Data Structures and Algorithms
Low-Stretch Spanning Trees via Smoothed Analysis of Dijkstra's Algorithm
Ioannis Dorkofikis, Bernhard Haeupler, Maximilian Probst Gutenberg et al. (+3 more)
Given an undirected weighted graph $G$, a $\gamma$-approximate low-stretch spanning tree (LSST) $T \subseteq G$ is a tree that approximates the distance metric of $G$ up to a $\gamma$-factor in expectation. Currently, existing
algorithms to find a provably good LSST carefully construct an approximate shortest-path tree from an arbitrary source. The resulting
algorithms are intricate. In contrast, practitioners observed that a much simpler heuristic performs surprisingly well: choose an arbitrary root, run Dijkstra's
algorithm, and use the resulting shortest-path tree as an LSST. In this paper, we give a smoothed analysis of shortest-path tree
algorithms, such as Dijkstra's
algorithm, that explains this behavior. We show that adding a small perturbation to the weights of the input graph suffices to turn the shortest path tree rooted at an arbitrary node in the resulting graph into an $\tilde{O}(1)$-approximate LSST. We further show that the set of perturbations can be computed efficiently from few low-diameter decompositions (LDDs). Thus, our proof is also constructive in the sense of giving a novel approach to computing LSSTs.
Read the full abstract ▾
All authors Ioannis Dorkofikis, Bernhard Haeupler, Maximilian Probst Gutenberg, Antti Roeyskoe, Aurelio Sulser, Gernot Zöcklein
High Energy Physics (Experiment)
Muon Detection and Direction Reconstruction with the Upgraded 1-ton Water Cherenkov Prototype Detector at CJPL-I
Yapeng Wang, Yuzi Yang, Shaomin Chen et al. (+35 more)
The 1-ton prototype detector for Jinping
neutrino experiment (JNE-1ton) has completed its hardware upgrade and has been successfully operated in both water and
liquid-scintillator modes at CJPL-I, which is situated under a 2400-m rock overburden. During the upgrade, the detector's mechanical support structure and PMT layout were redesigned. The original Hamamatsu PMTs were replaced with 8-inch MCP-PMTs from North Night
Vision, increasing the PMT count from 30 to 60. The detector was then operated in water mode for 85 days (Water-I) and 79 days (Water-II). From the accumulated water-mode data, we derive a
muon detection efficiency (Water-II) that is approximately 59% higher than that of the pre-upgrade
liquid scintillator detector. The measured
muon flux is $\phi_{\text{I+II}} = (3.55 \pm 0.43_{\mathrm{stat}}\pm 0.28_{\mathrm{syst}}) \times 10^{-10}~\mathrm{cm}^{-2}\mathrm{s}^{-1}$, which is consistent with the previous measurement. Owing to the characteristic angular dependence of Cherenkov radiation and the increased PMT coverage, the uncertainty in
muon direction reconstruction is approximately 6$^\circ$, which corresponds to a reduction to 30% of its previous value. Notably, one rare up-going
muon event was clearly identified. It is excluded from the
cosmic-ray
muon flux sample and opens a new window for
neutrino-induced event studies in the deepest underground laboratory in China. This work marks the first time that CJPL has employed a cost-effective water detector for
muon flux measurement.
Read the full abstract ▾
All authors Yapeng Wang, Yuzi Yang, Shaomin Chen, Wei Dou, Haoyang Fu, Guanghua Gong, Lei Guo, Ziyi Guo, XiangPan Ji, Jianmin Li, Jinjing Li, Bo Liang, Ye Liang, Ling Liu, Qian Liu, Zhiyi Liu, Wentai Luo, Ming Qi, Wenhui Shao, Haozhe Sun, Jian Tang, Yuyi Wang, Zhe Wang, Changxu Wei, Jun Weng, Yiyang Wu, Benda Xu, Chuang Xu, Tong Xu, Tao Xue, Haoyan Yang, Aiqiang Zhang, Bin Zhang, Xinshun Zhang, Yang Zhang, Zhicai Zhang, Lin Zhao, Yangheng Zheng
Computer Vision
VideoPhysEdit: Physical Counterfactual Video Editing via Rigid-Body Physical Scene Reconstruction
Conghan Yue, Yuanjie Chen, Yue Han et al. (+4 more)
Video editing has advanced substantially in recent years, with methods increasingly accounting for the visual consequences of edits, such as changes to shadows and occlusions. However, the physical consequences of edits, including changes to subsequent motion and interactions, remain less explored. We formulate this problem as physical counterfactual
video editing (PCVE), which aims to generate a counterfactual
video depicting the resulting motion and interactions given a source
video, a physical edit, and its execution frame. PCVE is challenging because it requires understanding scene physics and inferring the downstream motion and interactions induced by a physical intervention, while paired factual and counterfactual data and dedicated evaluation metrics are lacking. We introduce VideoPhysEdit, a new training-free pipeline for PCVE in rigid-body scenes. It makes physical
reasoning explicit through a novel physical scene reconstruction method that recovers a scene reproducing the observed motion and interactions under simulation, enabling the pipeline to apply physical edits as interventions and use the resulting trajectories to guide counterfactual
video generation. We further construct PCVE-RigidBench, a synthetic
benchmark with paired source and counterfactual target
videos and physical ground truth, and introduce the Physical Edit Score. VideoPhysEdit achieves substantially higher physical edit accuracy than open-source methods and commercial models while maintaining competitive visual fidelity. Its Physical Edit Score is 0.376, the only positive score among the compared methods. Qualitative comparisons on real
videos further show that VideoPhysEdit applies to real-world scenes and better depicts the downstream motion and interactions induced by the edits than the compared methods. Code: https://github.com/Hammour-steak/VideoPhysEdit
Read the full abstract ▾
All authors Conghan Yue, Yuanjie Chen, Yue Han, Ya Gao, Yunyan Xiao, WeiYao Zhang, Zhineng Chen
High Energy Physics (Phenomenology)
The $D\pi$ and $D^*\pi$ femtoscopy puzzle
Pablo Encarnación, Miguel Albaladejo, Albert Feijoo et al. (+1 more)
