arXiv:2601.05106v5 Announce Type: replace Abstract: Large language models (LLMs) exhibit strengths across diverse domains. However, achieving strong performance across these domains with a single general-purpose model typically requires scaling to sizes that are prohibitively expensive to train and deploy. On the other hand, while smaller domain-specialized models are much more efficient, they struggle to generalize beyond their training distributions. To address this dilemma, we propose FusionRoute, a robust and effective token-level multi-LLM collaboration framework in which a lightweight router simultaneously (i) selects the most suitable expert at each decoding step and (ii) contributes a complementary logit that refines or corrects the selected expert's next-token distribution via logit addition. Unlike existing token-level collaboration methods that rely solely on fixed expert outputs, we provide a theoretical analysis showing that pure expert-only routing is fundamentally limited: unless strong global coverage assumptions hold, it cannot in general realize the optimal decoding policy. By augmenting expert selection with a trainable complementary generator, FusionRoute expands the effective policy class and enables recovery of optimal value functions under mild conditions. Empirically, across both Llama-3 and Gemma-2 families and diverse benchmarks spanning mathematical reasoning, code generation, and instruction following, FusionRoute outperforms both sequence- and token-level collaboration, model merging, and direct fine-tuning, while remaining competitive with domain experts on their respective tasks.
Science Journals
arXiv:2601.08374v2 Announce Type: replace Abstract: MFEM is a widely used finite-element library, but its native linear-elasticity Partial Assembly (PA) path still applies an $O((p+1)^6)$ contraction in the element operator, leaving the CPU operator-throughput sweet spot near $p\approx 2$ in our baseline measurements. This work closes this implementation gap for MFEM linear elasticity on affine tensor-product hexahedral meshes by integrating four well-established tensor-product PA optimizations (sum factorization, Voigt notation, macro-kernel fusion, and slice-wise loop reorganization) into MFEM's native linear-elasticity PA path. The resulting operator is evaluated in high-order GMG-PCG solves using MFEM's geometric multigrid (GMG) components. On AMD EPYC 7713, the optimized operator achieves $7\text{--}83\times$ kernel speedup and $3.6\text{--}16.8\times$ end-to-end speedup across $p\in\{1,2,4,8\}$. At fixed problem size, the kernel-time operator throughput peaks around $p=6$ and remains high at $p=8$, shifting the operator-throughput sweet spot to $p\ge 6$. The same trend is reproduced on Huawei~Kunpeng~920 (ARMv8.2). These results are accompanied by per-stage ablation and hardware-counter characterization; the implementation will be released on GitHub.
arXiv:2606.14356v1 Announce Type: new Abstract: Applications in the 3D Computing Continuum, which unifies edge, cloud, and space, require combining multiple AI tasks such as object detection, time-series analytics, and natural language processing into Compound AI systems. These systems must satisfy stringent Service Level Objectives (SLOs) on accuracy, latency, and cost. A key mechanism for maintaining SLO compliance of Compound AI systems is runtime model selection, where AI models are dynamically switched for each workflow task. However, existing distributed and compound AI frameworks do not natively support runtime model selection. We present PLAIground, a framework that enables runtime model selection for Compound AI systems. PLAIground introduces Compoundable AI Model (CAIM) abstraction, which decouples task semantics from AI model implementations via Task and Data Contracts, enabling model switching without workflow changes. Additionally, PLAIground introduces Pixie, an SLO-driven runtime model selection algorithm, which dynamically selects the most suitable model for each task during execution. Our evaluation on two realistic Compound AI workflows demonstrates that Pixie achieves up to 91.3% accuracy while maintaining SLO compliance where fixed-model strategies either violate cost and latency budgets up to 21x or miss accuracy targets by 4%.
