arXiv subjects

Yi Yang

Publications and source records attributed to Yi Yang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Temporal chirality in Floquet photonic media

We reveal temporal chirality in time-modulated photonic media: reversing the orientation of a closed loop traced by the complex permittivity can leave the Floquet spectrum unchanged while strongly modifying the directional scattering response under the same excitation. This behavior is linked to two symmetry constraints on the evolution operator, namely pseudo-Hermiticity and pseudo-unitarity. We prove the general occurrence of temporal chirality based on a complex Bogoliubov parametrization of the scattering amplitudes. We illustrate these findings in a single-harmonic temporal-$\mathcal{PT}$-symmetric system: both modulation-loop orientations support the same vacuum-like Floquet photonic bands at the phase transition, but one suppresses backward photon generation whereas the other induces strong growth, as confirmed by full-wave simulations. Our results establish temporal chirality as a distinct notion in time-varying photonics and identify modulation-loop orientation as a control knob for Floquet wave dynamics.

physics.optics↗

arXiv2026

Point It, Strike It: Direction-Conditioned Dynamic Manipulation of Deformable Linear Objects

Goal-conditioned dynamic manipulation of deformable linear objects has mainly specified goals as positions for a rope tip to reach. Many tasks, however, depend on how the tip arrives. We therefore study single-swing rope striking with goals that specify the tip's 3D position and arrival direction, across the workspace and on different ropes. This is challenging because rope dynamics are hard to model, no demonstrations exist, distinct swings reach the same goal with different reliability, and the sim-to-real gap extends beyond the rope. To address these challenges, we extend the state-of-the-art DLO simulator DeformX with GPU acceleration, a stable Cosserat rod solver, and a cross-flow aerodynamic model, yielding DeformX2.0, which is more than $20{,}000\times$ faster. We then propose TRACE (Trace-rooted Adaptive Cross-Entropy), which generates striking data by warm-starting each new target from the stored swing whose tip path passes closest to it. Its cost penalizes rope bending and abrupt tip motion to favor repeatable swings. A conditional flow-matching policy trained on this data reaches 92.1% accuracy in simulation. Finally, we propose RECAP (Residual Calibration Policy), which fits the simulator's rope and rig parameters to a few calibration swings and adapts actions with a correction policy trained in simulation. On a real robot, across three ropes, RECAP raises success within 5cm from 72% to 87% for position goals, and within 10cm and 10° from 50% to 79% for goals that also specify the arrival direction.

cs.RO↗

arXiv2026

On-Policy Distillation Teaches New Skills but Not New Knowledge

On-policy distillation (OPD) strengthens language-model reasoning, yet whether students acquire new factual knowledge or compositional skill for multi-step reasoning remains unknown. We separate these capabilities using a controlled synthetic framework that measures the student's initial capabilities and independently controls the teacher's additional facts, compositional skill, or both. Across four models from three families, reverse-KL OPD reliably transfers compositional skill across unseen reasoning structures, but transfers minimal factual knowledge. Decoupling the distillation recipe reveals the source of this asymmetry: replacing reverse KL with forward KL restores factual transfer, whereas student rollouts specifically improve the execution of multi-step reasoning. Experiments on recent factual QA and competition mathematics show a similar asymmetry under reverse-KL OPD, yielding notable reasoning gains without factual memory expansion. Together, these results demonstrate that on-policy distillation does not expand a model's parametric knowledge, but instead teaches it to organize and compose the knowledge it already possesses.

cs.CL↗

arXiv2026

Beyond the Transcriptome: Chromatin-Informed Prediction of Cell-State-Dependent Perturbation Responses

