Papers

Time series forecasting requires extrapolating the dynamics of an observed process beyond the last available measurement. Yet recurrent forecasting models typically treat observations as inputs that directly control their latent dynamics. It leads to a regime change when these observations become unavailable at prediction time. Following a state-estimation perspective, we introduce Observer State-Space Models (OSSMs), a class of recurrent models that interprets the observed input time series as measurements of an underlying autonomous dynamical system. OSSMs explicitly separate latent-state propagation from measurement assimilation: a single transition governs the dynamics across both context and forecasting intervals, while available observations correct the estimated state through an observer. This formulation naturally exposes classical control-theoretic properties, including observability and convergence of the state estimation error. We further show that conventional and recent SSMs can be recovered as particular instances of our OSSM framework, thereby providing a unified interpretation of their recurrent dynamics and revealing modeling inconsistencies. We perform experiments across several benchmarks showing that OSSM achieves substantial improvements while maintaining the same parameter count and training setup as the corresponding SSM baseline. These results support a simple principle for recurrent forecasting: observations should correct the estimated latent state, rather than control the dynamics used to propagate it.

A central limitation of current data-free self-evolution methods for training search agents is their reliance on difficulty-based proposer rewards. These methods reward a proposer for generating questions that challenge a co-evolving solver, using solver difficulty as a proxy for question quality. Yet difficulty alone is insufficient to distinguish questions that require cross-passage evidence from those that are answerable via simpler shortcuts. In addition, measuring difficulty demands repeated solver rollouts for every candidate question, leading to substantial computational costs. In this paper, we introduce \methodname, the first self-evolving search framework that eliminates difficulty-based proposer rewards and directly optimizes for evidence necessity relative to shortcut contexts. Dr. Free samples relational chains from a knowledge graph and pairs them with aligned passages, giving question generation an explicit multi-hop structure. A generated question receives a positive information-gain reward only when the likelihood of the target answer under the complete evidence passages exceeds the maximum likelihood under all evaluated shortcut contexts. Because this signal is computed from teacher-forced likelihoods, it removes the need for pass-rate estimation and reduces proposer training time by over $7\times$. Experiments on seven open-domain QA benchmarks show that Dr. Free outperforms prior data-free search agents and the supervised baseline, with large improvements on multi-hop QA benchmarks.

Finite Fibonacci and Lucas sums involving zeta values at negative integers can have individual terms far larger than their total. In this paper, we study the distribution of these term magnitudes as the even degree increases. For non-negative weights of at most exponential growth, we prove a finite exponential expansion for the normalized absolute-term sum with an explicit remainder at every fixed order. The normalized magnitudes converge in total variation to a discrete distribution determined by the even exponential generating function of the weights. For the Fibonacci and Lucas sums, this distribution has its unique maximum at index eight and we prove that the same index gives the largest finite-sum term at every even degree at least twenty-six. We also derive the factorial growth of the summation condition number. The signed identities are placed within the established Bernoulli-polynomial reflection framework and extended to a trace-one recurrence family, including repeated roots and a vanishing scale. Exact rational calculations and precision-refined evaluations illustrate the proved bounds and distinguish a fixed cancellation profile from the growing length of the sum.

World models improve sample efficiency by training policies on imagined trajectories, but their usefulness depends on learning representations that capture the information needed for future control. We study whether self-supervised joint-embedding prediction (JEPA) can provide this learning signal for multi-agent reinforcement learning. We introduce MA-JEPA, a stochastic world model that replaces observation reconstruction with prediction of target representations, enabling model-based multi-agent reinforcement learning with centralized training and decentralized execution. A categorical latent state and a causal Transformer are trained with posterior and action-conditioned dynamics prediction objectives and are then used for actor-critic learning from latent imagination. A training-only joint predictor conditions on all agents' local states and actions to predict each agent's next local observation embedding. These predictions are passed through the same local posterior used during real interaction with a centralized critic that is used only for value learning, with execution remaining decentralized. Our experiments show that this architecture performs strongly on SMAC, matching or exceeding the strongest reported comparator mean win rate on four of eight evaluated maps.