arXiv:2606.14358v1 Announce Type: new Abstract: Synchronous online teaching environments pose specific challenges for the analysis of pedagogical activity as teaching takes place via videoconferencing platforms and interactions are multimodal. While pedagogical orchestration has been extensively studied in the context of face-to-face courses and at the level of instructional design, the analysis of teaching in videoconferencing environments remains under-explored and insufficiently instrumented in terms of methodology. This article proposes a multimodal observation grid designed to analyze the instrumentalised pedagogical orchestration of teachers during synchronous online classes. Based on theories of pedagogical orchestration, multimodality and professional gestures of teachers, this grid identifies a set of observable indicators related to communicational gestures, posture, gaze and the management of digital tools. These indicators are structured and ranked in order of priority according to their observability and analytical relevance. They are operationalised to consider the constraints associated with data that can be analysed in videoconference class contexts. The proposed grid aims to provide a reproducible methodological framework for the analysis of instrumental pedagogical orchestration, with a view to future empirical validation.
arXiv:2606.14624v1 Announce Type: new Abstract: Numerous transition-metal oxides have low-temperature antiferromagnetic (AFM) states and high-temperature paramagnetic (PM) phases, where the AFM state is usually insulating while the PM phase can be either insulating or metallic. Without involving strong correlation, we use symmetry-broken density-functional theory (DFT) to obtain the PM phases of insulating NaFeO3 vs the recently discovered metallic NaOsO3. We develop the understanding of insulating and metallic behaviors in paramagnetic oxides by analyzing the interactions between magnetic and positional symmetry breaking: The insulating gap is governed by the competition between the spin disorder that induces a distribution of different magnitudes of local magnetic moments and the polymorphous distribution of off-center atomic displacements. NaFeO3, on the other hand, has large positional displacement with small spin-disorder-induced moments distribution, leading to insulating PM phase, whereas NaOsO3 has a pronounced spin-disorder-induced moments distribution that forces the PM phase to become metallic. Our work identifies this symmetry-breaking competition as a general framework to bridge seemingly disparate metal-insulator behaviors in transition-metal oxides paramagnets without invoking strong correlation.
arXiv:2601.09337v2 Announce Type: replace Abstract: The Unmapped Tent Pitching (UTP) algorithm is a space--time domain decomposition method for the parallel solution of hyperbolic problems. It was originally introduced for the homogeneous one-dimensional wave equation in [Ciaramella, Gander, Mazzieri, 2024]. UTP is inspired by the Mapped Tent Pitching (MTP) algorithm [Gopalakrishnan, Sch{\"o}berl, Wintersteiger, 2017], which constructs the solution by iteratively building polytopal space--time subdomains, referred to as tents. In MTP, each physical tent is mapped onto a space--time rectangle, where local problems are solved before being mapped back to the original domain. In contrast, UTP avoids the nonlinear and potentially singular mapping step by computing the solution directly on a physical space--time rectangle that contains the tent, at the expense of redundant computations in the region outside the tent. In this work, we investigate several strategies to extend UTP to heterogeneous media, where the wave propagation speed is piecewise constant over two subregions of the domain. Among the considered approaches, the most efficient in terms of computational time is the one employing space--time subdomains with identical spatial and temporal dimensions in both regions, determined by the maximum propagation speed.
arXiv:2601.11273v3 Announce Type: replace Abstract: In the RAG paradigm, document ranking determines the evidence available to downstream generators. Through controlled analysis, we identify two phenomena underexplored by existing rankers: (i) downstream response quality depends not only on relevance but also on the composition and ordering of selected documents, and (ii) such preferences differ systematically across generators. However, existing rankers are trained purely on query--document relevance, leaving both phenomena unmodeled. To close this gap, we construct \textbf{PRISM}, a bilingual preference-aligned dataset built through a four-stage pipeline that compresses the combinatorial subset-and-ordering space by roughly four orders of magnitude and produces response-quality preference supervision conditioned on seven downstream generators. On a 13k-query subset of PRISM, we train \textbf{Rank4Gen}, a generator-aware ranker that performs joint document set selection and ordering. Experiments on five challenging RAG benchmarks show that Rank4Gen improves downstream QA quality on most evaluated generators, with per-generator F1 gains of up to $+2.08$ over the strongest set-selection baseline. Code is available at https://github.com/JOHNNY-fans/Rank4Gen.