Predicting transcriptional responses to genetic perturbations is central to understanding gene function. Existing predictors primarily rely on transcriptomic measurements, although chromatin accessibility provides complementary information about the cellular context in which perturbations act. Using this information requires linking chromatin context to specific perturbations and accounting for baseline differences between independently sampled control and perturbed populations. We propose ChromaPert, a chromatin-informed framework for predicting state-dependent perturbation responses. ChromaPert combines molecular and DNA-informed locus priors with paired control RNA--ATAC features to jointly represent target identity and measured cellular context. Its Chromatin-Guided Response Router (CGRR) builds a source bank of control-derived expression offsets and retrieves relevant sources using perturbation/locus or control-ATAC similarity. These offsets account for baseline differences, while conditional transport flow learns the remaining response. Across three unseen-target and held-out-state settings on K562 CAT-ATAC and Perturb-Multiome, ChromaPert achieves the highest all-gene response correlation among evaluated methods. When transferring known perturbations to held-out states, it improves correlation by 23.2% across all genes and 46.2% for the twenty most responsive genes over the strongest respective baselines. Correct RNA--ATAC pairing improves recovery of chromatin-associated response slopes over shuffled pairing, while cross-state predictions retain state-specific responses to the same perturbation.

q-bio.GN↗

arXiv2026

GRC-Pose: Generation-Reconstruction Correspondence for Prior-Free 6D Object Pose Tracking

Prior-free 6D object pose tracking seeks to recover the trajectory of an unseen object from a single RGB video without object-specific CAD models, posed reference images, or pose annotations. Geometric foundation models provide complementary object-centric and scene-centric cues, yet SAM3D CAD is indexed by an arbitrary object-local surface parameterization, whereas reconstructed evidence is expressed in a sequence-specific world frame with partial surface coverage. To exploit this complementarity, we formulate tracking as generation-reconstruction correspondence and introduce GRC-Pose, a correspondence-based framework that combines learned correspondence prediction with robust pose estimation. Concretely, GeoCorr-Matcher estimates weighted object-scene correspondences and per-match uncertainty for each pose candidate. FGH-Solver integrates these matches through multiple robust geometric estimators and sequence-level posterior inference, while a posterior-gated memory retains only inlier-supported observations through occlusion and viewpoint change. Extensive evaluation shows that with SAM3D CAD, GRC-Pose achieves state-of-the-art Average Recall and motion retention on HOT3D, improving the latter by 58% over prior art. On classical benchmarks including YCBInEOAT and LINEMOD, it remains highly competitive.

cs.CV↗

arXiv2026

A Common Antimatter Response in AMS-02 Positrons and Antiprotons

I develop a common history-overlap description that gives antiparticle time orientation a physical role beyond its use in field-theory amplitudes. In this framework, opposite orientation reduces the effective overlap with a finite, matter-defined environmental history, while standard local interactions, positive physical energies, and causal detector records are retained. The common geometry connects two contrasting cosmic-ray spectral patterns measured by AMS-02. For positrons, a geometric exposure estimate displaces the broad maximum of an empirical energy-weighted electron reference from 8--10 GeV to the 300--400 GeV region. A number-conserving calculation combines energy redistribution and spatial dilution to interpret the relative weighted height, of order $0.1$. For antiprotons, ordinary secondary production retains the parent-proton history. I extend the same three-direction overlap estimate to the accumulated logarithmic deformation after production. Within this approximation, an ordinary softening of a few tenths leaves a residual additional spectral index of order 0.01, yielding near preservation of the source index and a nearly flat antiproton-to-proton ratio in the production-scaling limit. The positron comparison uses no additional dominant source, and known astrophysical contributions remain allowed. The paired AMS spectra give this time-orientation framework concrete empirical direction: a common geometric description relates a lepton scale displacement to hadronic index preservation and defines explicit targets for independent theoretical and experimental tests.

hep-ph↗

arXiv2026

TacForcing: Streaming Action Generation with Execution-Time Tactile Feedback

Contact-rich manipulation requires adapting to contact states that can evolve substantially within an action horizon. However, chunk-based vision-language-action models predict complete action chunks from observations collected before execution, leaving tactile conditioning stale during execution. Existing tactile-reactive approaches typically rely on separate high-frequency controllers, which increase both architectural and training complexity. In this paper, we introduce TacForcing, a streaming action-generation framework incorporating execution-time tactile feedback. TacForcing replaces the standard action expert with a streaming expert that generates action blocks sequentially while preserving intermediate states of unfinished blocks. After each block is executed, the expert resumes generation from these states using newly acquired tactile feedback. To better align tactile conditioning with action execution, we further introduce Execution-Aware Tactile Attention (EATA), which restricts direct access to each tactile update to the next block scheduled for execution. Across six UniVTAC simulation tasks and six real-world contact-rich manipulation tasks on two robot platforms, TacForcing achieves average success rates of 65% in simulation and 66% in the real world, outperforming the strongest baselines by 6 and 15 percentage points, respectively.