We present a relativistic momentum-space framework for femtoscopic correlation functions including strong and Coulomb interactions in coupled channels. The approach extends the Vincent–Phatak prescription to the three-dimensional Bethe–Salpeter equation, combining a fully relativistic treatment with the correct Coulomb asymptotics through wave-function matching. We apply it to the $D\pi$ system using a coupled-channel heavy-
meson chiral interaction that reproduces the established two-pole structure of the $D_0^*(2300)$. Our correlation functions agree with previous theoretical calculations but differ significantly from those extracted by the ALICE Collaboration. We trace this tension to the assumption made in the experimental analysis that the genuine correlation function reaches unity above a relative momentum of $100$ MeV, whereas this region remains affected by the lower $D_0^*(2300)$ pole. Repeating the extraction over different fitting windows reveals a strong dependence of the extracted signal on the chosen range, demonstrating that the resulting $D\pi$ scattering parameters are not robust and motivating a dedicated reanalysis of the experimental data. We extend the analysis to the $D^*\pi$ system, where the same issue arises in close connection with heavy-
quark spin symmetry. Together with the unresolved tension between theoretical calculations and the ALICE $K\pi$ femtoscopic measurements, this highlights the need to clarify the origin of this puzzle, with the aim of establishing femtoscopy as a reliable quantitative tool for precision studies of low-energy
hadron interactions.
Read the full abstract ▾
All authors Pablo Encarnación, Miguel Albaladejo, Albert Feijoo, Juan Nieves
High Energy Physics (Phenomenology)
Parameter degeneracy and information loss in inverse flavor mapping for high-energy astrophysical neutrinos
Zi-Qiang Chen, Zhi-zhong Xing and Ye-Ling Zhou
Given a high-energy astrophysical
neutrino flux, its source
flavor ratios $\eta = \{\eta^{}_e, \eta^{}_\mu, \eta^{}_\tau\}$ are correlated with the ones $f = \{f^{}_e, f^{}_\mu, f^{}_\tau\}$ measured at a
neutrino telescope via $f = P \eta$, where the elements of $P$ consist of four
lepton flavor mixing parameters. But realistic inverse
flavor mapping $\eta = P^{-1} f$ encounters unavoidable parameter degeneracy and information loss, especially in or near the $\det P = 0$ limit allowed by current
neutrino oscillation data. In this work we explore the parameter correlation condition for $\det P = 0$ including the $\mu$-$\tau$ reflection
symmetry case, formulate the corresponding constraints on $f^{}_e$ and $f^{}_\mu$ versus $\eta^{}_e$ and $\eta^{}_\mu$, and illustrate to what extent the source
flavor distribution can be mapped by using the recent IceCube all-sky
neutrino flux data ranging from 5 TeV to 10 PeV under the assumption that the relevant sources have a common
flavor composition.
Read the full abstract ▾
Quantum Physics
The Overlap Gap Property: Separating Quantum and Quantum-Inspired Approximate Optimization Algorithms
Mark Goh and Thorge Müller
In this paper, we prove that the Mean-Field Approximate Optimization
Algorithm (MF-AOA), a
quantum-inspired classical
algorithm of the
Quantum Approximate Optimization
Algorithm (QAOA), is unable to give arbitrary near optimal solution for problems exhibiting the Overlap Gap Property (OGP). We show this by relating the MF-AOA to finite-
memory Approximate Message Passing (AMP)
algorithms, which are known to be limited in performance by the OGP. Thus, our paper argues that problems exhibiting the OGP are one such instance where the QAOA can outperform the MF-AOA and AMP
algorithms in general. We supplement this by numerically simulating the performance of the QAOA on instances of Max-$4$-XORSAT that exhibit the OGP and find that the QAOA is able to surpass the barrier, while our finite-size scaling analysis favors a super-
polynomial over a
polynomial depth dependence. In addition, we observe a non-differentiable point in the performance of the QAOA near by the OGP barrier, where the optimal QAOA parameters change from an adiabatic to a non-adiabatic schedule.
Read the full abstract ▾
Machine Learning
CTP-FL: Common-Trajectory Gradient Prediction for Federated Learning
Junkang Liu
Communication-efficient
federated optimization commonly spends several gradient evaluations between server updates. Existing local-update methods use this computation to advance an independent model on each client. Under heterogeneous data, however, these models evaluate gradients at different locations, making the aggregated update difficult to interpret as a gradient of the global objective. We study an alternative use of the same computation budget: evaluate the global objective along a shared, predicted path. We propose Common-Trajectory Predictive
Federated Learning (CTP-FL). At each round, all clients construct the same sequence of query points from the current global model and the previous aggregated direction, evaluate $K$
stochastic gradients along this sequence, and upload their average. The server then performs a single global update. Thus, CTP-FL uses $K$ mini-batch gradients per client and one model-sized vector in each communication direction, matching the per-round computation and communication of full-participation FedAvg-M. Shared query points make the aggregated direction an unbiased
estimator of the average global gradient along the predicted path. The remaining discrepancy from the gradient at the current model is controlled by the path length, without assuming bounded client-gradient dissimilarity or bounded gradients. For smooth non-convex objectives, we establish an $\mathcal{O}\!\left( \sqrt{L\Delta\sigma^2/(NKR)}+L\Delta/R \right)$ average-stationarity bound under full participation. The analysis isolates a testable trade-off: extending the prediction path provides more forward-looking gradient information but increases its displacement bias.