arXiv:2601.11626v2 Announce Type: replace Abstract: Large collections of matrices arise throughout modern machine learning, signal processing, and scientific computing, where they are commonly compressed by concatenation followed by truncated singular value decomposition (SVD). This strategy enables parameter sharing and efficient reconstruction and has been widely adopted across domains ranging from multi-view learning and signal processing to neural network compression. However, it leaves a fundamental question unanswered: which matrices can be safely concatenated and compressed together under explicit reconstruction error constraints? Existing approaches rely on heuristic or architecture-specific grouping and provide no principled guarantees on the resulting SVD approximation error. In the present work, we introduce a theory-driven framework for compression-aware clustering of matrices under SVD compression constraints. Our analysis establishes new spectral bounds for horizontally concatenated matrices, deriving global upper bounds on the optimal rank-$r$ SVD reconstruction error from lower bounds on singular value growth. The first bound follows from Weyl-type monotonicity under blockwise extensions, while the second leverages singular values of incremental residuals to yield tighter, per-block guarantees. We further develop an efficient approximate estimator based on incremental truncated SVD that tracks dominant singular values without forming the full concatenated matrix. Therefore, we propose three clustering algorithms that merge matrices only when their predicted joint SVD compression error remains below a user-specified threshold. The algorithms span a trade-off between speed, provable accuracy, and scalability, enabling compression-aware clustering with explicit error control.
arXiv:2601.12890v3 Announce Type: replace Abstract: Large Language Models (LLMs) have significantly advanced code analysis tasks, yet they struggle to detect malicious behaviors fragmented across files, whose intricate dependencies easily get lost in the vast amount of benign code. We therefore propose a graph-centric attention acquisition pipeline that enhances LLMs' ability to localize malicious behavior. The approach parses a project into a code graph, uses an LLM to encode nodes with semantic and structural signals, and trains a Graph Neural Network (GNN) under sparse supervision. The GNN performs an initial detection, and by interpreting these predictions, identifies key code sections that are most likely to contain malicious behavior. These influential regions are then used to guide the LLM's attention for in-depth analysis. This strategy significantly reduces interference from irrelevant context while maintaining low annotation costs. Extensive experiments show that the method consistently outperforms existing approaches on multiple public and custom datasets, highlighting its potential for practical deployment in software security scenarios.
arXiv:2601.12913v4 Announce Type: replace Abstract: This paper argues that interpretability research in Artificial Intelligence (AI) is fundamentally ill-posed as existing definitions of interpretability fail to describe how interpretability can be formally tested or designed for. We posit that actionable definitions of interpretability must be formulated in terms of *symmetries* that inform model design and lead to testable conditions. Under a probabilistic view, we hypothesise that four symmetries (inference equivariance, information invariance, concept-closure invariance, and structural invariance) suffice to (i) formalise interpretable models as a subclass of probabilistic models, (ii) yield a unified formulation of interpretable inference (e.g., alignment, interventions, and counterfactuals) as a form of Bayesian inversion, and (iii) provide a formal framework to verify compliance with safety standards and regulations.
arXiv:2601.12944v3 Announce Type: replace Abstract: We study the evolution of Tsallis entropy along the heat flow and establish concavity results in arbitrary dimensions. Extending earlier one-dimensional results, we prove that Tsallis entropy is concave along the heat flow for $q\in(0,3]$ in dimension one and for $q\in[1,3]$ in higher dimensions. The upper endpoint $q=3$ is sharp in every dimension. The proof is based on a nonlinear transformation of the heat equation, a sharp dimension-free functional inequality with constant $C_u=3$, and a rigorous justification of the integration-by-parts identities used in the argument. The sharp inequality is proved by an explicit integration-by-parts sum-of-squares identity, rather than by a computer-assisted semidefinite-programming search. As consequences, we recover a generalized de Bruijn identity, prove monotonicity of the associated $q$-Fisher information along the heat flow, and establish concavity results for Tsallis entropy power, including the Shannon entropy-power case and Costa's EPI as an endpoint. We also obtain an asymptotic entropy-power concavity statement for general initial data and a sharp auxiliary functional inequality which may be of independent analytic interest.