cs.RO↗

arXiv2026

Relic: From Multi-Agent Collaboration to Persistent Organizational Capability

Multiple agents may often conflict in an organization: for example, one coding agent changes an interface in a repository, but another continues to develop on the old version where existing tests become stale. A conversation can resolve the episode, but when the participants change, what makes the lesson continue to govern the team? We introduce Relic, which turns recurring collaboration failures into organization-owned, executable protocols. Members reflect on visible work, propose rules, and govern their adoption. Adopted protocols bind triggers, responsibilities, required evidence, and execution consequences to the runtime, while remaining open to revision and retirement. In one traced case, repeated integration friction produces an interface-review rule that governs later pull requests and is revised as work continues. Across 360 controlled runs over ten software workloads and three models, Relic raises complete-contract delivery from 14.06% to 19.76% (+5.71 percentage points) over a matched structured team without the protocol lifecycle, improving all four verified production endpoints in every model stratum. Under fresh-member transfer, behavioral correctness is 25.4% with no inherited protocol, 34.6% with the same rules provided as readable text, and 41.2% with executable bindings, a +6.5-point advantage over text alone. On the full CooperBench benchmark, after excluding 183 broken benchmark pairs, Relic achieves 371/469 (79.1%), establishing the best reported result among peer-structured systems. On the 47-pair same-model subset, Relic also exceeds Solo (28/47 vs. 26/47), reversing the coordination loss exhibited by the official peer baseline. Together, these results show how collaboration experience can become persistent organizational state that remains useful beyond the members who created it.

cs.AI↗

arXiv2026

What Shared Prefixes Hide: Trajectory Dropout for On-Policy Distillation

On-policy distillation (OPD) trains a student model on its own trajectories using dense token-level feedback from a stronger teacher model. Since each update is conditioned on the reasoning prefix already generated by the student, the prefix also shapes how effectively teacher feedback is converted into learning. We find that shared prefixes can lead to weak token-level updates, a phenomenon we call Prefix-Induced Supervision Attenuation (PISA). This attenuation arises in two common cases. (i) High student confidence can weaken corrective gradients even when the teacher disagrees. (ii) Tokens that rely on earlier reasoning can receive learning signals as weak as those for simple local continuations. To solve this problem, we propose Trajectory Dropout, a simple training-time intervention that exposes these weakened signals. The student first performs a standard full-context rollout to generate a complete trajectory. During training, we randomly drop a certain proportion of the student's reasoning trajectory, while the teacher continues to observe the complete trajectory for token-level supervision. This intervention strengthens corrections for overconfident predictions and introduces additional supervision at prefix-sensitive positions. Trajectory Dropout consistently improves average performance across teacher--student model pairs of different scales and six mathematical reasoning benchmarks, while also yielding gains on two out-of-domain benchmarks. It can also be flexibly integrated into existing OPD variants with negligible computational overhead, further improving their performance. These results demonstrate that Trajectory Dropout provides a simple mechanism for strengthening token-level supervision across model scales and OPD objectives.

cs.AI↗

arXiv2026

UniBYD: A Unified Framework for Learning Robotic Manipulation Across Embodiments Beyond Imitation of Human Demonstrations