Systematic out-of-distribution (OOD) generation remains a critical bottleneck for continuous-time generative models. While standard joint classifier-free guidance (CFG) routinely fails to synthesize unobserved concept combinations, exact decomposed scoring generalizes robustly at the cost of severe computational overhead. In this work, we reveal that compositional binding is not a uniform process but a highly localized phase transition. We identify the semantic bifurcation window - the precise temporal interval where joint and decomposed vector fields meaningfully diverge. Exploiting this dynamic, we propose surgical guidance, a hybrid sampling strategy that restricts exact multi-pass scoring strictly to this critical window. On an OOD bi-digit MNIST testbed, surgical guidance achieves state-of-the-art compositional fidelity at a fraction of the inference cost, yielding a +5.3% absolute improvement in pairwise accuracy over the joint baseline by intervening during just the first 15% of the diffusion trajectory. Furthermore, our empirical analysis uncovers a fundamental topological divide: diffusion models (SDEs) force conceptual resolution immediately at peak noise, whereas Conditional Flow Matching (ODEs) delays structural binding until intermediate features emerge, establishing a new temporal framework for accelerating large-scale generative decoding.

Multimodal graph predictors combine text, images, and relations to classify connected entities. How much of this input is needed to preserve their predictions? We study budgeted representation selection, which chooses a subset of candidate text and image vectors under a separate capacity for each modality. Predictions from the complete candidate input define the classes to preserve. The challenge is that a representation's contribution depends on the other selected inputs, while graph propagation extends its effects across nodes. Our empirical study shows that candidate rankings change with the selected input, while predicted probabilities remain informative after the class stops changing. Updating scores improves selection, and exchanging inputs can improve a subset whose capacity is already filled. These findings lead to GraphSelect, which starts from individual candidate gains and refines the subset through jointly evaluated exchanges. It screens promising removals and additions, accepts an exchange when it reduces the prediction loss, and updates the scores. Experiments on six graphs show higher mean objective recovery than six attribution and explanation methods adapted to the selection task. Across nine trained architectures on two graphs, retaining 20% of the candidate representations per modality gives a mean accuracy drop of 0.10 percentage points relative to full candidate input, preserving classification performance with substantially fewer text and image representations.

Let $T$ be a convolution Calderón–Zygmund operator satisfying Hytönen's non-degeneracy condition. We prove that, for every $p\in(1,\infty)$ and $m\ge3$, there exists a positive constant $C$ such that, for any $t\in[1,\infty)$, \begin{equation*} \sup_{[W]_{\mathcal A_p}\le t} \|T\|_{L^p(W)\to L^p(W)} \ge C t^{1+\frac{1}{p(p-1)}}, \end{equation*} where the supremum is taken over all matrix weights $W:\ \mathbb R^n\to\mathbb C^{m\times m}$ with their weight constants $[W]_{\mathcal A_p}\le t$. If, in addition, the odd part of the kernel is non-degenerate (such as the Hilbert transform and the Riesz transforms), the same conclusion holds for every $m\ge2$.

Recipes such as s1 and LIMO teach a model to reason with little data by showing it the same thousand or fewer worked solutions many times over. Judged when that training ends, the repetition looks harmless. But reasoning models are often trained again, and we find that repetition leaves their reasoning fragile to that next stage, even when the stage has nothing to do with reasoning. We fine-tuned Qwen3.5-9B-Base on its own correct solutions to competition math problems, either drilling a few hundred of them about eight times each or showing many more once; with the same amount of training, both solve about 95% of held-out problems. A single pass of ordinary instruction tuning leaves the once-trained model where it was, while the drilled one falls to 86.0%, and harsher later stages take it to 59.3% or below. A third model that visited the drilled problems just as often, with a new solution at every visit, was unharmed, so the damage comes from seeing the same texts again rather than from having few problems. The break recurs with a stronger model's traces, in further training runs and on other models and tasks. It is also cheap to undo: the reasoning is suppressed rather than erased, and five updates of reasoning training bring almost all of it back, as does brief training on the reasoning format with almost no mathematics. Fresh solutions prevented the damage, and so did replaying 6.25% of the original solutions in a gentler later stage, so our claim concerns later training without such replay. Sharpening alone does not explain the break, since a model sharpened three-quarters as much without repetition was unharmed. On a skill the base model could not perform within a token budget, repetition mainly cost learning.