Read the full abstract ▾
Quantum Physics
Asymptotically Good Quantum Codes with Addressable Transversal T Gates
Tongyin Lin, Bujiao Wu and Bin Cheng
Designing
quantum codes with both transversal non-Clifford gates and good error-correcting parameters is an important goal in fault-tolerant
quantum computation. Here, we construct an explicit family of binary CSS codes with constant rate and linear distance that admit fully addressable transversal $T$ gates. Specifically, any prescribed tensor product of logical powers of $T$ is implemented by a tensor product of physical powers of $T$, without any subsequent correction. Our construction combines algebraic-geometry codes with suitable binary
embedding to obtain generalized divisibility of binary codes. This divisibility then enables fully addressable transversal $T$ gates. Furthermore, we formulate the minimum binary
embedding length problem as a minimum-weight problem over an affine space and numerically improve the constants in the asymptotic rate and relative-distance bounds.
Read the full abstract ▾
Machine Learning
Explaining Hyperbolic Neural Networks via Geometry-Aware Relevance Propagation
Ping Xiong, Shanglin Li, Yi Ding et al. (+2 more)
Hyperbolic neural networks introduce
geometric operations that require explicit treatment in relevance propagation. Equivalent
geometric realizations can produce different feature attributions, even when local relevance is conserved. We study this problem through
Geometric Representation Invariance (GRI), a specialization of Implementation Invariance, and zero-
curvature consistency, which requires identity relevance propagation when a
geometric module approaches the identity. We propose LRP-radial-all for origin-centered radial modules, treating
geometric scaling as modulation and assigning relevance entirely to the signal branch. The rule conserves relevance, is invariant to equivalent radial factorizations, and satisfies zero-
curvature consistency, yielding GRI for a specified Poincaré-Lorentz logarithmic-map construction. In contrast, a conservative LRP-half baseline can violate both consistency criteria. Experiments on
hyperbolic MNIST, sEEG, and CIFAR-10 classifiers assess attribution fidelity, qualitative explanations, and runtime. LRP-radial-all achieves competitive attribution fidelity across
datasets with runtime comparable to Gradient$\times$Input and substantially lower than Integrated Gradients. These findings motivate geometry-aware propagation rules that distinguish relevance conservation from consistency across equivalent computations.
Read the full abstract ▾
All authors Ping Xiong, Shanglin Li, Yi Ding, Thomas Schnake, Shinichi Nakajima
Soft Condensed Matter
Landau-de Gennes corrections to the Oseen-Frank limit: Anchoring-induced tilt modes
Prabakaran Rajamanickam
An asymptotic analysis of the Landau–de Gennes framework is performed to compute higher-order corrections to the Oseen–Frank limit in a bounded three-dimensional domain under appropriately scaled surface anchoring energy. A systematic decomposition of the $\mathbf{Q}$-tensor into three mutually orthogonal subspaces-the uniaxial scalar,
geometric tilt vector, and transverse anisotropy tensor-reveals that the leading $\mathcal{O}(\varepsilon)$ correction to the Oseen–Frank director field $\mathbf{n}_0(\mathbf{x})$ is dominated by a non-vanishing tilt field $\mathbf{p}_1(\mathbf{x})$, where $\varepsilon$ represents the ratio of the nematic coherence length to the characteristic domain size. This macroscopic variation constitutes a soft mode released from the boundary once the surface anchoring energy is retained at its physical scaling rather than driven to an infinite strength. We show that this tilt field is governed by the linear Jacobi equation, $\mathcal{J}_{\mathbf{n}_0}(\mathbf{p}_1)=\mathbf{0}$, subject to a non-trivial, anchoring-driven Dirichlet boundary condition, where $\mathcal{J}_{\mathbf{n}_0}$ is the on-shell Jacobi operator of the harmonic map $\mathbf{n}_0$ on $\mathbb{S}^2$. The two fields are accompanied at $\mathcal{O}(\varepsilon^2)$ by an off-shell correction to both the uniaxial scalar and the transverse anisotropy tensor, passively induced by the elastic non-uniformity $(\nabla\mathbf{n}_0\neq\mathbf{0})$ and the boundary-driven tilt $(\mathbf{p}_1\neq\mathbf{0})$. Through $\mathcal{O}(\varepsilon^2)$, the tilt enters the energy only through surface terms, not the bulk, providing a pathway for the system to lower its energy . Under the conventional
benchmark of rigid Dirichlet conditions, this response is annihilated outright, demonstrating that corrections built upon infinite energy barriers obscure the underlying physics of anchoring-driven tilt modes.
