Roncero, Karl Horn, Edvardas Narevicius, Markus Meuwly
Sound
Misrecognition or Abstraction? Rethinking Outputs of Sound Event Recognition
Naoya Tomida, Yuki Okamoto and Keisuke Imoto
Conventional general sound recognition systems typically output deterministic sound event labels, implicitly assuming that the target sound class can be correctly identified from the input audio. However, in real listening situations, the sound event class is not always clearly identifiable. Human listeners may nevertheless understand their surroundings from an ambiguous sound without identifying its exact sound event class. This motivates a discussion of how the outputs of sound recognition systems should be redesigned under such uncertainty. As a basis for this discussion, this paper proposes an output representation for sound event recognition that combines a sound event class, its confidence score, and an onomatopoeic description of the sound. The proposed representation preserves conventional class-based recognition while providing an additional onomatopoeic description of acoustic characteristics that can remain informative even when the class prediction is uncertain. Experiments using ESC-50 and ESC-50-Onomatopoeia show that the proposed method achieves sound recognition performance comparable to that of a conventional recognition-only system. In addition, an LLM-as-a-judge evaluation and subjective listening experiments indicate that the proposed output is preferred over conventional deterministic outputs based on the sound event label, particularly when used to support understanding of the surrounding environment. These results suggest that such output representations can make sound event recognition more informative and communicative under uncertainty.
Read the full abstract ▾
High Energy Astrophysical Phenomena
Observational Evidence of an Internal Shock Collision in the Relativistic Jet of the Nearby Radio Galaxy 3C 111
Kenzo Kawamura, Motoki Kino, Mareki Honma et al. (+21 more)
We report 10 epochs of 22 GHz East Asia To Italy: Nearly Global VLBI (EATING VLBI) observations of the nearby
radio galaxy 3C~111 obtained from 2020 to 2023. Combined with VLBA data at 15 and 43 GHz, these observations yield a 64-epoch, multi-frequency sequence of high-resolution
radio images. We used high-resolution
Bayesian Stokes~I imaging and Gaussian component modeling to derive light curves and
knot kinematics. We find evidence that a faster, later-emitted
knot catches up with a slower, earlier
knot at a deprojected distance of about 10 pc from the central engine. The
knot component responsible for this interaction propagated at relativistic speed and exhibited a rise-and-decay
evolution over approximately six months. The Stokes~I flux
evolution supports the internal shock interpretation, while archival polarization images reveal a temporary increase in fractional linear polarization in the interaction region near the Stokes~I flux maximum. The timing of the 2021 GeV
flare suggests that this earlier high-energy event is more naturally associated with activity near the core. These results provide direct evidence for an internal shock in an active
galactic nucleus jet and support a hybrid picture in which internal and standing shocks coexist in relativistic jets and may play different roles in their variability.
Read the full abstract ▾
All authors Kenzo Kawamura, Motoki Kino, Mareki Honma, Kazuhiro Hada, Paul Tiede, Evgeniya Kravchenko, Marcello Giroletti, Rocco Lico, Kazunori Akiyama, Mieko Takamura, Filippo D'Ammando, Monica Orienti, Gabriele Giovannini, Fumie Tazaki, Bong Won Sohn, Salvatore Buttaccio, Lang Cui, Giuseppe Maccaferri, Andrea Melis, Alexey Melnikov, Carlo Migoni, Matteo Stagni, Kiyoaki Wajima, Takeshi Sakai
Computer Vision
Retrieval Geometry Shapes Cache-Based Clip Adaptation
Mahir Shahriar Tamim, Md. Samiul Alim, Azmine Toushik Wasi et al. (+5 more)
Cache-based test-time adaptation improves CLIP predictions by storing and retrieving examples from the target stream while keeping the model frozen. However, existing methods largely treat the feature space used for image-image
retrieval as fixed, leaving open how much adaptation depends on the
retrieval space itself. We study this question by fixing the
memory and changing only the
retrieval encoder, finding that the same
memory can yield very different gains: across sixteen
retrieval spaces, ImageNet-A cache gain ranges from at most +0.44 points for CLIP and MAE to +19.7 +/- 0.4 for DINOv2-L, while label-free
retrieval-space selection retains 98% of oracle gain on ImageNet-V2. These results show that
memory quality depends not only on which examples are stored, but also on how they are retrieved. Motivated by this finding, we propose MARC (
Memory Augmented
Retrieval for CLIP), a training-free system that uses frozen CLIP for prediction and DINOv2-B for
retrieval with a single fusion weight. A single-view cache repairs 1074 +/- 21 baseline errors, compared with 878 +/- 4 for a 64-view ensemble, at roughly one seventh of the cost. Across four ImageNet distribution shifts, MARC reaches a 67.91% OOD average and, at matched DINOv2-B scale and eight views, achieves 64.17 +/- 0.31% versus 62.75 +/- 0.15% for a graph-based cache system while running 2.6 times faster. Overall, our results establish
retrieval space as a first-order design choice for robust cache-based adaptation in remote sensing, scientific imaging, and changing visual environments.
