High Energy Physics (Phenomenology)
A global analysis of ${\bar B}\to DP$ and ${\bar B} \to {\bar D} P$ decays
Bhubanjyoti Bhattacharya, Rida Boumris and David London
Recently, a fit to the 15 measured branching ratios of ${\bar B} \to D P$ decays ($B = \{B^+, B^0, B_s^0\}$, $D = \{D^0, D^+, D_s^+\}$, $P \in \{\pi, K\}$) was performed under the assumption of flavor SU(3)
symmetry [SU(3)$_F$]. A terrible fit was found: there was a $> 5\sigma$ disagreement with the SU(3)$_F$ limit of the Standard Model [SM$_{\rm{SU(3)}_F}$]. It was also found that this discrepancy can be removed by adding $\sim 20\%$ SU(3)$_F$-breaking effects. Now, the ${\bar B} \to D^0 P$ amplitudes can contribute to other observables (direct and indirect CP asymmetries) if they interfere with the amplitudes for ${\bar B} \to {\bar D}^0 P$ decays. This happens if the $D^0$ and ${\bar D}^0$ decay to the same final state (e.g., $K^+ K^-$, $\pi^+ \pi^-$, $K^\pm \pi^\mp$, etc.). In this paper, we perform a combined fit of ${\bar B} \to D P$ and ${\bar B} \to {\bar D} P$ decays, including all observables. We again find a $> 5\sigma$ disagreement with the SM$_{\rm{SU(3)}_F}$, and we find that the data can be explained if $\sim 60\%$ SU(3)$_F$-breaking effects are included.
Read the full abstract ▾
Computer Vision
Can Learning to Answer Spatial Questions Improve Novel-View Generation?
Jahyeok Koo, Hyeonseo Yu, Seungho Jang et al. (+4 more)
Generating novel views from limited observations requires preserving both the appearance and spatial structure of a scene. Although recent approaches exploit fine-grained
geometric cues, they do not explicitly model high-level semantic and spatial properties such as object identity, attributes, and inter-object relationships. In this work, we investigate whether 3D
language understanding can provide complementary supervision for novel-view generation. We introduce Ask4NV, built on a pretrained unified
multimodal model with a Mixture-of-
Transformers (MoT) architecture. Our model incorporates camera geometry into both VAE and ViT representations and leverages understanding features during generation through joint self-
attention with a modality-aware
attention masking strategy. Experiments demonstrate strong image quality and
geometric consistency in novel-view generation, alongside competitive 3D
language understanding. Further analysis shows that joint training improves cross-view correspondence in VAE representations and similarity in a 3D-aware feature space.
Read the full abstract ▾
All authors Jahyeok Koo, Hyeonseo Yu, Seungho Jang, Jaewoo Jung, Jiwon Kang, Seokju Cho, Seungryong Kim
Robotics
SafeDriveVLA: Navigation-Conditioned World Model Dreaming for Conflict-Aware End-to-End Autonomous Driving
Shaoyuan Xie, Zihan Zhang, Jingxuan Wang et al. (+6 more)
Vision-
Language-Action (VLA) models are gaining
attention in autonomous driving, inheriting commonsense priors from large
language models (LLMs) and widening the navigation interface from a fixed command set to free-form natural-
language instructions. However, this open-ended interface exposes a new vulnerability surface: an inattentive driver may issue instructions that are unsafe given the current scene, and a malicious user may issue dangerous ones on purpose. We present the first
benchmark suite that evaluates navigation following and conflict awareness together, along with
adversarial robustness. Existing VLAs fall short on both: SimLingo completes no
language-instructed lane change in closed loop, and all models degrade sharply under conflicting signals. We propose SafeDriveVLA, the first driving VLA that (1) explicitly predicts a three-level navigation-compliance mode to decouple conflict
reasoning from action generation, and (2) uses navigation-conditioned world model dreaming to inform this prediction. Unlike prior world models that roll out expert or planned actions, ours dreams of strictly following the navigation signal, showing when following is safe. SafeDriveVLA reaches 88.16 Driving Score (DS) and 68.64% Success Rate (SR) on Bench2Drive, lifts closed-loop
language following by 49.3% and reduces collisions under unsafe instructions from 60 to 38 compared to the best SimLingo variant.
Video demonstrations are available on our project page.