Read the full abstract ▾
Image and Video Processing
Memory- and Bandwidth-Efficient SPAD-LiDAR Ranging via Coarse-to-Fine Spline Sketching
Zhenya Zang, Istvan Gyongy and Mike Davies
This work presents an optimized compression framework for direct time-of-flight (dToF) light detection and ranging (
LiDAR), using an accurate, compact timestamp-encoding strategy to address the high data-rate bottleneck from single-
photon timing information in single-
photon avalanche diode (SPAD) arrays. We propose a hardware-friendly timestamp-to-depth framework in which a sparse single-
photon encoding strategy, namely coarse-to-fine spline sketches (CFSS), projects
photon timestamps into fine-grained, low-dimensional representations, called sketch values. CFSS converts timestamps on-the-fly to sketch values and accumulates them with lower
memory consumption, without constructing histograms, thereby saving on-chip/device
memory. CFSS's closed-form solution for ToF
retrieval ensures iteration-free reconstruction. Compared with the conventional sketched-
LiDAR framework, the proposed method retains the same compression ratio in the single-peak case, ranging from hundreds$\times$ to thousands$\times$ depending on the accuracy-complexity trade-off, while improving depth-estimation accuracy. Both synthetic and real
datasets are used to validate the performance. The CFSS framework is further extended to two-peak scenarios, enabling recovery of objects obscured by partially occluding camouflage and uniformly scattering semi-transparent media. Firmware implementation is also investigated by clearly separating online hardware processing from offline software workloads.
Read the full abstract ▾
Mesoscale and Nanoscale Physics
Capacitance sensing in bilayer graphene with gate reflectometry
Tamás Kalmár, Gergő Fülöp, Bálint Szentpéteri et al. (+5 more)
In the search for topological states in
bilayer graphene, capacitance measurement methods are widely utilized in bridge type readouts or source-drain reflectometry setups. In this paper we demonstrate an alternative readout method based on gate reflectometry. We probe the capacitance of a BLG flake encapsulated between hBN and WSe$_2$ through a lumped-element LC resonator coupled to the top gate electrode of the
heterostructure. The microwave readout reliably tracks the opening of a displacement field-induced band gap in the device. Compared to simultaneous DC transport measurements, the RF signal proves to be less affected by the percolating transport channels caused by sample disorder. Furthermore, we use the gate based readout to identify a small-angle
alignment between the hBN and BLG flake. From the fixed frequency microwave readout, we extract the device capacitance change between the insulating and metallic state of the
bilayer and utilize a comprehensive capacitance network model to explain its origin. Finally, we propose an advanced, multiplexed double-resonator readout scheme capable of simultaneously measuring top and bottom gate capacitances to extract the layer polarizability of the system. This work establishes gate reflectometry as a highly scalable and powerful methodology for exploring topological phases and internal charge dynamics in van der Waals
heterostructures.
Read the full abstract ▾
All authors Tamás Kalmár, Gergő Fülöp, Bálint Szentpéteri, Olivér Kürtössy, Kenji Watanabe, Takashi Taniguchi, Szabolcs Csonka, Péter Makk
Probability
A non-trivial dynamics on the directed landscape
Manan Bhatia
We prove the existence of a non-trivial continuous stationary reversible dynamics on the directed landscape as a subsequential scaling limit of Brownian last passage percolation under the Ornstein–Uhlenbeck dynamics. Strong passage-time stability estimates from the companion paper Bhatia '26, together with static endpoint regularity, give tightness at the critical dynamical scale $n^{-1/3}$. To establish non-triviality of the subsequential limiting dynamics, we show that, in dynamical BLPP, the fractional overlap between geodesics at times zero and $a n^{-1/3}$ tends to one as first $n\to\infty$ and then $a\downarrow0$. The proof refines the excursion argument of Ganguly-Hammond '24, using strong passage-time stability and the fact that a fixed dynamical time is increasingly subcritical at finer spatial scales.
Read the full abstract ▾
Classical Analysis and ODEs
Sharp weighted $L^1$ bounds for Coulomb fields on the sphere
Evgueni Doubtsov, Ioann Vasilyev and Grigory Voinov
Let $d \ge 2$, let $\alpha_1,\dots, \alpha_n > 0$, and put $A = \sum_k \alpha_k$. We determine, up to constants depending only on $d$, the smallest possible $L^1$ norm in the unit ball $\mathbb{B}^d\subset \mathbb{R}^d$ of the Coulomb field generated by charges of strengths $\alpha_k$ placed on the unit sphere $\mathbb{S}^{d-1}$. More precisely, we prove \[ \inf_{x_1,\dots,x_n \in \mathbb{S}^{d-1}} \int_{\mathbb{B}^d} \left| \sum_{k=1}^n \alpha_k \frac{x_k-x}{|x_k - x|^d} \right| dx \asymp_d A^{-1/(d-1)} \sum_{k=1}^n \alpha_k^{d/(d-1)}. \] The lower bound follows from an $L^1$ Lipschitz trace estimate and an explicit weighted tent function on the sphere. For the upper bound we use a partition of the sphere into
cells of masses $\alpha_k/A$ and diameters $O_d \left((\alpha_k/A)^{1/(d-1)} \right)$. For the unit weights the sharp order is $n^{(d-2)/(d-1)}$, and in dimension three it is $\sqrt{n}$.
Read the full abstract ▾
Category Theory
More on action representability and normalizers
James Richard Andrew Gray
We show that under mild assumptions a functor is a prefibration if and only if it has terminal objects in its fibers and admits precartesian liftings of monomorphisms. We use this to show that the following conditions on a pointed category $\mathbb{C}$ are equivalent: (a) The category of morphisms of $\mathbb{C}$ admits generic split extensions; (b) Each functor category of $\mathbb{C}$ with finite domain category admits generic split extensions; (c) $\mathbb{C}$ admits normalizers and generic split extensions; (d) The kernel functor from the category split extensions in $\mathbb{C}$ to $\mathbb{C}$ is a prefibration. In addition we show that the category of morphisms of a pointed protomodular $\mathbb{C}$ admits generic split extensions if and only if for each morphism $f:X\to Z$ the functor sending each object $B$ to the isomorphism class of split extensions in $\mathbb{C}^{\mathbf{2}}$ of $(B,B,1_B)$ with kernel $(X,Z,f)$, is representable.