Read the full abstract ▾
All authors Mahir Shahriar Tamim, Md. Samiul Alim, Azmine Toushik Wasi, Shahriyar Zaman Ridoy, Meharun Nesa, Mohammad Abu Yousuf, Alex Lamb, Mohammad Ali Moni
Computer Vision
Accurate Motion Estimation with Bézier Control Point for Efficient Frame Interpolation
Shuhao Han, Chenyang Wu, Chun-Le Guo et al. (+4 more)
In frame interpolation tasks, motion ambiguity in the training set causes models to generate blurry intermediate frames. Moreover, the assumption of uniform motion between frames during inference further leads to inaccuracies in the generated intermediate frames. To tackle these challenges, we propose an Accurate motion estimation
algorithm with Bézier Control point, ABC-Inter, for efficient frame Interpolation. Specifically, ABC-Inter designs an Accurate Flow estimation Module (AFM) by decoupling two-frame features and mapping to corresponding coordinates to better estimate the optical flow between the two frames. Furthermore, ABC-Inter eliminates motion ambiguity in the training set by introducing Bézier control points that are computed using the input frames and the intermediate ground-truth (gt) frames. This allows the model to estimate accurate optical flow between two frames during the training process, thereby solving the blurriness problem in the generated intermediate frames during inference. Benefiting from the more accurate flow estimation between two frames, we can introduce additional frames and directly use multiple flows to calculate Bézier control points for modeling non-uniform motion without retraining the model. Simultaneously, to realize the estimation of non-linear motion using only two frames, we also introduce a new Bézier control point estimation module which achieves better motion estimation between the two frames by performing fine-tuning on the model in the second stage. Experimental results demonstrate that our ABC-Inter achieves state-of-the-art performance on multiple
benchmark datasets and exhibits excellent visual perception.
Read the full abstract ▾
All authors Shuhao Han, Chenyang Wu, Chun-Le Guo, Zheng-Peng Duan, Zhen Li, Ming-Ming Cheng, Chongyi Li
Sound
OmniEcho: Audio-Visual Spatial Understanding for Omni-Modal Embodied Agents
Ruixun Liu, Yuxuan Wang, Jiacheng Xie et al. (+10 more)
Humans can effortlessly localize the direction of a sound source and integrate it with visual cues for
reasoning, yet this remains challenging for
embodied agents. In particular, it is still unclear how to effectively evaluate and model spatial audio understanding in
embodied settings. To address this gap, we introduce OmniEchoBench, a unified
benchmark for spatial audio-visual perception and audio-
vision-
language navigation. OmniEchoBench comprises six tasks over 197 real-world spatial audio-visual scenes, 2,972 question-answer pairs, and 900
navigation samples with first-order ambisonics (FOA) audio collected from 30 real-world environments. To enable scalable training supervision, we develop a controllable rendering pipeline for spatial audio. It preserves
geometric consistency among sound sources, visual observations, and
agent trajectories. Building on this, we propose OmniEcho, a spatially aware omni-modal model. It introduces an FOA spatial encoder alongside a pretrained semantic audio pathway. Extensive experiments show that OmniEcho achieves state-of-the-art performance on spatial audio-visual perception. For our sound-guided
navigation, OmniEcho reaches a performance level close to that of traditional
vision-
language navigation. These results demonstrate that spatial audio can serve as a valuable signal for
embodied scene
reasoning and
navigation, while also highlighting fine-grained spatial localization and distance estimation as important open challenges. Our code and data will be available in https://github.com/PKU-VaLuE-Lab/OmniEcho/tree/main
Read the full abstract ▾
All authors Ruixun Liu, Yuxuan Wang, Jiacheng Xie, Yuhuan You, Donghua Cai, Junming Lin, Xiong-Hui Chen, Zhifang Guo, Yunfei Chu, Qize Yang, Xize Cheng, Jin Xu, Yiwu Zhong
Quantum Physics
Quantum scattering by intersecting {\delta}-potential barriers: from Gaudin's kaleidoscope to quantum Galperin billiards
Yi-Cong Yu, Wen-Jie Qiu and Xiaoming Cai
We develop a general framework for the scattering of a plane wave by a class of Gaudin kaleidoscope models: straight $\delta$-potential barriers intersecting at a common point. The projected Lippmann-Schwinger equations split into a singular part, a finite pole closure generated by a
geometric moving rule, and a regular remainder, the outgoing state being an atomic measure on the circle. At the special angles $\pi/N$, where the intersecting barriers generate the dihedral group $D_N$, the number of channels stays fixed, whereas at generic angles channels are created and destroyed, opening smoothly from zero weight as a critical direction is crossed. This leaves no singular trace in the probabilities, being carried instead by the phase shifts, a
quantum-classical correspondence beyond the reach of the coordinate Bethe ansatz. The same equations solve the
quantum Galperin billiards exactly, the method of images reducing them to a single linear system whose phase shifts follow in closed form.