3D Visual Query Localization (3DVQL) retrieves the latest contiguous occurrence of a queried object in an RGB–point-cloud sequence and predicts a 9-DoF cuboid for every response frame. The query is captured independently of the search sequence, so its annotated pose may differ from how the object appears in the search frames. The benchmark baseline predicts cuboids after feature modeling, leaving their geometry unused for subsequent feature refinement. We investigate whether complete intermediate cuboids can improve query and proposal representations before final decoding. We introduce Progressive Geometric Learning for 3DVQL (PGL-3D), a predict–select–refine–re-predict framework that uses intermediate cuboids to guide the aggregation of search evidence and update query and proposal representations. A shared head first predicts a complete cuboid for every proposal. Query–Tube–Memory (QTM) then selects reference observations by combining proposal association, cuboid quality, frame response, and target absence, since association confidence alone establishes neither target presence nor geometric accuracy. The center, size, and orientation of each selected cuboid define soft pooling weights over query-conditioned proposal features. The pooled memory updates the query and proposal representations, and the head re-predicts from the updated features. A training-only objective, ST-D9O, supervises cuboid geometry at every stage by adding boundary, signed-distance, and soft-overlap terms to parameter regression. PGL-3D achieves a mean stAP of $0.270 \pm 0.004$ on 3DVQL, compared with $0.044$ reported for LaF. Ablations support the benefits of geometry-guided feature updates, while stage-wise analyses show improved cuboid accuracy. Replacing the geometry objective in our PROT3D reproduction with ST-D9O improves mAO on GSOT3D from $21.63\%$ to $25.78\%$. Our code and models will be released.

The Favard length of a planar set is the average length of its orthogonal projections. We provide sharp estimates for the Favard length decay of a random variant of the non-homogeneous four corner Cantor sets introduced by Garnett in the 1970s. As a corollary, we obtain examples which show that Mattila's classical lower bound on the Favard length in terms of the Riesz capacity is sharp for all possible decay rates. We also obtain a new family of examples of sets with positive analytic capacity and zero Favard length.

Success-driven imitation is a fundamental aspect of social learning, yet group interactions beyond pairwise require more subtle imitating rules since success can be observed at either the group or individual level. How such heuristic imitating rules based on these distinct levels of social information compete and coevolve with collective behavior, particularly cooperation, remains poorly understood. Here, we address this issue by studying the coevolutionary dynamics of cooperation in public goods games and individual- and group-based imitating rules on hypergraphs. We distinguish three rules according to whether payoff-biased imitation operates at the individual level, the group level, or both. Our results identify a mutual reinforcement between cooperation and group-biased imitation: preferentially learning from successful groups promotes cooperation, while cooperators in turn favor group-biased imitation. We confirm this synergy between cooperation and group-biased imitation across diverse synthetic and empirical higher-order network populations. We further show that the subtle scale and organization of group interactions critically shape this coevolution, with intermediate group sizes providing the greatest advantage for cooperation. Our results reveal how learning whom to imitate can jointly evolve to shape collective cooperation.

For $3\le n\le6$, we determine the sharp isoperimetric bound for the mass-normalized first weighted Neumann eigenvalue on bounded Euclidean domains. For every finite nonnegative nonatomic measure $\mu$ on $\overline\Omega$, \begin{equation*} \overline\lambda_1^N(\Omega,\mu)\le\frac{n(n-1)}{n-2}\omega_n\sin^2\vartheta_n\left(\frac{|\Omega|}{\omega_n}\right)^{(n-2)/n}, \end{equation*} where $\omega_n$ is the volume of the unit ball and $\vartheta_n\in(\pi/2,\pi)$ is the angle at the first stationary radius of the regular rotational harmonic-map profile. The bound is attained on balls by a smooth positive radial density. In fact, we affirmatively resolve Question~1.12 posed by Vinokurov \cite{Vinokurov2026}. Furthermore, for admissible domains with finite boundary measure, we also prove the sharp strict Steklov bound \begin{equation*} \sigma_1(\Omega)|\partial\Omega|\,|\Omega|^{(2-n)/n}<\frac{n(n-1)}{n-2}\omega_n^{2/n}\sin^2\vartheta_n, \end{equation*} whose constant is approached by $C^1$ perforated domains.