Read the full abstract ▾
All authors Shaoyuan Xie, Zihan Zhang, Jingxuan Wang, Jiashu Qu, Xiaoqing Liang, Lingdong Kong, Junchi Lu, Henrik I. Christensen, Qi Alfred Chen
Computer Vision
Representation Conditioning for Generation
Saeed Razavi, Xingjian Leng and Liang Zheng
Representation-
alignment methods effectively accelerate the
diffusion transformer training. REPA, for example, uses a pretrained
vision encoder to supervise an intermediate layer through an MLP projection head which is discarded after training. It aims to import semantics from the
vision encoder to the
diffusion backbone. While this objective is achieved at early layers by back-propagating the
alignment loss, there is no guarantee that later layers, supervised by the velocity prediction loss alone, can well access such semantics. We propose a simple solution, representation conditioning (RECON), to improve semantics in later layers. During training, we feed the pretrained
vision encoder
tokens into later layers; during inference, because we no longer have the clean image, we reuse the MLP head (as a proxy to the
vision encoder) and project its output
tokens to the later layers. Our method adds no parameters and leads to much faster convergence: RECON is $63\times$ faster than SiT and $25\times$ faster than REPA. Code is available at https://github.com/saeidrazavi/RECON
Read the full abstract ▾
Robotics
Hierarchical Residual Policy Learning for Real-World Mobile Manipulation with Sparse Human Guidance
Zhefei Gong, Hefeifei Jiang, Zehao Wang et al. (+1 more)
Mobile Manipulation (MoMa) is a cornerstone of autonomous robotic systems, yet real-world fine-tuning remains a formidable challenge because of high-dimensional action spaces and intricate coupling between heterogeneous modules. In this paper, we propose HiRe-MoMa (Hierarchical Residual Mobile Manipulation), a systematic framework for stable and sample-efficient real-world fine-tuning of complex MoMa systems. At its core, HiRe-MoMa adopts a Hierarchical Residual Architecture that fine-tunes a frozen base policy. A Policy-Agnostic Residual learns additive corrections atop the base policy, while Hierarchical Decoupling splits the residual into mobile-base and manipulator-arm heads with unidirectional conditioning, mitigating gradient interference and aligning corrections with the platform's physical hierarchy. To improve sample efficiency in data-expensive real-world settings, we propose Expert-Guided Value
Alignment, which leverages sparse human interventions with under 20% training coverage to reshape the critic's value landscape, effectively guiding autonomous exploration through expert priors. Extensive real-world experiments on the TIAGo++ platform demonstrate that HiRe-MoMa consistently enhances task performance with only around 30 minutes of online interaction, regardless of the base policy architecture. To our knowledge, this is among the first frameworks to enable efficient on-
robot fine-tuning for mobile manipulation in the real world, providing a practical paradigm for deploying complex robotic systems in dynamic, human-centric environments. Further details can be found on our project website: https://hire-moma.github.io/
Read the full abstract ▾
All authors Zhefei Gong, Hefeifei Jiang, Zehao Wang, Jiachen Li
Robotics
Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion
Huixuan Cheng, Bike Zhang, Shankar Sastry et al. (+1 more)
Recent advances in
robot learning have driven rapid progress in humanoid control, enabling increasingly agile behaviors in the real world. However, humanoid control networks are typically obtained through conventional training procedures, while comparatively little
attention has been given to directly modeling the distribution of their learned weights and generating functional networks in weight space. In this work, we investigate whether high-performing humanoid control networks can be generatively modeled directly through their weights. We adopt a latent
diffusion framework that couples a linear autoencoder with a
Diffusion Transformer. The autoencoder compresses actor network parameters into a compact latent space, and the
Diffusion Transformer models the distribution of these latents. Trained on networks collected from reinforcement learning for the Unitree G1
robot, our approach generates deployable networks that perform various motions in the real world, including walking, dancing, and
spin kick. Beyond generation, we further study the learned weight space to better understand the structure captured by these humanoid control networks. We identify a structural principle in the first-layer weights that consistently reflects the
robot's kinematic prior. Notably, this kinematic prior emerges under both Reinforcement Learning (RL) and Imitation Learning (IL) training, suggesting that it is not specific to a particular optimization process but may instead reflect common knowledge associated with humanoid embodiment. Together, these results demonstrate the feasibility of generative modeling in humanoid network weight space, while providing initial insights into the information encoded by their learned weights. Additional results are available on our project website: https://humanoid-weight-
diffusion.github.io/
Read the full abstract ▾
All authors Huixuan Cheng, Bike Zhang, Shankar Sastry, Koushil Sreenath
Superconductivity
Nodal nematic superconductivity in moiré graphene
Shuwen Sun, Grgur Palle, Kenji Watanabe et al. (+4 more)
Unconventional superconductivity can emerge in various forms in strongly correlated systems. Historically, the most extensively studied examples include nodal and triplet pairing, which were experimentally detected in cuprates and superfluid helium-3, respectively. Beyond these canonical cases, there are also nematic
superconductors, in which the rotational
symmetry of the
lattice is broken in the superconducting state. While there are several examples of nematic
superconductors whose pairing emerges inside a preexisting nematic normal state, experimental evidence for superconducting states that intrinsically break rotational
symmetry on their own has been exceedingly scarce. Here, we report the experimental realization and identification of an intrinsic multi-component nodal nematic superconducting state in magic-angle twisted trilayer
graphene (MATTG). By integrating tunneling spectroscopy with
quantum transport measurements, we directly probe the
symmetry of low-energy superconducting quasiparticles. Deep inside the superconducting state, we observe a pronounced two-fold anisotropy in the zero-bias tunneling conductance under an in-plane
magnetic field, which is strongly suppressed together with superconductivity as the temperature or perpendicular
magnetic field is increased. Remarkably, the major axis of anisotropy rotates by up to $90^{\circ}$ across the superconducting dome, indicating an intrinsic, field- and density-tunable
symmetry of the low-energy excitations. These behaviors imply a nodal superconducting state in MATTG that spontaneously breaks the six-fold rotational
symmetry of the moiré superlattice. Our results establish moiré
graphene as a versatile platform for realizing and interrogating nodal nematic superconductivity, opening a new avenue for exploring unconventional
symmetry-breaking superconducting states in two-dimensional materials.