arXiv:2601.18707v2 Announce Type: replace Abstract: Machine learning-based surrogate models have emerged as more efficient alternatives to numerical solvers for physical simulations over complex geometries, such as car bodies. Many existing models incorporate the simulation mesh as an additional input, thereby reducing prediction errors. However, generating a simulation mesh for new geometries is computationally costly. In contrast, mesh-free methods, which do not rely on the simulation mesh, typically incur higher errors. Motivated by these considerations, we introduce SMART, a neural surrogate model that predicts physical quantities at arbitrary query locations using only a point-cloud representation of the geometry, without requiring access to the simulation mesh. The geometry and simulation parameters are encoded into a shared latent space that captures both structural and parametric characteristics of the physical field. A physics decoder then attends to the encoder's intermediate latent representations to map spatial queries to physical quantities. Through this cross-layer interaction, the model jointly updates latent geometric features and the evolving physical field. Extensive experiments show that SMART is competitive with and often outperforms existing methods that rely on the simulation mesh as input, demonstrating its capabilities for industry-level simulations.
arXiv:2601.19115v2 Announce Type: replace Abstract: With large-scale text-to-image (T2I) diffusion models achieving significant advancements in open-domain image creation, increasing attention has been focused on their natural extension to the realm of text-driven image-to-image (I2I) translation, where a source image acts as visual guidance to the generated image in addition to the textual guidance provided by the text prompt. We propose FBSDiff, a novel framework adapting off-the-shelf T2I diffusion model into the I2I paradigm from a fresh frequency-domain perspective. Through dynamic frequency band substitution of diffusion features, FBSDiff realizes versatile and highly controllable text-driven I2I in a plug-and-play manner (without need for model training, fine-tuning, or online optimization), allowing appearance-guided, layout-guided, and contour-guided I2I translation by progressively substituting low-frequency band, mid-frequency band, and high-frequency band of latent diffusion features, respectively. In addition, FBSDiff flexibly enables continuous control over I2I correlation intensity simply by tuning the bandwidth of the substituted frequency band. To further promote image translation efficiency, flexibility, and functionality, we propose FBSDiff++ which improves upon FBSDiff mainly in three aspects: (1) accelerate inference speed by a large margin (8.9$\times$ speedup in inference) with refined model architecture; (2) improve the Frequency Band Substitution module to allow for input source images of arbitrary resolution and aspect ratio; (3) extend model functionality to enable localized image manipulation and style-specific content creation with only subtle adjustments to the core method. Extensive qualitative and quantitative experiments verify superiority of FBSDiff++ in I2I translation visual quality, efficiency, versatility, and controllability compared to related advanced approaches.
arXiv:2606.13903v1 Announce Type: cross Abstract: We prove a degree-four Sum-of-Squares lower bound for the standard vector-coordinate formulations of mutually unbiased bases. For every dimension $d$ and every proposed number $m$ of bases, we construct a degree-four pseudoexpectation satisfying the orthonormality constraints and the cross-unbiasedness constraints in the quartic equality formulation. The construction is expectation over $m$ independent Haar-random orthonormal bases. We also prove that the same pseudoexpectation satisfies the degree-four localizing constraints for the natural $2\times 2$ Hermitian semidefinite formulation of the cross-coherence inequalities. Consequently, degree-four vector-coordinate SoS cannot refute the existence of $m$ mutually unbiased bases, even when $m>d+1$. In particular, under the two vector-coordinate encodings explicitly described in Randomstrasse101 Open Problem 23, degree-four SoS cannot prove that seven mutually unbiased bases do not exist in $\mathbb C^6$. We contrast this with a centered projector-coordinate Gram formulation, where degree-four SoS already recovers the elementary upper bound $m\le d+1$, giving a simple separation between vector-coordinate and projector-coordinate degree-four relaxations.