Driven by the rapid growth of immersive teleconferencing and generative spatial audio, efficient low-bitrate coding of first-order Ambisonics (FOA) has become increasingly important. In this work, we develop a lightweight parametric FOA codec that retains the standard Directional Audio Coding analysis and synthesis while redesigning the spatial metadata quantization scheme. Rather than quantizing direction-of-arrival (DOA) and diffuseness independently, we combine them into a 3-D directivity vector and jointly quantize these vectors across frequency bands via residual vector quantization (RVQ). The RVQ codebooks are optimized within minutes via stage-wise k-means without backpropagation, yielding a constant-bitrate (CBR) representation that decouples metadata rate from the number of frequency bands. Evaluations show that our approach outperforms low-bitrate perceptual codecs in FOA reconstruction, remains competitive with neural codecs on the downstream Sound Event Localization and Detection task, and maintains robust performance when paired with different external monaural codecs. Given its lightweight training, strong performance, and CBR design, we consider the proposed method to be a favorable and reproducible baseline for future FOA codec research.

The rapid expansion of large-scale AI data centers (AIDC) is introducing new stability challenges, particularly in weak or low-inertia networks characterized by fast, step-like demand variations and strict requirements on voltage and dynamic performance. This paper investigates the use of grid-forming (GFM) Enhanced STATCOMs (E-STATCOMs) to support reliable integration of such facilities. A power-admittance-based linear modelling framework is developed to capture system interactions and is validated through detailed EMT simulations. The results demonstrate that E-STATCOMs provide fast, well-damped responses to abrupt load changes while effectively mitigating low-frequency oscillations and interactions with network resonances. By enabling tunable dynamic behavior via a load balancer, virtual impedance, and coordinated active-reactive power support, the proposed approach allows precise shaping of system response and improved regulation at the point of connection. These features make E-STATCOMs a flexible and scalable solution for integrating large data centers into weak grids and long transmission systems, supported by a design-oriented framework that facilitates parameter selection and performance assessment without extensive reliance on EMT studies to meet grid codes and AIDC interconnection requirements.

Let $M^n\to\mathbb{S}^{n+q}(1)$ be a closed connected minimal immersion, where $n\ge3$, and set $Q=S+\lambda_2$, with $S=|h|^2$ and $\lambda_2$ the second largest eigenvalue of Lu's fundamental matrix. We determine the sharp codimension range for Lu's second-gap conjecture. For every $2\le q\le n$, there exists $\gamma_{n,q}>0$ such that, if $Q$ is constant and $Q>n$, then $Q\ge n+\gamma_{n,q}$. Conversely, for every $q\ge n+1$, we construct closed connected homogeneous minimal embeddings, followed when necessary by totally geodesic inclusions, with constant scalar curvature and constant $Q$-values dense in $(n,2n)$. Thus, in every dimension $n\ge3$, Lu's conjecture holds precisely for $q\le n$. Combined with the theorem of Peng-Terng for hypersurfaces and the recent resolution of the two-dimensional case, this gives a complete resolution of Lu's second-gap conjecture: for every $n\ge2$, the conjecture holds exactly when $q\le n$ and fails when $q\ge n+1$. In codimension two we further obtain the explicit admissible gap $\gamma_{n,2}=\exp(-10^{16}n^2)$.

Transferring human demonstrations to dexterous robots remains challenging because differences in hand morphology and contact dynamics often cause retargeted motions to fail at producing the intended object behavior. We present FoLD, a framework for learning dexterous manipulation of articulated objects through explicit force guidance. FoLD compute compensatory force fields from human demonstrations together with the robot's current interaction state, yielding a force prior that promotes the demonstrated object motion. This force prior informs a residual policy that adapts retargeted hand motions to the contact requirements of the task. We evaluate FoLD on a public benchmark for articulated object manipulation, where it consistently outperforms state-of-the-art baselines across tasks and embodiments. We further validate FoLD on real dexterous robot platforms, demonstrating successful transfer of human manipulation skills to robot execution. Here is the link of our project page: https://gghgghgghgg.github.io/FoLD-project-page/.