In embodied intelligence, the embodiment gap between robotic and human hands brings significant challenges for learning from human demonstrations. Although some studies have attempted to bridge this gap using reinforcement learning, they remain confined to merely reproducing human manipulation, resulting in limited task performance. Moreover, current methods struggle to support diverse robotic hand configurations. In this paper, we propose UniBYD, a unified framework that uses a dynamic reinforcement learning algorithm to discover manipulation policies aligned with the robot's physical characteristics. To enable consistent modeling across diverse robotic hand morphologies, UniBYD incorporates a unified morphological representation (UMR). Building on UMR, we design a dynamic PPO with an annealed reward schedule, enabling reinforcement learning to transition from offline-informed imitation of human demonstrations to online-adaptive exploration of policies better adapted to diverse robotic morphologies, thereby going beyond mere imitation of human hands. To address the severe state drift caused by the incapacity of early-stage policies, we design a hybrid Markov-based shadow engine that provides fine-grained guidance to anchor the imitation within the expert's manifold. To evaluate UniBYD, we propose UniManip, the first benchmark for cross-embodiment manipulation spanning diverse robotic morphologies. Experiments demonstrate a 44.08% average improvement in success rate over the current state-of-the-art. Our project page is https://zhanheng-creator.github.io/UniBYD.

cs.RO↗

arXiv2026

Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference

The quadratic computational complexity of standard attention mechanisms presents a severe scalability bottleneck for LLMs in long-context scenarios. While hybrid attention mechanisms combining Full Attention (FA) and Sparse Attention (SA) offer a potential solution, existing methods typically rely on static allocation ratios that fail to accommodate the variable retrieval demands of different tasks. Furthermore, head-level dynamic sparsity often introduces severe computational load imbalance and synchronization long-tails, which hinder hardware acceleration during autoregressive decoding. To bridge this gap, we introduce Flux Attention, a context-aware framework that dynamically optimizes attention computation at the layer level. By integrating a lightweight Layer Router into frozen pretrained LLMs, the proposed method adaptively routes each layer to FA or SA based on the input context. This layer-wise routing preserves high-fidelity information retrieval while ensuring contiguous memory access, translating theoretical computational reductions into practical wall-clock speedups. As a parameter-efficient approach, our framework requires only 12 hours of training on 8$\times$A800 GPUs. Extensive experiments across multiple long-context and mathematical reasoning benchmarks demonstrate that Flux Attention achieves a superior trade-off between performance and inference speed compared with baseline models, with speed improvements of up to $2.8\times$ and $2.0\times$ in the prefill and decode stages.

cs.LG↗

arXiv2026

RAVEL: Asynchronous Rolling Inference for Flow-Based Vision-Language-Action Models

Flow-based vision-language-action (VLA) models are highly effective for generalist robot manipulation, yet their reliance on computationally expensive VLM encoding and multi-step iterative action generation imposes a significant latency bottleneck. The resulting inference latency makes it difficult for robots to respond quickly, especially in dynamic environments. We address this limitation with RAVEL (Rolling Asynchronous VLA Enabling Low-Latency Control), an asynchronous inference framework that addresses the computational bottlenecks of both the VLM backbone and the action expert. To reduce the delay from multi-step action denoising, RAVEL allows near-term actions to be executed after a single denoising step by carrying partially denoised future actions forward in a rolling buffer. To avoid blocking on slow VLM encoding, RAVEL decouples VLM encoding from rolling action generation, allowing the action expert to operate continuously using the latest available VLM context, while a lightweight Fast Observation Pathway (FOP) directly conditions the action expert on current observations. Across simulated and real-world manipulation tasks, RAVEL consistently achieves substantially lower response latency while maintaining the task capability of the underlying VLA, enabling high-frequency and responsive closed-loop control.

cs.RO↗

arXiv2026

PDE-JEPA: Predictive Representation Learning of Latent Dynamics Modeling for Parametric PDEs

Physical trajectories contain more than snapshots of a system: they also reveal how its states evolve under governing conditions. However, representation learning for parametric partial differential equations (PDEs) has largely relied on reconstruction-based objectives that emphasize recovering observed physical fields. In this paper, we investigate predictive representation pretraining as an alternative to reconstruction-based learning. We find that predictive representations preserve rich physical information, yet this advantage alone does not ensure accurate field evolution. Based on these observations, we introduce PDE-JEPA for parametric PDE dynamics. Specifically, we first train an encoder using a masked-latent prediction to capture the underlying regularities of PDE dynamics. To explicitly adapt the pretrained representation toward a more dynamics-aligned state space, we then introduce a geometry projector that aligns latent trajectory geometry with the evolution geometry of physical fields. Finally, building on this geometry-aligned latent space, we further develop a physics-structured latent predictor that decomposes the dynamics into parameter-independent evolution and parameter-dependent response components. Extensive experiments on nine widely used PDE benchmarks demonstrate that our framework outperforms existing state-of-the-art methods by an average of 33.4\% in-distribution, while achieving an average improvement of 51.4\% when extrapolating to unseen governing parameters. The project page is available \href{https://tanpig-x.github.io/PDE-JEPA/}{here}.