Read the full abstract ▾
Machine Learning
A Multimodal Autonomic Sensing Framework for Objective Assessment of Patient Responses to Dental Pulp Stimulation
Youngsun Kong, Yubin Choi, Dongjin Song et al. (+3 more)
Patient responses to dental pulp testing, ranging from no sensation to intense pain, provide important information for assessing pulp status in endodontic diagnosis. However, pain is a subjective sensory and emotional experience that varies considerably across individuals and can be difficult to communicate. We investigated whether complementary autonomic signals could support objective assessment of responses during dental examination. Forty-nine patients underwent cold pulp testing, yielding no-response, mild-response, and intense-response conditions. The framework integrated ECG-derived skin nerve activity (SKNA) and R-R intervals (RRI), together with electrodermal activity (EDA), using temporal convolutional network encoders with
attention-based mid-level fusion. Individual baseline signals and subject-level
covariates, including anxiety scores and biological sex, were also incorporated. The framework achieved 80.2% balanced accuracy, 75.2% sensitivity, and 85.2% specificity for binary classification of no response versus mild or intense response. For three-class classification, it achieved 60.0% balanced accuracy and a 58.8% macro-averaged F1 score. Ablation and
attention-weight analyses indicated that EDA contributed most strongly to model performance, followed by RRI, while SKNA improved balanced accuracy by approximately five percentage points. Age was significantly associated with model performance. These findings support the feasibility of
multimodal autonomic sensing for objective, non-invasive assessment of responses to dental pulp stimulation.
Read the full abstract ▾
All authors Youngsun Kong, Yubin Choi, Dongjin Song, Dong-Guk Shin, I-Ping Chen, Ki Chon
Computer Vision
Style-Driven Data Synthesis and Degradation-Aware Enhancement for Ultrasound Image Restoration
Yu-Kai Wang, Chun-Xin Tan, Manh-Hung Nguyen et al. (+1 more)
Low-cost handheld ultrasound devices can be widely deployed compared to professional hospital ultrasound machines. However, their images suffer from compound degradation that can mislead clinical judgment. Motivated by this observation, mapping handheld low-quality (LQ) to hospital high-quality (HQ) images has been considered a valuable research question. Conventionally, the mapping requires pixel-aligned LQ-HQ pairs. This requirement is unsatisfactory in practical scenarios because real scans at different times are never pixel-aligned. This paper addresses the challenge with a two-stage framework. The first stage generates pixel-aligned LQ-HQ
datasets, and the second stage trains an enhancement model that improves LQ images. The first stage trains a cycle-consistent style-transfer model on unaligned real LQ-HQ pairs to learn a HQ-to-LQ model. Then, the model transforms real HQ images into pixel-aligned LQ images. Based on the
dataset generated by the first stage, the second stage uses the Dual Degradation-Guided (DDG) Low-Rank Adaptation (LoRA) method to fine-tune an LQ-to-HQ model based on aligned pairs. In this stage, the model is based on the well known PiSA-SR framework but inserts a degradation-conditioned correction
matrix. Experimental results on the USenhance2023
dataset show that the FID metric is improved by 16.7% over the strongest baseline while other metrics indicate that our enhanced outputs are well aligned with the real HQ distribution. The source code of our method is available at https://github.com/Jason0411202/DDG_LoRA.
Read the full abstract ▾
All authors Yu-Kai Wang, Chun-Xin Tan, Manh-Hung Nguyen, Ching-Chun Huang
Algebraic Topology
Gros Topoi as Partially Lax Limits of Petit Topoi
Fabio Neugebauer and Qi Zhu
We prove that the sheaf $\infty$-topos associated to a
geometric site in the sense of Lurie can be written as a partially lax limit of smaller sheaf $\infty$-topoi. This is thus a formalization of Lurie's
vision for fractured $\infty$-topoi to axiomatize the relation between gros and petit topoi. As a consequence, we realize the $\infty$-topos of $\kappa$-small condensed anima as a partially lax limit of sheaf $\infty$-topoi over extremally disconnected spaces, marked at the open
embeddings. Moreover, we deduce a gros version of the étale exodromy
theorem.
Read the full abstract ▾
Sound
RemixIT-TSE: Progressive Synthetic-to-Real Adaptation for Target Speech Extraction via Target-Aware Supervision and Remixing
Yu Wang, Haixin Guan, Shuang Wei et al. (+1 more)
Target
Speech Extraction (TSE) in real-world conversational scenarios suffers from severe performance degradation due to the domain gap between synthetic training data and complex acoustic environments, where signal-level ground truth is typically unavailable. To address this challenge, we make the first attempt to extend RemixIT from
speech enhancement to TSE and propose a progressive synthetic-to-real adaptation framework for real-world TSE with two fine-tuning stages. The first stage leverages region-wise speaker similarity and silence constraints within a target-aware adaptation framework to jointly optimize the model using synthetic and weakly supervised real-world data, injecting real-world traits while preserving synthetic-learned capabilities. The second stage further adapts the model using only real-world data through our RemixIT-TSE, where quality-filtered teacher pseudo targets, which guarantee reliable student training, provide signal-level supervision via SI-SNR loss. Experiments on the real conversational evaluation set (EVAL-2) of the SLT 2026 REAL-TSE Challenge, the proposed method achieves a 6.53% relative TER reduction, together with relative improvements of 21.84% in speaker similarity, 9.89% in DNSMOS-P808, and 4.10% in target-activity F1 over the source-domain baseline, demonstrating its effectiveness under unseen real-world conditions. Source code at https: //github.com/YuWang-
Speech/RemixIT-TSE.