Read the full abstract ▾
All authors Shuwen Sun, Grgur Palle, Kenji Watanabe, Takashi Taniguchi, Rafael M. Fernandes, Pablo Jarillo-Herrero, Jeong Min Park
Computer Vision
VITAL: From Visual Motion to Precise Action
Ao Liang, Youquan Liu, Mingxiao Li et al. (+7 more)
World-action models aim to turn visual anticipation into driving decisions, yet typically train action experts on trajectory representations defined independently of visual dynamics. This motivates deriving action targets directly from visual motion. We first examine a simpler setting in which trajectories are reconstructed from recorded future images. Even without the need for future prediction, reconstruction captures the dominant maneuver but leaves centimeter-scale errors. We therefore introduce VITAL (
Vision-Induced
Tokens for Action Learning), an action tokenizer that couples a visually derived motion scaffold with a compact residual encoding the trajectory error left by that scaffold. This representation makes visual motion the principal prediction target and gives small metric corrections a separate, normalized learning scale. Using flow matching, the action expert jointly predicts both components from the current image, ego state, and navigation command, without generating future images at inference. The same residual coordinates support bounded reinforcement post-training that refines execution while keeping the predicted visual scaffold fixed. With only one action denoising step, VITAL achieves 90.8 PDMS on NAVSIM v1 navtest, within 0.13 points of its ten-step score at 3.29x the end-to-end inference speed. Without retraining, it achieves 91.1 EPDMS on NAVSIM v2 navtest and 38.6 on navhard. Project page: https://worldbench.github.io/vital
Read the full abstract ▾
All authors Ao Liang, Youquan Liu, Mingxiao Li, Wei Chow, Lingdong Kong, Yue Liao, Dongyue Lu, Yu Yang, Wei Yin, Runnan Chen
Computer Vision
CommerceGen-Studio: An End-to-End Framework for Commercial-Grade Main-Image Generation
Xiaoli Xu, Zhengtao Wu, Zhengze Xu et al. (+8 more)
We present CommerceGen-Studio, an end-to-end framework for generating commercial-grade main images for e-commerce Stock Keeping Units (SKUs). Our framework is organized along a three-stage pipeline of information extraction, semantic rewriting, and multi-condition generation:
Prompt-Eval-Refine-Loop
Agent (PERL-
Agent) distills heterogeneous SKU metadata into concise selling-point text lists through an iterative closed loop of VLM and LLM roles scored by a reference-free rubric. Layout Thinker Optimization (LTO) aligns the rewriter with both caption and rendered-image quality via alternating domain-specific policy optimization. SKU-DiT, trained under Reinforce-Distill-Reinforce (RDR) progressive training, conditions jointly on the white-background product image, the trademark crop, and the structured layout to hold subject fidelity, trademark integrity, and character-level text accuracy together. The framework runs in production at a scale of millions of main images per day with a latency of a few seconds per image, which we believe makes it among the first systems of this kind to operate at commercial scale. We also release CommerceGen-Bench, a
benchmark of hard cases collected from live traffic across all product categories. On it, our framework achieves the highest overall scores among the evaluated open-source baselines on both editing tracks, with and without
prompt enhancement. It also matches much larger closed-source models, and beats several of them at Instruction Following and Product Integration - the two properties that determine whether an image can be published.
Read the full abstract ▾
All authors Xiaoli Xu, Zhengtao Wu, Zhengze Xu, Tao Liu, Yongchao Du, Zuan Gao, Xingjian Wang, Hao Yan, Mengting Chen, Wei Li, Bo Zheng
High Energy Physics (Theory)
Heavy-Heavy-Light-Light correlators from holography on the worldsheet
Matías Colaneri, Diego H. Correa, Víctor I. Giraldo-Rivera et al. (+1 more)
We study correlation functions of light local operators and defect changing operators on the conformal line set by the 1/2 BPS Wilson line of $\mathcal N=4$ super Yang-Mills. We consider the defect changing operators associated with geometrical and internal cusp angles, which at strong coupling have a large scaling dimension and back-react on the dual $AdS_2$ world-sheet. We set up a holographic computation of correlation functions, in which the light operators are described by fluctuations around the cusped classical string solution, so that correlators of two defect changing operators and an arbitrary number of light protected operators are obtained from the on-shell action of these fluctuations. We implement this prescription in detail for the case of certain 4-point functions, with two light and two defect changing operators. Finally, since for small cusp angles the defect changing operators become light, we manage to derive relations satisfied by integrated correlators of light operators, valid at any coupling. We verify these relations at strong coupling for both internal and geometrical cusps.
Read the full abstract ▾
All authors Matías Colaneri, Diego H. Correa, Víctor I. Giraldo-Rivera, Nicolás A. Ivanovich
High Energy Physics (Phenomenology)
Transient PULSE: Code for the Transient Propagation of UltraLight (pseudo)Scalars to Earth
Jason Arakawa, Charles Cheung, Muhammad Hani Zaheer et al. (+2 more)
We present Transient PULSE, a software package that numerically implements a previously developed source-to-detector framework for transient ultralight bosonic signals from astrophysical bursts. Predicting these signals relies on connecting the source emission spectrum to its
evolution during propagation and the resulting observables at a terrestrial detector. The package discretizes a user supplied energy spectrum into momentum modes, propagates them through a specified matter density profile and constructs the time-frequency energy density distribution at Earth, accounting for dispersion and in-medium screening effects. This treatment accommodates non-Gaussian source spectra and provides inputs for detector specific analyses. A complementary analytical module estimates coupling sensitivities, arrival time delays and screening thresholds from parameterized source and detector properties.