arXiv:2601.19420v3 Announce Type: replace Abstract: Spontaneous emission, as a fundamental radiative process and a versatile information carrier, plays a vital role in light-emitting devices, optical information modulation and encryption, super-resolution fluorescence imaging. Engineering the photonic environment surrounding photon emitters enables control over their emission properties. However, simultaneously achieving precise engineering of both excitation enhancement and quantum-yield modulation at the nanoscale remains elusive, highlighting substantial room for advancing the precise orchestrating of photoluminescence. Here, we introduce silicon Mie voids - air-defined cavities that invert the conventional solid-particle geometry - to achieve independent tuning of photoluminescence within a single subwavelength unit, while minimizing optical losses. Full-wave simulations and experiments on both gradient and uniform Mie-void arrays jointly validate this quantitative framework for spontaneous emission tuning, which disentangles excitation enhancement arising from local field confinement in air and quantum-yield enhancement resulting from strengthened emitter-resonator coupling, while confirming the accelerated radiative decay enabled by the modified optical LDOS. Leveraging this flexible mechanism, we realize a multimodal nanophotonic pattern with near-diffraction-limited pixels that encode the EPFL logo in the bright field and the SJTU logo in both dark field and photoluminescence maps. These results establish Mie voids as a powerful platform for high-density multimodal encrypted displays and open new avenues for advancing state-of-the-art nanophotonic devices.
arXiv:2601.19810v2 Announce Type: replace Abstract: Unsupervised pre-training can equip reinforcement learning agents with prior knowledge and accelerate learning in downstream tasks. A promising direction, grounded in human development, investigates agents that learn by setting and pursuing their own goals. The core challenge lies in how to effectively generate, select, and learn from such goals. Our focus is on broad distributions of downstream tasks where solving every task zero-shot is infeasible. Such settings naturally arise when the target tasks lie outside of the pre-training distribution or when their identities are unknown to the agent. In this work, we (i) optimize for efficient multi-episode exploration and adaptation within a meta-learning framework, and (ii) guide the training curriculum with evolving estimates of the agent's post-adaptation performance. We present ULEE, an unsupervised meta-learning method that combines an in-context learner with an adversarial goal-generation strategy that maintains training at the frontier of the agent's capabilities. On XLand-MiniGrid benchmarks, ULEE pre-training yields improved exploration and adaptation abilities that generalize to novel objectives, environment dynamics, and map structures. The resulting policy attains improved zero-shot and few-shot performance, and provides a strong initialization for longer fine-tuning processes. It outperforms learning from scratch, DIAYN pre-training, and alternative curricula. Code is available at: https://github.com/Octavio-Pappalardo/ulee-jax
arXiv:2601.21179v2 Announce Type: replace Abstract: This work studies the challenging problem of acquiring high-quality underwater images via 4-D light field (LF) imaging. To this end, we propose GeoDiff-LF, a novel diffusion-based framework built upon SD-Turbo to enhance underwater 4-D LF imaging by leveraging its spatial-angular structure. GeoDiff-LF consists of three key adaptations: (1) a modified U-Net architecture with convolutional and attention adapters to model geometric cues, (2) a geometry-guided loss function using tensor decomposition and progressive weighting to regularize global structure, and (3) an optimized sampling strategy with noise prediction to improve efficiency. By integrating diffusion priors and LF geometry, GeoDiff-LF effectively mitigates color distortion in underwater scenes. Extensive experiments demonstrate that our framework outperforms existing methods across both visual fidelity and quantitative performance, advancing the state-of-the-art in enhancing underwater imaging. The code will be publicly available at https://github.com/linlos1234/GeoDiff-LF.