Read the full abstract ▾
All authors Yu Wang, Haixin Guan, Shuang Wei, Yanhua Long
Artificial Intelligence
Tool Mediation Alters Refusal Mechanisms in Large Language Models
Abel Rodríguez, Giuseppe Garofalo, Lieven Desmet et al. (+1 more)
Large
language models (LLMs) are increasingly deployed with access to external tools, yet harmful tool-mediated interactions are less likely to be refused when compared to regular conversational ones. As this change in refusal behavior remains underexplored, we investigate its underlying mechanisms across a diverse set of open-weight
language models. We find that information about the harmfulness of a request remains strongly encoded in the model's representations and transfers across conversational and tool-mediated inputs. Evidence from representation geometry and
neuron-level analysis further indicates that the two interaction modes systematically distribute harm-related computation differently. Crucially, while conversational inputs can be refused at relatively low levels of perceived harmfulness, tool-mediated inputs remain permissive until harmfulness crosses a substantially higher effective refusal threshold. Moreover, tool-mediated refusal is also more brittle: progressively weakening the refusal computation disrupts tool-mediated refusal at lower intervention strengths than conversational refusal, even when benign capabilities remain intact. Together, our findings indicate that tool mediation does not simply reduce the internal perception of harm, but instead impacts its conversion into refusal. Overall, this suggests tool-mediated environments may intrinsically reduce robustness of models to harmful requests, and that conventional safety evaluations may not fully transfer to LLM
agents.
Read the full abstract ▾
All authors Abel Rodríguez, Giuseppe Garofalo, Lieven Desmet, Vera Rimmer
Chemical Physics
Non-equilibrium cavity pumping of electronic molecular polaritons
Yassir El Moutaoukal, Rosario R. Riso and Henrik Koch
Strong light-matter coupling in optical cavities offers a non-invasive route to modify molecular properties, and it is usually reached by collectively coupling many molecules to the same mode. Here we consider the photonic counterpart of the collective regime, in which a single molecule interacts with a cavity mode holding many
photons, a steady state of the pumped cavity. Since the eigenstates of the Pauli-Fierz
Hamiltonian carry no transverse electric field, such a state cannot be described as an excited state of the light-matter system. We therefore constrain the
photon number of the strong coupling QED Hartree-Fock wave function with a Lagrange multiplier and self consistently minimize the energy with respect to all parameters. The constrained reference carries the field of a coherent state with an average number of
photons, generated by a cavity mode whose effective free-field frequency is lowered by the multiplier. We apply this framework to benzene and a benzene-water complex, where the pumped cavity polarizes the molecules and modifies their interaction as with a static classical field. For hydrogen peroxide, the field can reshape the torsional potential and the far-infrared torsional spectrum. The classical driven field limit thus emerges from a quantized mean field description, and the
photon density becomes a control parameter for cavity-modified chemistry.
Read the full abstract ▾
Artificial Intelligence
DuplexCadence: Exact State and Execution from a Speech Model's Declared Timelines
Haixiao Gao, Yimin Zheng, Linyou Xiao et al. (+1 more)
Full-duplex
speech models support streaming interaction that listens and speaks at the same time. Serving them is governed by a strict, repeating deadline: conversation advances on a one-second cadence, and every second of input must be turned into a second of
speech before the next second arrives. Because stages within a session run in strict sequence, per-invocation overhead cannot be batched away. Profiling reveals that the autoregressive stages of a duplex second already fit within the period, whereas the
token-to-audio synthesis tail is what causes overruns. This tail stage suffers from orchestration slack where the GPU is left waiting as thousands of tiny, regular operations are issued one by one, while also wasting substantial
memory by over-provisioning state at static implementation constants. Existing remedies, such as graph recording and demand-sized allocation, fail because streaming state dynamics violate their prerequisites. The root cause is that the runtime lacks the model's native clocks: the per-region counters that govern advancement rates and retention policies. We propose DuplexCadence, which explicitly declares native clocks to the runtime and derives two mutually enabling rules: demand-sized state allocation at a stable address, and exact-shape graph replay without padding. The former eliminates idle
memory and stabilizes tensor pointers, while the latter removes orchestration slack without padding overhead. Evaluated on four released models across three decoder architectures with bit-for-bit identical output, DuplexCadence reaches $2.85\times$ the stock runtime's speed at $38.8\%$ lower peak
memory. On the live duplex path, mean SPEAK time falls from $14\%$ over the one-second cadence to $2\%$ under it, enabling models to reliably keep up with interactive
speech while markedly expanding multi-
Read the full abstract ▾
All authors Haixiao Gao, Yimin Zheng, Linyou Xiao, Zeke Xie
Quantum Physics
Accurate and bounded approximation of quantum correlation functions
Alexander van Lomwel and Florian Mintert
The exact classical simulation of non-integrable
quantum systems rapidly becomes intractable beyond small system sizes, yet its role is central to predicting many-body dynamics. Approximate methods alleviate this cost but typically lack rigorous guarantees on their deviation from the exact result. For the infinite-temperature correlation function, we derive a rigorous error bound for approximating the full dynamics within a computable finite-sized region. Applied to autocorrelations, this finite-region approach is especially accurate across one-dimensional systems and beyond, and can further provide rigorous error bounds for other approximate methods, demonstrated here for
matrix-product-operator simulations.