Read the full abstract ▾
All authors Jason Arakawa, Charles Cheung, Muhammad Hani Zaheer, Volodymyr Takhistov, Marianna Safronova
Number Theory
The existence of small defining polynomials for number fields
Manjul Bhargava and Sameera Vemulapalli
We prove that every number field of degree $n$ and absolute discriminant $D$ has a defining integral
polynomial of degree $n$ whose coefficients are all $O_n(\sqrt D)$ in size, thus resolving a 1998
conjecture of Ruppert. Our proof compares ideal
lattices with their trace duals, via which we handle all number fields with the exception of CM fields whose maximal totally real subfields have small discriminant. We then treat the remaining fields by averaging completed Epstein zeta functions over all possible ideal classes and archimedean scalings, for which we make critical use of Hecke's formula and the subconvexity
theorem of Michel and~Venkatesh. Our proof does not use the Quasi-Riemann Hypothesis.
Read the full abstract ▾
Instrumentation and Methods for Astrophysics
International Astrophysical Consortium for High-energy Calibration: Summary of the 18th IACHEC Workshop
K. K. Madsen, C. E. Grant, V. Burwitz et al. (+14 more)
In this report we summarize the activities of the International Astronomical Consortium for High Energy Calibration (IACHEC) from the 18th IACHEC Workshop at Pelham, Germany. Fifty-six scientists directly involved in the calibration of operational and future high-energy missions gathered during four days to discuss the status of the cross-calibration between the current international complement of X-ray observatories, and the possibilities to improve it. This summary consists of reports from the Working Groups with topics ranging across: the identification and characterization of standard calibration sources, multi-observatory cross-calibration campaigns, appropriate and new statistical techniques, calibration of instruments and characterization of background, preservation of knowledge, and results for the benefit of the astronomical community.
Read the full abstract ▾
All authors K. K. Madsen, C. E. Grant, V. Burwitz, H. Earnshaw, K. Forster, M. Guainazzi, V. L. Kashyap, H. L. Marshall, C. B. Markwardt, F. Meeker-Fürst, E. D. Miller, L. Natalucci, P. P. Plucinsky, M. Shidatsu, H. Suzuki, Y. Terada, A. Tümer
Computer Vision
DynThinker: Unifying Static 3D and Dynamic 4D Spatial Reasoning
Qihang Cao, Haoyuan Li, Zihan Guo et al. (+8 more)
Achieving general spatial intelligence requires a unified understanding of both static 3D and dynamic 4D. However, we find that existing models struggle to excel at both: 3D post-training can weaken dynamic understanding, while 4D-specialized models often lack strong static 3D
reasoning. Our analysis reveals that this spatial–dynamic interference is highly non-uniform across parameter groups, with the balance between static and dynamic capabilities particularly sensitive to changes in late-layer MLPs. To address this interference, we propose DynThinker, a unified framework for jointly strengthening static and dynamic spatial
reasoning. Decoupled Exponential Moving Average (D-EMA) applies configurable group-wise averaging to constrain drift in interference-sensitive pretrained parameters while preserving flexibility for capability enhancement. Spatial-Dynamic Fusion (SDF) complements D-EMA by selectively integrating spatial and temporal priors from a dynamic geometry encoder into intermediate VLM representations under semantic guidance. Across ten metrics spanning nine static 3D and dynamic 4D
benchmarks, DynThinker achieves the highest average score among all evaluated models. Notably, DynThinker achieves 56.0 on ReVSI, 67.2 on DSR-Bench, and 51.2 and 40.1 under the sample-wise and group-wise evaluation protocols of DSI-Bench, respectively, setting new state-of-the-art results among the compared open-source models across all four evaluations within a unified model.
Read the full abstract ▾
All authors Qihang Cao, Haoyuan Li, Zihan Guo, Sensen Gao, Zihui Gao, Huixin Zhang, Fan He, Jianhua Han, Hang Xu, Xiaodan Liang, Jia-Wang Bian
Artificial Intelligence
Co-Evolving Acting and Forecasting for Language Agents
Haofei Yu, Yixing Chen, Yining Zhao et al. (+6 more)
Training
language agents is bottlenecked by the high cost of long-horizon rollouts, yet standard agentic reinforcement learning (RL) learns only from a terminal scalar reward, leaving the supervision in intermediate steps largely unused. Prior work exploits this supervision through value targets that estimate per-step success probabilities or observation targets that fit environment response. Both are hard to construct well: (1) value targets come from the same single scalar signal for each step in the trajectory, and (2) observation targets, such as web pages and terminal logs, are long and noisy and reach only one step ahead. We propose \name, which adds a third kind of multi-step action targets. On successful on-policy trajectories, \name trains the policy backbone to generate the action sequences that follows each history. Each action is also supervised when forecast from the preceding histories, forming a temporal ensemble that the shared backbone absorbs during training, and this enables acting and forecasting to co-evolve. The policy still acts one step at a time at inference, with no overhead. Across five
benchmarks (ALFWorld, ScienceWorld, WebShop, $\tau^2$-bench, and AppWorld), \name (1) achieves stronger aggregate performance than GRPO and RL baselines with predictive auxiliary targets across simulated environments, while improving overall performance on both realistic applications; (2) delivers up to $1.9\times$ and $2.0\times$ higher sample efficiency than ECHO and GRPO, respectively, on ScienceWorld; and (3) achieves these gains with a single policy backbone, without additional models, separate pretraining, or inference overhead. The codebase is available at https://github.com/ulab-uiuc/CoPE.