Read the full abstract ▾
Cryptography and Security
The Anatomy of Address Poisoning on Ethereum: Funding Mechanisms, Scam Signatures, and Laundering via Tornado Cash
Son Hoang Dau, Thanh Nguyen, Phuong Duy Huynh et al. (+5 more)
Address-Poisoning Transfer (APT) is a prevalent blockchain phishing scam in which a scammer poisons a victim's address book by generating a transfer with a phishing address that looks similar to a benign address that the victim has previously interacted with. Although simple, APT phishing attacks have cost users millions of dollars in recent years, which has captured the
attention of the research community (Ye et al. WWW'24, Guan-Li CCS'24, Chen et al. NDSS'25, Tsuchiya et al. USENIX'25). In this work, we go beyond detection and investigate three important and underexplored aspects of APT: scam funding mechanisms, scam signatures, and scam proceeds laundering via public services. In particular, we propose five families of scam signatures that capture key aspects of APT operations, which are useful for address clustering. We also conduct the first investigation into usage of Tornado Cash for funding APTs and laundering scam proceeds.
Read the full abstract ▾
All authors Son Hoang Dau, Thanh Nguyen, Phuong Duy Huynh, Hong Yen Tran, Nicholas Huppert, Jeff Nijsse, Huong Ha, Xun Yi
Computer Vision
ScaleBlind: Point Cloud Completion under Unknown Scale
Shenghui Wu, Chen Wang, Yuan Feng et al. (+3 more)
Point cloud completion aims to infer a complete 3D shape from a partial point cloud and serves as a fundamental building block for downstream tasks such as reconstruction, editing, and simulation. Despite the recent progress, existing learning-based methods often implicitly rely on access to the ground-truth shape scale (GT-scale) during both training- and testing-time normalization, assuming privileged information that is unavailable in real-world inference. This hidden assumption limits practical deployment and can lead to severe completion artifacts, e.g., over- or under-completion and nested shells, once the oracle GT-scale cue is removed. We observe that the recent foundation image generation models exhibit a strong capability of understanding objects and geometries, and producing multi-view consistent renderings, making them promising priors for GT-scale-free 3D completion. Motivated by this insight, we propose ScaleBlind, a novel framework that leverages foundation-model-based image completion to recover global scale directly from partial inputs and then faithfully produces the 3D completion. Specifically, ScaleBlind dreams out complete multi-view appearances from rendered partial views, lifts the inferred missing regions back into 3D to obtain a geometry-aware coarse completion, and further refines it via a powerful cross-modal fusion network with the original partial point cloud. By harnessing 2D foundation priors, our method eliminates the need for accessing GT-scale information at inference. Moreover, it provides a principled bridge between 2D generative priors and 3D point cloud completion. Extensive experiments demonstrate the superiority of our framework, making ScaleBlind the new state-of-the-art for the point cloud completion task.