Read the full abstract ▾
Machine Learning
SymbolicArena: A Unified Infrastructure for Benchmark Distillation and Dynamic Evaluation in Symbolic Regression
Ziwen Zhang, Xiju Wu, Yuheng Jing et al. (+9 more)
Symbolic regression (SR) seeks concise and interpretable mathematical expressions from data for scientific equation discovery. Existing SR
benchmarks face a tradeoff between evaluation cost and
benchmark validity. Repeated evaluation of large task pools is expensive, and compact
benchmarks lack systematic evidence of preserved task diversity and
algorithm discriminability. SymbolicArena provides a unified infrastructure for
benchmark distillation and dynamic evaluation. The framework standardizes 664 heterogeneous tasks with executable ground truth expressions and distills the Full Task Set into Core50, a validated
benchmark of 50 tasks. The
distillation process preserves task coverage and
algorithm discrimination under explicit balance constraints. SymbolicArena applies a unified execution protocol to heterogeneous SR
algorithms and produces comparable outputs and search trajectories. Multi Axis Evaluation characterizes numerical quality, symbolic quality, and search behavior. Core50 reduces evaluation workload by 92.5% and maintains agreement with Full Task Set evaluations. Experiments show that SymbolicArena achieves 72.6% to 86.7% lower approximation error than alternative selectors, further supporting its fidelity to the Full Task Set. Evaluation reveals a substantial gap between numerical fitting and symbolic recovery across current SR methods, suggesting that reliable equation recovery remains an open challenge.
Read the full abstract ▾
All authors Ziwen Zhang, Xiju Wu, Yuheng Jing, Runxiang Wang, Boxiao Wang, Yifan Zang, Yifan Zhang, Yang Wang, Kai Li, Yifan Zhang, Huilin Xu, Jian Cheng
Quantum Physics
From Quantum to Classical Dissipative Spin Dynamics
Alireza Ataei, Leila Ataei, Olle Eriksson et al. (+1 more)
The Landau-Lifshitz-Gilbert (LLG) equation provides the fundamental description of dissipative magnetization dynamics. A recently proposed
quantum analog, the
quantum Landau-Lifshitz-Gilbert (QLLG) equation, extends this framework to
quantum spin systems. However, whether the QLLG equation recovers the LLG equation in the classical large-
spin limit for interacting
spin systems remains an open question. Here, we demonstrate this connection by deriving the LLG equation as the classical large-
spin limit of the QLLG equation. We establish this correspondence analytically and verify it numerically for an interacting
spin system. These results provide a
quantum-to-classical description of dissipative
spin dynamics, offering a microscopic foundation for the LLG equation. From numerical calculations provided here, we also conclude that
quantum effects become much less noticeable for systems with larger number of
spins in the
Hamiltonian.
Read the full abstract ▾
All authors Alireza Ataei, Leila Ataei, Olle Eriksson, Vahid Azimi Mousolou
General Economics
The missing price feedback: Why studies overstate local grid peaks from synchronized home batteries under real-time pricing
Lion Hirth
Existing studies consistently find that home batteries optimized against real-time electricity prices create new load peaks in local distribution grids. This is because all assets respond to the same price signal in sync. However, most of these studies are simulations, and they all treat wholesale prices as exogenous, i.e., unaffected by the battery dispatch. This paper argues that assuming prices to be non-responsive biases the result: in reality, charging in low-price hours raises the wholesale price, which dampens the incentive to charge. In
equilibrium, charging is spread out across hours. I test this by simulating households with rooftop
solar and home batteries under a spot-based retail tariff with a model of
equilibrium wholesale prices, calibrated to German data for 2025. With exogenous prices, my results confirm the literature findings: synchronous charging pushes the coincident import peak of the local grid to 83 % above its no-battery level. With endogenous prices, the same fleet leaves the peak 6 % below the no-battery
benchmark. This finding is robust across alternative price functions and exogenous price paths, different measures of the coincident peak, 22 scenario variations and three historical years. I conclude that while new local peaks remain possible, particularly in grids that contain local clusters of home batteries, the risk and magnitude are considerably smaller than the existing literature suggests.
Read the full abstract ▾
Artificial Intelligence
Sol-H3: Recursive Self-Improvement for MiniMax-H3 Inference Acceleration on Sol-Engine across Cloud and Edge
Yitong Li, Jincheng Yu, Junsong Chen et al. (+6 more)
Video diffusion models are rapidly scaling and exhibiting enhanced generation capabilities. Among these recent advancements, MiniMax-H3 stands out as a highly capable, production-level open-source model. However, its 33-billion parameters and multi-step iterative denoising process introduce substantial computational overhead. Consequently, their practical production is hindered by generation latency in the cloud deployment like NVIDIA-GB200, alongside strict
memory limits that pose further challenges at the edge device like DGX-Spark. To address these diverse hardware bottlenecks from cloud to edge device, we present a full-stack inference pipeline that integrates efficient algorithmic design with optimized operator implementations. Algorithmically, we introduce a cross-resolution two-stage generation scheduler that exploits the step-wise nature of
diffusion: early low-resolution steps rapidly establish the global layout, while later high-resolution steps focus refinements of local and perceptual details. These stages are connected by a learned latent-to-latent mapping module, completely eliminating the computationally expensive VAE decode-reencode cycle for resolution transferring cross different resolutions. For operator implementation, we deploy a Recursive Self-Improvement (RSI) loop that searches kernel fusions and
memory layouts, evaluating latency together with numerical agreement. Together, these optimizations deliver up to 30x end-to-end speedup and 20% lower
memory: a 5-second 1344x768
video with audio is generated 3.5x faster than real time on an 8xGB200 node, and in under a minute fully
memory-resident on a single DGX Spark.