Read the full abstract ▾
All authors Haofei Yu, Yixing Chen, Yining Zhao, Xuyan Ye, Yuxiang Liu, Zhigang Hua, Yan Xie, Shuang Yang, Jiaxuan You
Robotics
CoWear: Towards Collaborative Wearable Inertial Localization with Multiple Smart Devices
Zhuoyuan She, Zhanpeng Zhang, Yapeng Chen et al. (+3 more)
Consumer smart devices observe the same human motion from different body locations, yet existing inertial-localization
benchmarks rarely provide synchronized heterogeneous wearables with device-specific motion ground truth. We present CoWear, a
dataset and reference framework for collaborative inertial localization using a smartphone, smartwatch, and smart glasses. The
dataset contains over 30 participants, over 300 walking sessions, and more than 30 hours and 100 km of aggregate device trajectories, with each device individually tracked by motion capture. To enable collaboration across distinct body motions, we train role-specific
estimators to predict the displacement of a common reference device over synchronized temporal windows. Their predictions are transformed to the world frame using gyroscope integration and fused according to learned uncertainty. Experiments using phone, glasses, and watch as reference targets show that multi-device fusion consistently improves both displacement and trajectory rollout accuracy over individual-device estimation. CoWear provides a
dataset and reproducible baseline for studying collaborative inertial localization across heterogeneous consumer wearables.
Read the full abstract ▾
All authors Zhuoyuan She, Zhanpeng Zhang, Yapeng Chen, Jiale Wang, Xin Xie, Ming Xia
Mesoscale and Nanoscale Physics
Landau quantization of $C_3$ high-fold fermions in honeycomb lattices
Luisa Madail and Joaquín Fernández-Rossier
We present the theory of the Landau levels of $C_3$ high-fold
fermions in honeycomb
lattices. These
fermions arise as a generalization of the tight-binding model in a honeycomb
lattice with one orbital per site, known to describe
graphene, to a family of models with several orbitals per site transforming according to representations of $C_3$, relevant for triangulene two-dimensional
crystals. The models are defined by the number of orbitals $n_a$ and $n_b$ at the $a$ and $b$ sites of the unit
cell. They feature energy bands that combine flat bands and Dirac cones, that are isomorphic to those of
graphene, the Lieb, $p_{xy}$ honeycomb, and the Kagome tight-binding models. Combining tight-binding and $k\cdot p$ methods, here we compute the Landau levels of the $(1,1)$, $(1,2)$, $(2,2)$ and $(3,3)$ models, we compare them to those of the reference models, and we explore the similarities and differences with the Landau levels in
graphene. Our findings reveal that the nature of $C_3$ high-fold
fermions combine Dirac-type Landau quantization, originating from the linear crossings and associated relativistic band dispersion, with anomalous Landau quantization rooted in the
quantum geometry of the singular flat bands.
Read the full abstract ▾
General Relativity and Quantum Cosmology
Constraining the Progenitor Orbits of Sequential Black-Hole Mergers in Hierarchical Triples: Peters-Based Estimates
Felix M. Heinze and Zhoujian Cao
Two black-hole merger events may originate from sequential mergers in a pre-existing hierarchical black-hole triple if their spatial localizations overlap and the remnant mass of the first merger is compatible with a component mass of the second. In this scenario, the inner binary merges first, leaving a remnant that remains bound to the tertiary and subsequently undergoes a second merger. We present a method to constrain the progenitor triple's outer orbit by modeling the gravitational-wave-driven
evolution of the remnant-tertiary binary using Peters' equations. Given an assumed upper bound on the recoil velocity of the first merger, the method requires only the black-hole masses, the time delay between the mergers, and the eccentricity at a reference frequency - quantities that can be inferred from the merger events themselves. We investigate the potential constraining power by constructing a constraint atlas spanning a range of system properties and identify the conditions under which the method places tight and reliable bounds on the progenitor triple's outer orbital parameters. By imposing an upper bound on the detector-band eccentricity, we also obtain meaningful constraints for candidate hierarchical mergers proposed in the literature, including GW190514-GW190521. We provide numerical tools for applying the method to user-specified systems and evaluating the reliability of the resulting Peters-based constraints. This framework offers a computationally inexpensive first step toward reconstructing hierarchical-merger progenitors and provides a foundation for extensions incorporating more sophisticated relativistic treatments.
Read the full abstract ▾
Functional Analysis
The $\mathrm{U}$-Derivative Associated with Linear Recurrence Sequences
Ronald Orozco López
We associate a canonical operator to every linear recurrence sequence admitting a distinct-root Binet representation \(\mathrm{U}_n=\sum_{j=1}^{k}A_j\lambda_j^n.\) The construction is based on the spectral Binet operator \(\mathcal{B}_\mathrm{U}=\sum_{j=1}^{k}A_jT_{\lambda_j}, \) where \(T_{\lambda_j}f(x)=f(\lambda_jx)\) is the dilation operator associated with the characteristic root \(\lambda_j\). The corresponding recurrence derivative is defined by \(\mathrm{D}_\mathrm{U}=x^{-1}\mathcal{B}_\mathrm{U}.\) Its fundamental property is the monomial identity \(\mathrm{D}_\mathrm{U}(x^n)=\mathrm{U}_nx^{n-1},\) which establishes a direct correspondence between recurrence sequences and degree-lowering operators. We develop the operator theory of \(\mathrm{D}_\mathrm{U}\), including iteration, a spectral product rule, the associated Leibniz defect, and a quotient rule. The construction reveals a natural bifurcation determined by the value \(\mathrm{U}_0\), producing the Ward and non-Ward regimes as intrinsic operatorial phenomena. The product structure is shown to be governed by \(k-2\) residual spectral channels, leading to notions of spectral cancellation and spectral balance. As an application, we obtain a complete qualitative classification of second-order recurrence calculi. Using Horadam sequences as a universal second-order model, we prove that every such calculus is determined by the invariant triple \((\mathrm{W}_0,\Delta, Z_\mathrm{W}),\) where \(\mathrm{W}_0\) identifies the Ward/non-Ward type, \(\Delta\) determines the spectral geometry, and \(Z_\mathrm{W}\) determines regularity or singularity.