Read the full abstract ▾
All authors Shenghui Wu, Chen Wang, Yuan Feng, Guangshun Wei, Yuanfeng Zhou, Changjian Li
Human-Computer Interaction
Alignment and Divergence between Humans and AI in Interpersonal Privacy Decisions
Hanxiang Zeng, Shuning Zhang, Xinyuan Zhou et al. (+5 more)
AI assistants increasingly mediate interpersonal communication on behalf of their primary user, but they risk violating the
privacy expectations of third-party information owners. Resolving these tensions requires understanding how humans anticipate interpersonal
privacy boundaries. Therefore, we conducted a dyadic study (N=76) and a matched evaluation of AI models across 18 information types and 3 recipient relationships. We found that data owners'
privacy judgments are highly contextual and relationship dependent. While familiar data co-owners show meaningful
alignment with owners' expectations, they significantly overestimate the need for permission. Interestingly, greater familiarity within the owner-co-owner dyad was associated with both higher disclosure acceptability and lower co-owner misalignment, whereas our exploratory four-item empathy measure was not. In contrast, AI models significantly underperform human co-owners in anticipating the data acceptability, even when provided with within-dyad examples. These findings underscore a core HCI design challenge to develop
privacy-aware AI that respects multi-stakeholder information boundaries.
Read the full abstract ▾
All authors Hanxiang Zeng, Shuning Zhang, Xinyuan Zhou, Tianqi Song, Yuhan Yuan, Yuting Yang, Shuai Ma, Xin Yi
Functional Analysis
Toeplitz and Hankel operators on Bergman spaces with doubling weights
Yongjiang Duan, Kunyu Guo, Junhan Hong et al. (+1 more)
In this paper, we focus on the boundedness and compactness of the Toeplitz operators and the Hankel operators on the $\mathcal{D}$-weighted Bergman spaces $A_{\omega}^p$ ($1\leq p<\infty$). In particular, the boundedness of Toeplitz operators with $\mathrm{BMO}_\omega^p$-symbols on $A_{\omega}^p$ is characterized in terms of the Berezin-type transform. Several sufficient conditions for the Toeplitz operators $T_f^\omega$ with locally integrable symbols to be bounded (resp. compact) for each $1\leq p<\infty$ are presented. Moreover, sufficient and necessary conditions for the boundedness of the Toeplitz operator $T_{\bar{f}}^\omega:A_\omega^1\rightarrow A_\omega^1$ and the Hankel operator $H_{\bar{f}}^\omega:A_\omega^1\rightarrow L_\omega^1$ with co-analytic symbol are established.
Read the full abstract ▾
All authors Yongjiang Duan, Kunyu Guo, Junhan Hong, Fugang Yan
Data Structures and Algorithms
The Inverse Lyndon Array
Clelia De Felice, Pietro Negri, Manuel Sica et al. (+2 more)
The Lyndon array stores, at each position of a word, the length of the longest Lyndon factor starting at that position and plays an important role in combinatorics on words, for example, in the construction of fundamental data structures such as the suffix array. In this paper, we introduce the Inverse Lyndon array, the analogous structure for inverse Lyndon words, namely words that are lexicographically greater than all their proper nonempty suffixes. Unlike standard Lyndon words, inverse Lyndon words may have non-trivial borders, which introduces a genuine theoretical difficulty. We show that the Inverse Lyndon array can be characterized in terms of the next greater suffix array together with a border-correction term, and we prove that this correction coincides with a longest common extension (LCE) value. Building on this characterization, we adapt the nearest-suffix framework underlying Ellert's linear-time construction of the Lyndon array to the inverse setting, obtaining an O(n)-time
algorithm for general ordered alphabets. Finally, we show that the Inverse Lyndon array can also be used to reconstruct the canonical inverse Lyndon factorization in linear time.
Read the full abstract ▾
All authors Clelia De Felice, Pietro Negri, Manuel Sica, Rocco Zaccagnino, Rosalba Zizza
General Relativity and Quantum Cosmology
Optimal all-angle reconstruction of the Hellings-Downs curve
Jing-Hong Han and Zhi-Chao Zhao
Pulsar timing arrays (PTAs) detect nanohertz gravitational waves through spatial correlations between the timing residuals of different
pulsars. For an isotropic, unpolarized
stochastic background in general relativity, the ensemble-mean correlation follows the Hellings–Downs (HD) curve; measuring this angular pattern tests the gravitational-wave origin of the signal. Standard bin-by-bin reconstructions optimize the weights within each angular bin separately. We reconstruct the curve jointly using the full
pulsar-pair covariance, retaining a free amplitude in every bin. The resulting all-angle best linear unbiased
estimator minimizes the variance of every bin value and every linear combination of bins. Applied to the public NANOGrav 15 yr data products, our method reduces the bin standard deviations by up to 12.9%, with a median reduction of 9.5%. For a future Square Kilometre Array Observatory (SKAO)-like PTA, the predicted reduction reaches 36.4%, with a median of 29.6%, enabling significantly more precise measurements of the gravitational-wave background.