Read the full abstract ▾
All authors Yitong Li, Jincheng Yu, Junsong Chen, Haopeng Li, Shuchen Xue, Haozhe Liu, Ping Luo, Song Han, Enze Xie
Artificial Intelligence
Using Context Is Not Enough: Test-Time Training for Personalized Reward Modeling
Bohao Wang, Xiaoyan Zhao, Yang Zhang et al. (+4 more)
Reinforcement learning from human feedback (RLHF) aligns large
language models (LLMs) with human preferences, yet most pipelines learn a single reward model that overlooks individual differences in preferences. Personalized reward models (PRMs) address this by conditioning rewards on user-specific feedback, most commonly through in-context learning (ICL), where a user's historical comparisons are supplied as contextual preference pairs. However, we identify a key limitation of ICL-based PRMs: they fail to capture the preference relations conveyed by contextual pairs. To address this, we propose Preference-Aligned Test-Time Training (P-TTT), which explicitly encodes these relations into user-specific fast weights for personalized reward prediction. P-TTT introduces sequence-level update and apply operations to match the response-level granularity of preference feedback, together with a preference-aligned objective that directly uses pairwise preference relations to guide fast-weight adaptation. Notably, P-TTT is simple to implement and computationally efficient, updating fast weights within a single forward pass without inference-time backpropagation. Extensive experiments show that P-TTT more effectively captures historical preference relations and outperforms state-of-the-art methods by a large margin.
Read the full abstract ▾
All authors Bohao Wang, Xiaoyan Zhao, Yang Zhang, Jinghang Guo, Chun Chen, Can Wang, Jiawei Chen
Computation and Language
A mechanistic study of language model introspection
Jiahong Zou, Xiangkun Sun, Lingkai Kong et al. (+1 more)
Large
language models (LLMs) can sometimes report perturbations to their internal activations—even when the input provides no evidence that an intervention occurred. How do models detect and localize such internal changes? We study this question using a controlled task that keeps the input text fixed. We either inject a concept vector into the hidden state at one of ten
token positions or apply no intervention. The model is asked to identify the perturbed position or report that no intervention occurred. Across three model families, we identify two small groups of
attention heads with distinct roles in introspective reporting. Middle-layer gate heads influence whether the model reports a change, while router heads in a later layer help select the position to report. Interventions on gate heads can suppress position reports even when router heads supply location information. We further examine why reporting accuracy varies across concepts. Concept vectors that are localized more accurately produce stronger
attention-score and output responses in gate heads, which is associated with better
alignment of the induced key and value changes in their QK and OV computations. Together, these findings identify
attention-head mechanisms supporting introspective detection and localization.
Read the full abstract ▾
All authors Jiahong Zou, Xiangkun Sun, Lingkai Kong, Tonghan Wang
Artificial Intelligence
DoAtlas-2: A Foundation for Self-Evolving Causal Biomedical Discovery
Yulong Li, Rong Xia, Yuxuan Zhang et al. (+12 more)
We introduce DoAtlas-2, a foundation for self-evolving causal biomedical discovery that organizes knowledge around causal mechanisms and advances through external evidence from human populations. DoAtlas-2 integrates 771 research resources covering more than 720,000 participants in 48 countries, from longitudinal clinical phenotypes, medical imaging, and continuous physiological signals to eight molecular layers, together with an evidence network of approximately 4.7 million literature-derived records over 93,566 concepts and 149,383 candidate causal relations. DoAtlas-2 autonomously formulates research questions from evidence gaps and unresolved mechanisms, prespecifies their causal designs, and generates validated analyses. Supporting, challenging, and unresolved results continuously revise mechanistic interpretations, the causal evidence state, and the discovery frontier, so that DoAtlas-2 self-evolves within a closed loop of hypothesis generation, empirical testing, and renewed discovery. DoAtlas-2 has systematically evaluated 2,031 research questions. In the Human Phenotype Project (HPP), it formulated 4,014 candidate pathway questions across vascular, early-glycemic, and hepatic-metabolic systems, and screening of the first 1,079 yielded statistical support for 756. Representative studies identify blood pressure as a convergence node linking adiposity, hepatic, and lipid phenotypes to vascular outcomes, and show that an adiposity-inflammation-blood-pressure pathway is largely attenuated by joint adjustment for body mass index (BMI) and smoking. The discovered vascular network constitutes a completely interpretable predictive foundation, admitting exact attribution of every prediction and closed-form mediation effects. DoAtlas-2 thereby unifies causal mechanism discovery, population-evidence testing, and interpretable prediction within one continuously evolving foundation.
Read the full abstract ▾
All authors Yulong Li, Rong Xia, Yuxuan Zhang, Jianxu Chen, Xiwei Liu, Haochen Xue, Maosheng Li, Yuhang Liu, Yibo Yuan, Yutong Xie, Chong Li, Jionglong Su, Hagai Rossman, Eran Segal, Imran Razzak