Read the full abstract ▾
Computer Vision
GeoThinker-$F$: Endogenous Predictive World Representations for Fast Spatial Intelligence
Haoyuan Li, Qihang Cao, Huanyuan Zhou et al. (+10 more)
Existing geometry-aware
Multimodal Large
Language Models (MLLMs) increasingly rely on external 3D encoders to provide structural priors, but executing these additional encoder paths at inference introduces substantial latency and computational overhead. We propose GeoThinker-$F$, an efficient framework that internalizes spatiotemporal perception by transforming causal MLLM hidden states into predictive latent world representations. Our key insight is that intermediate semantic representations contain sufficient information to predict useful representations of current geometry and future semantics, without requiring an exact reproduction of features from a dedicated 3D encoder. To realize this idea, GeoThinker-$F$ adopts a decoupled two-stage training strategy. In the first stage, the MLLM backbone is frozen while lightweight Latent Frame Prediction (LFP) heads learn from an offline 3D geometry expert and next-frame semantic targets. In the second stage, the external expert is removed, and the model is jointly optimized to generate and selectively integrate its self-predicted priors through a specialized Spatial-Grounded Temporal (SGT) Fusion module. Extensive evaluations show that GeoThinker-$F$ eliminates runtime dependence on external 3D geometry encoders while maintaining competitive performance across diverse spatial intelligence
benchmarks. It achieves up to a $3.6\times$ inference speedup, reducing latency from $1.45\,\mathrm{s}$ to $0.40\,\mathrm{s}$ while retaining strong spatial
reasoning performance.
Read the full abstract ▾
All authors Haoyuan Li, Qihang Cao, Huanyuan Zhou, Xin Hu, Sijia Chen, Zihan Guo, Kun Xiang, Yunshuang Nie, Tao Tang, Xinlong Zheng, Jia-Wang Bian, Xianming Liu, Xiaodan Liang
Quantum Physics
Logical CCZ state cultivation
Yuta Hirasaki and Jong Yeon Lee
Logical CCZ states enable Toffoli gates, a central building block of fault-tolerant
quantum arithmetic. We introduce an end-to-end protocol for direct CCZ-state preparation on bivariate tricycle and bicycle (BT/BB) codes. Our factory combines the cup-product CCZ gates native to BT codes with two novel ingredients, both enabled by engineering the logical coupling between BT and BB codes through projection-preserving lifts. Index-2 lifted BB ancillas enable selective $Z$-basis initialization, allowing the native cup-product gates to directly prepare two disjoint logical CCZ states. The prepared CCZ states are faithfully transferred to index-3 lifted BB
memories with high code distance. We use decoder-based postselection to obtain high-fidelity output states with low overhead. We perform exact simulations of our factory under a
circuit-level noise model motivated by neutral-atom hardware, obtaining joint infidelities of $1.84\times10^{-5}$ at $21.94\%$ acceptance and $4.10\times10^{-7}$ at $8.89\%$ acceptance for single-
qubit, CZ, and CCZ error rates of $0.01\%$, $0.1\%$, and $0.5\%$, respectively. At the output-infidelity target of our smallest factory, its estimated spacetime cost per CCZ state is lower than that of $T$-state cultivation followed by CCZ conversion, providing an efficient route to logical CCZ states for fault-tolerant
quantum computation.
Read the full abstract ▾
Computer Vision
Feed-Forward Multi-View Feature Coding for 3D Gaussian Splatting
Seungmi Choi, Ji-Hyoun Choi, Suhyun Jo et al. (+2 more)
Feed-forward 3D Gaussian Splatting (3DGS) methods enable efficient reconstruction of 3D scenes from sparse multi-view images in a single forward pass, without scene-specific optimization. However, existing feed-forward scene-coding pipelines first generate a complete Gaussian representation and then encode its geometry and appearance attributes separately, often involving non-differentiable operations. The resulting 'generate-then-compress' pipeline requires complex attribute-specific codec designs and limits rate-distortion performance by restricting the exploitation of cross-attribute redundancy. We propose multi-view feature coding for Gaussian splatting (MFGS), an end-to-end feed-forward framework for scene coding. Instead of compressing an explicit Gaussian representation, MFGS extracts multi-view features from the input images and encodes them into a compact bitstream. Through end-to-end rate-distortion training, MFGS learns compact multi-view features that jointly represent geometry and appearance, enabling the receiver to predict Gaussian attributes from the decoded features. We further introduce importance-guided Gaussian selection to accommodate different rendering budgets, allowing the receiver to adjust the number of Gaussians used for rendering from the same bitstream. Across RE10K, ACID, and DL3DV, MFGS achieves 99.76-99.86% bitrate savings over the generate-then-compress pipeline, as measured by BD-rate, while reducing total GMACs by 52-55% and providing an approximately 12 times end-to-end speedup.