Read the full abstract ▾
High Energy Physics (Theory)
Spin density from second order transport in fluids
Domingo Gallegos, Roi Klein and Amos Yarom
We study the hydrodynamics of a charged fluid with
spin degrees of freedom. We argue that, in a conformal field theory in 3+1 dimensions, the
spin density is related to a linear combination of standard second-order transport coefficients, and an additional term quadratic in the gauge- and mixed gauge-gravitational anomaly coefficients. We compute the
spin density in several free and
holographic theories and compare the results.
Read the full abstract ▾
Analysis of PDEs
Riesz transform on manifolds with mixed ends for $1<q<2$
Bo Li, Tianjun Shen and Zihan Wang
Let $M_1$, $\cdots$, $M_\ell$ be complete
manifolds of the same dimension, where $2\le \ell\in\mathbb{N}$. Suppose that each $M_i$ satisfies two side Gaussian bounds. If some of these
manifolds are
parabolic and there exists a constant $1\le n_i\le 2$ such that for some $x_i\in M_i$ with $1 \le r \le R < \infty$, $$c_i\left(\frac{R}{r}\right)^{n_i}\le \frac{Vol_{M_i}(B(x_i,R))}{Vol_{M_i}(B(x_i,r))}\le C_i\left(\frac{R}{r}\right)^{n_i},$$ by assuming that all the
manifolds satisfy the relative connectedness of the annuli ($RCA$) condition introduced by Grigor'yan and Saloff-Coste, we show that the Riesz transform $\nabla \mathcal L^{-1/2}$ is bounded on $L^q(M)$ for each $1<q<2$ on the gluing
manifold $M=M_1 \# M_2 \# \cdots \# M_\ell$.
Read the full abstract ▾
Computer Vision
Enhancing Shrimp Disease Detection via Deep Learning and Data Refinement for Resilient Aquaculture
Vinh Canh-Thanh Truong, Hai-Binh Pham and Ngoc Hong Tran
Shrimp diseases continue to cause devastating losses in the aquaculture industry, driving a critical need for robust, automated detection. This work contributes the first application of
Vision Transformers (ViT) and Self-Supervised Learning (SSL) to the shrimp farming domain, addressing both performance bottlenecks and data labeling challenges. We propose two deep learning pipelines to classify four key diseases: Healthy, Black Gill (BG), White Spot Syndrome Virus (WSSV), and a co-infection of both using a
dataset of 4,348 images. First, our supervised transfer-learning approach leverages ImageNet-pretrained ViT-Small/16 and EfficientNet backbones. Second, we introduce a contrastive learning framework (SimCLR) with a ViT-Small encoder to extract robust representations from unlabeled images prior to fine-tuning. Our results establish strong new baselines for sustainable aquaculture monitoring. The supervised approach achieves an outstanding 96% accuracy with fast convergence, outperforming traditional generic models, while the label-efficient SSL approach reaches a highly competitive 85% validation accuracy.
Read the full abstract ▾
Cryptography and Security
CLOADER: Evading Security Mobile Defenses via Runtime Obfuscation and Adaptive Hooking Tactics
Nhat-Anh Huynh, Minh Quang Luu and Ngoc Hong Tran
We propose a stealth framework that eliminates detection of hooking tools such as Frida and Xposed in secured mobile environments by replacing static configurations with dynamic evasion tactics. In contrast to existing approaches that apply these techniques independently, the framework introduces a unified runtime control layer that systematically coordinates network, temporal, and code-level evasive transformations. The solution integrates randomized port allocation, runtime code obfuscation, delayed execution triggers, and self-integrity checks to disrupt signature-based scans, timing heuristics, and tampering attempts. A custom Android loader, CLoader, enforces these mechanisms to isolate hooking activities from security monitors while maintaining complete interception and modification capabilities. Validation across enterprise anti malware systems, hardened applications, and device management platforms demonstrates a 90% bypass rate in our evaluation
matrix. This approach enables reliable penetration testing and malware analysis in locked-down mobile ecosystems by masking network, temporal, and code-level fingerprints without architectural overhauls.
Read the full abstract ▾