arXiv:2601.21632v5 Announce Type: replace Abstract: Open-source AI is scaling rapidly, and model hubs now host millions of artifacts. Each foundation model can spawn large numbers of fine-tunes, adapters, quantizations, merges, and forks. We take the position that compute efficiency alone is insufficient for sustainability in open-source AI. Lower per-run costs can accelerate experimentation and deployment, increasing aggregate footprint unless impacts are measurable and comparable across derivative lineages. However, the energy use, water consumption, and emissions of these derivative lineages are rarely measured or disclosed in a consistent, comparable way, leaving aggregate ecosystem impact largely invisible. We argue that sustainable open-source AI requires a coordination infrastructure that tracks impacts across model lineages, not only base models. We propose Data and Impact Accounting (DIA), a lightweight, non-restrictive transparency layer that (i) standardizes carbon-and-water reporting metadata, (ii) integrates low-friction measurement into common training and inference pipelines, and (iii) aggregates reports via public dashboards to summarize cumulative impacts across releases and derivatives. DIA makes derivative costs visible and supports ecosystem-level accountability while preserving openness. Project Page: https://vectorinstitute.github.io/ai-impact-accounting/
arXiv:2601.22108v2 Announce Type: replace Abstract: Continued pretraining is optimized with fixed self-supervised tasks but selected by downstream performance, creating a coarse feedback loop in which practitioners evaluate checkpoints, change data mixtures or objectives, and restart runs, while individual updates remain blind to target capabilities. We ask whether a small set of verifiable downstream examples can provide step-level feedback without directly supervising the learner. We introduce V-pretraining, which decouples a learner trained only with a self-supervised loss from a lightweight task designer that constructs targets or views for unlabeled batches. Given the current learner and batch, V-pretraining scores a candidate construction by predicting the first-order reduction in downstream loss after the induced self-supervised update. The designer maximizes this value; the learner then applies the update with targets or views detached, so downstream labels never update learner parameters. We instantiate V-pretraining as adaptive top-K soft targets for language modeling and learned views or masks for self-supervised vision. Across both modalities, V-pretraining improves target capabilities without degrading generalization. Under wall-clock-matched continued pretraining, it improves GSM8K Pass@1 for Qwen models using 1,024 GSM8K examples only as feedback, including a +7.4 point single-run gain for Qwen2.5-0.5B. In vision, it improves DINOv3 transfer to ADE20K semantic segmentation and NYUv2 depth estimation while preserving ImageNet linear accuracy, suggesting that feedback-guided task construction can improve target capabilities without collapsing general-purpose representations.
arXiv:2601.22436v3 Announce Type: replace Abstract: Self-evolving large language model (LLM) agents continually improve by accumulating and reusing past experience, yet it remains unclear whether they faithfully rely on that experience to guide their behavior. We present the first systematic investigation of experience faithfulness, the causal dependence of an agent's decisions on the experience it is given, in self-evolving LLM agents. Using controlled causal interventions on both raw and condensed forms of experience, we comprehensively evaluate four representative frameworks across 13 LLM backbones and 9 environments. Our analysis uncovers a striking asymmetry: while agents consistently depend on raw experience, they often disregard or misinterpret condensed experience, even when it is the only experience provided. This gap persists across single- and multi-agent configurations and across backbone scales. We trace its underlying causes to three factors: the semantic limitations of condensed content, internal processing biases that suppress experience, and task regimes where pretrained priors already suffice. These findings challenge prevailing assumptions about self-evolving methods and underscore the need for more faithful and reliable approaches to experience integration.
arXiv:2601.22954v2 Announce Type: replace Abstract: Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to purely autoregressive language models because they can decode multiple tokens in parallel. However, state-of-the-art block-wise dLLMs rely on a "remasking" mechanism that decodes only the most confident tokens and discards the rest, effectively wasting computation. We demonstrate that recycling computation from the discarded tokens is beneficial, as these tokens retain contextual information useful for subsequent decoding iterations. In light of this, we propose Residual Context Diffusion (RCD), a module that converts these discarded token representations into contextual residuals and injects them back for the next denoising step. RCD uses a decoupled two-stage training pipeline to bypass the memory bottlenecks associated with backpropagation. We validate our method on both long CoT reasoning (SDAR) and short CoT instruction following (LLaDA) models. We demonstrate that a standard dLLM can be efficiently converted to the RCD paradigm with merely ~300 million tokens. RCD consistently improves frontier dLLMs by 4-11 percentage points in accuracy with minimal extra computation overhead across a wide range of benchmarks. Notably, on the most challenging AIME tasks, RCD nearly doubles baseline accuracy and attains up to 4-5x fewer denoising steps at baseline's peak accuracy.