cs.AI↗

arXiv2026

Synthetic SU(3) non-Abelian gauge fields in photonic synthetic frequency dimensions

Synthetic gauge fields provide a powerful route to engineer photonic wave dynamics, especially in synthetic frequency dimensions formed by modulated resonator modes. While most previous realizations have focused on Abelian or lower-rank non-Abelian structures, a photonic implementation of SU(3) gauge fields in the synthetic frequency dimension has remained lacking. Here we propose a route to realize SU(3) gauge structure using three coupled ring resonators with modulation- and interference-engineered mode conversion. We formulate a tight-binding Hamiltonian with matrix-valued hopping generated by the Gell-Mann matrices and show how its evolution can be decomposed into experimentally implementable optical elements. The resulting platform supports SU(3)-structured band dispersions with color-resolved texture and wave-packet dynamics shaped by SU(3) gauge fields. Full-wave-inspired numerical modeling further confirms the expected spectral response and Zitterbewegung-like motion in SU(3) subspaces. Our results introduce SU(3) gauge fields into synthetic frequency photonics and establish a route toward higher-rank non-Abelian control of light in modulated resonator systems.

physics.optics↗

arXiv2026

TTRSD: Test-Time Reinforcement Learning with Self-Distillation for Vision-Language Models

Test-time reinforcement learning enables vision-language models (VLMs) to adapt using unlabeled inputs. However, repeated sampling under fixed visual conditions can reinforce shared perceptual errors, while sequence-level rewards fail to isolate visual perception the foundational bottleneck that anchors multimodal reasoning risking the degradation of pre-trained reasoning capabilities. We propose TTRSD, a test-time reinforcement learning framework combining multi-view answer-level self-distillation with visual contrastive token selection. A shared policy aggregates teacher predictions across original, cropped, and downsampled views into an answer distribution. Student trajectories generated from the original image receive rewards based on the support for their final answers in this distribution. To allocate this feedback precisely toward perceptual bottlenecks, we compare the log-probabilities of the same sampled tokens under original and visually ablated inputs while holding their textual prefixes fixed, selecting visually sensitive positions for policy-gradient updates. TTRSD separates update direction, determined by group-relative advantages, from update position, determined by visual sensitivity, without requiring ground-truth labels, external verifiers, or a separate teacher. With only 20 unlabeled adaptation samples, TTRSD improves performance across seven benchmarks and three VLMs, raising InternVL3-2B's MMMU accuracy from 35.79% to 49.32%(+13.53%), demonstrating cross-dataset generalization while preserving inherent reasoning integrity.

cs.CV↗

arXiv2026

The 91T/99aa-like Type Ia Supernova 2019vrq, Part I: Photometry, Spectroscopy, and Spectropolarimetry of a Nearly Standard Candle

Among the various major subtypes of Type Ia supernovae (SNe), the luminous 1991T-like (Filippenko et al. 1992; Phillips et al. 1992) and 1999aa-like (Krisciunas et al. 2000; Garavini et al. 2004) events stand out as particularly important for constraining the explosion mechanism. Their white dwarf (WD) progenitors are thought to have masses close to or exceeding the Chandrasekhar limit. We present spectrophotometric and polarimetric time-series observations of the overluminous Type Ia SN2019vrq. Between -9 and +12 days relative to the light-curve peak, its continuum polarization is consistent with zero, as seen in most normal-bright SNeIa and despite a possible rise up to 0.2% toward longer wavelengths as seen in subluminous SNeIa. Polarization across major spectral features is marginal ($\lesssim$0.3%) with the exception of the high-velocity CaII near-infrared triplet, which reaches $\sim$0.5%. This low polarization indicates a substantially spherically symmetric electron-scattering photosphere and minimal large-scale ejecta asymmetry. Early-time spectra reveal hot, doubly ionized ejecta dominated by SiIII and blended iron-group features. Intermediate-mass-element lines, such as CaII and SiII, are more clearly seen than in 91T-like events, though these lines remain weaker than in normal SNeIa. The intrinsically higher peak luminosity of overluminous 91T/99aa-like SNeIa (Yang et al. 2022; Phillips et al. 2024), their comparatively short formation timescale for a massive WD progenitor - a channel that becomes more common toward high redshift, and the moderate viewing-angle dependence of their spectrophotometric properties as inferred from their persistently low polarization, together suggest that this subclass may serve as a promising standard candle at high redshift.