Read the full abstract ▾
All authors Seungmi Choi, Ji-Hyoun Choi, Suhyun Jo, Tae-Gwon Park, Hui Yong Kim
High Energy Physics (Theory)
Post-Newtonian theory in de Sitter and anti-de Sitter spacetimes
R. Loganayagam and Avi Wadhwa
We develop the post-Newtonian framework for the near-zone dynamics of a binary in de Sitter and anti-de Sitter spacetime. We consider a binary interacting via a free scalar field in a fixed (A)dS background in $D+1$ spacetime dimensions, with $D\ge3$ and odd. Using multipole expansion and re-summation techniques, we solve for the field degrees of freedom to obtain an effective particle description in the form of a Fokker action. We derive the equations of motion for the binary up to second post-Newtonian (2PN) order and obtain the corresponding 2PN
Hamiltonian and conserved charges in phase space. We show that the conformal
algebra, corresponding to the isometry
algebra of the (A)dS background, is realised in phase space as a
curvature-dependent deformation of the Poincare
algebra at each PN order. Our results provide a systematic extension of post-Newtonian effective dynamics to finite-
curvature backgrounds. This could serve as a useful framework for studying bound-state dynamics in (A)dS as well as in the corresponding holographic CFT.
Read the full abstract ▾
Machine Learning
Your Transformer Is Secretly Elastic
Walter Simoncini, Dheeraj Varghese, Gertjan J. Burghouts et al. (+1 more)
Scaling up
transformers has produced models with broad capabilities, deployed across heterogeneous hardware and fluctuating workloads. Yet, model families offer only a few fixed sizes (e.g., S, B, L), forcing users to commit to a model based on hardware and latency constraints. This rigidity is often suboptimal under varying query loads. We introduce a retraining-free method that elastifies pre-trained
transformers into a single model that adapts on-the-fly across a continuum of computational budgets. Our key contribution is a differentiable relaxation of the functional unit-ranking problem: unlike prior work, which optimizes a single importance weight per MLP block or
attention head, we rank every individual MLP
neuron and
attention head jointly, on a common scale, by minimizing the model's loss under soft
pruning across many sparsity targets, all at once. Across 27
vision and text
transformer encoders ranging from 22M to 8B parameters, our method yields smooth degradation up to 60% sparsity and outperforms one-shot baselines under deep
pruning. For example,
pruning DINOv3 ViT-H+/16 to 50% sparsity removes 427M parameters while reducing ImageNet-1k linear-probing accuracy by only 0.8%, without the need for weight correction or retraining. At 60% sparsity, a pruned AugReg ViT-B/16 model achieves up to $1.80\times$ higher throughput and $1.26\times$ lower single-query latency on an A100 GPU. Elastic models are produced in minutes, with construction time scaling sub-linearly with parameter count.
Read the full abstract ▾
All authors Walter Simoncini, Dheeraj Varghese, Gertjan J. Burghouts, Cees G.M. Snoek
Game Theory
Price of Anarchy of First-Price Auctions in Agentic Markets
Shubhendu Trivedi, Jacob Eisenstein, Alekh Agarwal et al. (+3 more)
An agentic
market is a decentralized serving-time
market where AI
agents bid to claim each user
prompt. An
agent's bid is based on its expected value for the
prompt, which is the expectation of the reward for its response minus its inference costs. First-price auctions (FPAs), in which the highest bidder pays its bid and receives the
prompt, are a natural way to allocate
prompts in such a
market, since they are simple and transparent to run, and do not require central knowledge of private capabilities and costs. However, FPAs are not truthful: participants may bid less than their true value in the hope of a larger profit, and the
prompt may then be allocated to someone other than the bidder who values it the most. This welfare loss is quantified by the $\textit{price-of-anarchy}$ (PoA). In this paper, we argue that agentic
markets are likely to possess features that reduce the price-of-anarchy. The first is a $\textit{competitive floor}$, which occurs when multiple bidders usually value the
prompt above some baseline share of the optimal welfare (e.g., because the best models are often distilled into cheaper alternatives, which perform similarly). We show that competition among them then limits the welfare loss, which can be driven to zero as the floor rises. The second feature is $\textit{near-independence}$ of values after conditioning on the
prompt, which occurs when most of the predictable variation in the expected value is determined by the difficulty of the
prompt. As the value distribution approaches conditional independence, we can derive an efficiency guarantee that approaches a factor of $1 - 1/e^2$ of the optimal welfare. Thus, both features give rise to welfare lower bounds that are significantly better than the well-known general case of $1 - 1/e$. We validate these results empirically using auctions with strategic bidders simulated on the EmbedLLM
prompt routing
dataset.
Read the full abstract ▾
All authors Shubhendu Trivedi, Jacob Eisenstein, Alekh Agarwal, Ian Gemp, Michael Kaisers, Adam Fisch
Machine Learning
Afshin Khadangi
We introduce Continuum
Attention (CA), a causal
Transformer attention operator that replaces discrete key and value lookup with integration over a continuous, piecewise linear
memory curve. Each query induces an exponential relevance density along the curve. Integrating the density and interpolated values yields exact normalized
attention from closed form exponential moments, including half
token boundary contributions. The gradients follow analytically from moments of order at most two, enabling parallel GPU implementations while retaining quadratic sequence complexity and the standard
Transformer decoder. Compared with ordinary softmax and focal
attention, CA reduces mean validation negative log likelihood by 0.0200 and 0.0480 nats per byte, respectively, winning all paired comparisons. An additional disjoint validation slice yields a 0.0457 nats per byte advantage over the strongest selected softmax baseline, with a paired 95% confidence interval. These findings motivate continuous integration as a promising
attention alternative while leaving larger scale generalization open.