arXiv:2602.00593v4 Announce Type: replace Abstract: Despite progress on general tasks, vision-language models (VLMs) still struggle with challenges that demand both fine-grained visual grounding and external knowledge, a synergy overlooked by existing benchmarks that evaluate these abilities in isolation. To fill this void, we introduce Pix2Fact, a visual question-answering benchmark designed to assess expert-level visual perception and knowledge search. Pix2Fact comprises 1,000 high-resolution (4K+) images spanning eight scenarios. Its questions and answers are meticulously crafted by PhD-holding annotators from top global universities across diverse disciplines. Each question requires detailed visual grounding and the integration of external knowledge. Evaluating ten state-of-the-art VLMs, including proprietary models such as Gemini-3.1-Pro and GPT-5.4, we find that Pix2Fact poses a formidable challenge: the most advanced model (Gemini-3.1-Pro) achieves only 51.7% average accuracy, even with access to visual ground truth and search tools. Our analysis attributes this low accuracy to three factors, frequent visual grounding errors even with visual ground truth, shallow search harnessing, and VLM's inability to retrieve long-tail, unstructured local information. This striking gap exposes the limitations of current models in assisting humans with real-world scenarios that demand overwhelming visual comprehension. We believe Pix2Fact will serve as a critical benchmark to drive the next generation of language-vision agents that seamlessly integrate fine-grained perception with robust knowledge search.
arXiv:2602.00845v3 Announce Type: replace Abstract: Agentic reasoning enables large reasoning models (LRMs) to dynamically acquire external knowledge, but yet optimizing the retrieval process remains challenging due to the lack of dense, principled reward signals. In this paper, we introduce InfoReasoner, a unified framework that incentivizes effective information seeking via a synthetic semantic information gain reward. Theoretically, we redefine information gain as uncertainty reduction over the model's belief states, establishing guarantees, including non-negativity, telescoping additivity, and channel monotonicity. Practically, to enable scalable optimization without manual retrieval annotations, we propose an output-aware intrinsic estimator that computes information gain directly from the model's output distributions using semantic clustering via bidirectional textual entailment. This intrinsic reward guides the policy to maximize epistemic progress, enabling efficient training via Group Relative Policy Optimization (GRPO). Experiments across seven question-answering benchmarks demonstrate that InfoReasoner consistently outperforms strong retrieval-augmented baselines, achieving up to 5.4% average accuracy improvement. Our work provides a theoretically grounded and scalable path toward agentic reasoning with retrieval. The code is available at https://github.com/dl-m9/InfoReasoner
arXiv:2602.01948v2 Announce Type: replace Abstract: Macro-micro manipulators combine a macro manipulator with a large workspace, such as an industrial robot, with a lightweight, high-bandwidth micro manipulator. This enables highly dynamic interaction control while preserving the wide workspace of the robot. Traditionally, position control is assigned to the macro manipulator, while the micro manipulator handles the interaction with the environment, limiting the achievable interaction control bandwidth. To solve this, we propose a novel control architecture that incorporates the macro manipulator into the active interaction control. This leads to a increase in control bandwidth by a factor of 2.1 compared to the state of the art architecture, based on the leader-follower approach and factor 12.5 compared to traditional robot-based force control. Further we propose surrogate models for a more efficient controller design and easy adaptation to hardware changes. We validate our approach by comparing it against the other control schemes in different experiments, like collision with an object, following a force trajectory and industrial assembly tasks.
arXiv:2602.03177v2 Announce Type: replace Abstract: Ground reaction forces (GRFs) provide fundamental insight into human gait mechanics and are widely used to assess joint loading, limb symmetry, balance control, and motor function. Despite their clinical relevance, the use of GRF remains underutilised in clinical workflows due to the practical limitations of force plate systems. In this work, we present a force-plate-free approach for estimating GRFs using only marker-based motion capture data. This kinematics only method to estimate and decompose GRF makes it well suited for widespread clinical depolyment. By using kinematics from sixteen body segments, we estimate the centre of mass (CoM) and compute GRFs, which are subsequently decomposed into individual components through a minimization-based approach. Through this framework, we can identify gait stance phases and provide access to clinically meaningful kinetic measures without a dedicated force plate system. Experimental results demonstrate the viability of CoM and GRF estimation based solely on kinematic data, supporting force-plate-free gait analysis.