astro-ph.SR↗

arXiv2026

The 91T/99aa-like Type Ia Supernova 2019vrq, Part II: 3D, Non-LTE, Low-Amplitude, Pulsating Delayed-Detonation Models of a Promising Standard Candle

We analyze the overluminous, 91T/99aa-like Type~Ia SN 2019vrq, based on light curves(LCs) and spectropolarimetric time-series. We employ 3D-radiation-hydrodynamical, full non-LTE simulations of low-amplitude, radially-pulsating-off-center-delayed-detonations(PDD) of a possibly rotating near-M(Ch) mass white dwarf (WD) to reproduce the LCs and spectra. The progenitor originates from a 7 Mo main-sequence star of solar metallicity. The explosion yields 0.86Mo of 56Ni and 0.023Mo of 58Ni. The latter falls a factor of 10 below that of `classical' delayed-detonations for 91T/99aa-like SNe, a diagnostic that is directly testable with JWST. The slow deflagration leaves a bound, pulsating WD. The detonation is triggered at 0.8 Mo. LCs and spectra require an outer 0.11 Mo of unburned material with twice-solar Fe, plausibly the ashes of an earlier, unsuccessful explosion, and low-level mixing of nuclear-statistical-equilibrium(NSE) elements. The spectra reflect early high ionization followed by recombination, with the photosphere shifting from intermediate-mass-element- to NSE-dominated layers about a week before maximum. The early high-velocity (HV) CaII IR3 line (likely produced by an aspherical density shell at 24,000km/s of 0.01 Mo) arises from an ionization sandwich rather than a double structure in abundances. After Ca recombines, the CaII IR3 wing reaches 33,000km/s well beyond the HV component. The low polarization is due to low scattering in an iron-group-dominated photosphere, consistent with asphericities <20% and resulting in a directional luminosity dependence 10-15% from the outer layers, and a dispersion of 35% in total. The LC-shape provides a further probe of asphericity, consistent with the locally tested polarimetry limits and relevant for high-z cosmology.

astro-ph.SR↗

arXiv2026

Tomography analysis of the intermediate-luminosity Type Iax SN 2024pxl

We present an abundance tomography analysis of SN 2024pxl, an intermediate luminosity Type Iax supernova ($M_{r}=-16.82 \pm 0.19$ mag), with the most-detailed follow-up in the Type Iax subclass to-date. As one of the few intermediate luminosity Type Iax objects, SN 2024pxl may link the two extremes of the peculiar thermonuclear supernova subclass. To test this hypothesis, we analyze its spectral evolution through the first 100 days after the explosion and aim to probe the structure of its ejecta. We conduct an abundance tomography analysis using synthetic spectra produced with the one dimensional radiative transfer code TARDIS. The fit of the spectral time series provides a radial scan of physical properties and probes the stratification of chemical elements throughout most of the SN ejecta. The observed spectral evolution is well fit with the final model, similar to the general predictions of the pure deflagration scenarios, but significant modifications are required in the density function of the inner ejecta and in the chemical profiles of the outermost regions. The constrained physical characteristics, such as the photospheric velocities and the time of maximum light, are also consistent with other SNe~Iax. Despite their small number, intermediate luminosity Type Iax SNe are not outliers in the subclass but demonstrate the continuous nature of SNe Iax through their luminosity range. Following this observation, we argue that all SNe~Iax share the same progenitor and explosion origin.

astro-ph.HE↗