Read the full abstract ▾
Machine Learning
Learnable Hodge-Laplacian Driven Mixture of Experts for Steady-PDE Surrogates on Coarse Unstructured Meshes
Stavros Nousias, Akis Nousias and Konstantinos Gkrispanis
Current approaches for steady partial differential equations surrogate models on unstructured meshes apply a single mixing mechanism to the whole solution. This paper proposes an architecture that partitions the solution among vanilla
transformer experts, each focused on a different spectral scale. The partition is built from a discrete Hodge
Laplacian $L_0=\star_0^{-1} d_0^{\top}\star_1 d_0$, that serves as a fitting layer measuring high-frequency content. A gating mechanism takes the mesh as input and produces the per-vertex weights that recombine the expert outputs. AirFRANS is a steady Reynolds-averaged Navier–Stokes (RANS) airfoil
benchmark. We evaluate our approach in a reduced remeshed version of the
benchmark with each unstructured mesh counting 9.6K faces and 4.9K vertices. We compare our approach with the Transolver on the same
dataset and
dataset slit. Our approach reduces relative $L_2$ error by 31% and interior mean squared error by 47%.
Read the full abstract ▾
Computer Vision
Player-1: Learning to Play from Synthetic Experience
Mingyu Ouyang, Yiran Wang, Siyuan Hu et al. (+9 more)
Computer-use
agents have demonstrated strong capabilities in interpreting screens and executing keyboard and mouse actions. However, games, with their visually rich worlds and diverse mechanics, remain challenging and offer a demanding next step toward general digital control. Current game-
agent learning pipelines often rely on large-scale demonstrations with limited coverage of dynamic game states, while scalable environments for interactive training remain difficult to obtain. We introduce Player-1, a post-training framework for learning game control beyond demonstrations through synthetic experience. We extend the computer-use action space with relative pointer motion, sustained inputs, and composite actions. For supervised initialization, we fine-tune computer-use models on 6,044 curated gameplay trajectories that pair recorded actions with task instructions and visual
reasoning. We construct Player-1-Gym, sixteen synthetic 2D and 3D environments designed around control demands shared across games. Programmatically verifiable rewards and randomized observations and control dynamics support multi-environment reinforcement learning with PPO. The resulting Player-1-9B and Player-1-35B-A3B improve MineDojo-300 success over their SFT policies by 27.8 and 17.7 percentage points, respectively. Player-1-9B reaches 38.33% success compared with 12.63% for the strongest evaluated baseline, with further gains over SFT on ViZDoom-10 and GameWorld. Further analyses show that broader synthetic environment coverage improves transfer, while the use of composite actions that combine multiple keyboard and mouse inputs increases during RL. We demonstrate that learning from synthetic experience substantially improves game control beyond supervised initialization, with gains that generalize beyond the RL training suite.
Read the full abstract ▾
All authors Mingyu Ouyang, Yiran Wang, Siyuan Hu, Xingyilang Yin, Huiqiang Sun, Guangzhi Wang, Junfu Pu, Cheng Cheng, Honghao Chen, Ying Shan, Mike Zheng Shou, Hwee Tou Ng
Probability
Supercritical sharpness for Ising double random currents
Trishen S. Gunaratnam, Christoforos Panagiotis, Romain Panis et al. (+1 more)
We show that the percolation model induced by the double random current representation of the Ising model on $\mathbb Z^d$, with $d\geq 3$, satisfies local uniqueness of the macroscopic cluster with high probability for every $\beta>\beta_c$, uniformly in boundary conditions (i.e. sources). Such a result is sometimes called supercritical sharpness in percolation theory and yields a detailed description of the model via standard renormalisation techniques. As applications, we show the exponential decay of truncated correlations for the Ising model and exponential mixing for the FK-Ising measure, results previously obtained by Duminil-Copin, Goswami and Raoufi. However, our proof is considerably simpler, as it replaces the use of a highly technical multi-valued map principle with a simple sprinkling argument.
Read the full abstract ▾
All authors Trishen S. Gunaratnam, Christoforos Panagiotis, Romain Panis, Franco Severo
Computation and Language
Latent Compaction with Test-Time Training
Yingfa Chen, Xingyu Shen, Chaojun Xiao et al. (+1 more)
Long contexts enlarge key–value (KV) caches, increasing
memory use and
attention costs during generation. Compressing these caches must preserve information for unseen questions. We introduce **Latent Compaction**, which learns a compact document-specific cache through test-time training. We initialize cache entries using reconstruction importance, teacher
attention, and key-space diversity. Training first fits
attention outputs and log-normalizers to those of the full cache, with reconstruction-selected entries held fixed. It then jointly optimizes the entire cache through end-to-end
distillation from the full-cache model. Model weights remain frozen throughout, and training uses only generic
prompts about the document, without access to downstream questions or answers. Experiments on RULER, LongBench, and Phonebook Lookup demonstrate substantial gains over budget-matched compression baselines. On a RULER, for example, our method achieves 96.7% accuracy at 32$\times$ compression, compared with 40.0% for KVZip+. The method also transfers across model scales and families, with
retrieval gains on both Qwen3 and MiniCPM5. These results show that treating the KV cache as a neural network and optimizing it with test-time training can preserve
retrieval accuracy under aggressive compression. The code is released at https://www.github.com/THUNLP/latent-compaction
Read the full abstract ▾
All authors Yingfa Chen, Xingyu Shen, Chaojun Xiao, Xu Han