Automatic relevance determination (ARD) is the standard device for input selection in Gaussian process (GP) regression. By giving the covariance kernel a separate lengthscale for every input and learning those lengthscales by maximizing the marginal likelihood, ARD lets the data decide which coordinates matter: irrelevant inputs receive very large lengthscales and are effectively switched off. We trace this mechanism to the Bayesian Occam's razor embodied in the marginal likelihood, derive the gradient through which it prunes inputs, and emphasize that ARD delivers effective rather than exact sparsity. We review the algorithms used in practice and the rules that turn lengthscales into selections, and we survey the asymptotic theory, distinguishing the fixed-domain identifiability obstruction on the lengthscales from the high-dimensional selection-consistency guarantees recently established for hierarchical GP priors, and noting what remains open for plain ARD. We compare ARD with spike-and-slab priors, sparse axis-aligned and global-local shrinkage priors including the Bayesian lasso and horseshoe, penalized likelihood kriging, sensitivity and projection criteria, and additive kernels. We argue that ARD endures because of its seamless integration with kernel learning, universal software support, and low cost, and we close with its limitations and remedies.

Automatic relevance determination (ARD), the default tool for variable selection in Gaussian-process (GP) regression, ranks inputs by inverse lengthscales – which measure how fast a function varies, not how much an input contributes to prediction – and offers no calibrated rule for deciding which inputs to keep. The prediction-centred alternative, the derivative sensitivity $\nu_j = \mathbb{E}[(\partial f/\partial x_j)^2]$, is available in closed form from a fitted GP, but turning it into a selection rule is harder than it looks: at a null input the estimator is a degenerate quadratic form, so Wald and Bernstein-von Mises cutoffs are anti-conservative, and the natural residual bootstrap is mis-scaled. We show that a studentized multiplier bootstrap of the GP derivative process repairs both, prove its validity through an invariance principle for quadratic forms, and obtain asymptotic family-wise and false-discovery-rate control across inputs. Over 100 replications the rule controls FDR wherever inputs are truly null, while uncalibrated derivative rankings breach the target by up to 2x and a Bernstein-von Mises cutoff by 2.2x; at matched FDR it loses no power; it holds under a Matérn kernel and input correlation up to 0.99; on real data with planted and authentic null inputs it admits 5-12x fewer spurious inputs; it costs 5-18% of the GP fit; and a block-averaged variant retains validity at cost linear in $n$.

Fine-tuning anguage models (LLMs) with first-order optimizers requires a memory several times larger than that required for inference. Memory-efficient zeroth-order optimization (MeZO) sidesteps this cost by estimating gradients from forward passes only. However, for BitNet architectures, a family of LLMs with ternary {-1,0,1\} weights and 8-bit activations, fine-tuning requires updating full-precision latent weights, and thus the memory footprint of MeZO no longer matches that of inference. A promising solution is to finetune only a subset of the latent weights, but existing sparse zeroth-order (ZO) methods either ignore the ternary structure or require first-order gradient information to build a sparse mask, which is at odds with the purpose of ZO fine-tuning. We propose TerMeZO, a sparse MeZO scheme that exploits the geometry of the ternary quantizer itself to identify the latent weights that are more likely to change values during fine-tuning, at no additional data or memory cost. Our convergence analysis shows that TerMeZO can converge faster than full-parameter MeZO, owing to its optimized reduction of the fine-tuning effective dimension. We run extensive experiments on BitNet models ranging from 1B to 3B parameters, spanning classification, instruction-following, and mathematical reasoning tasks. TerMeZO matches or exceeds the performance of full-parameter MeZO while substantially reducing the fine-tuning memory footprint.

Resolving indirect calls is central to call-graph construction for C. Scalable type-based analyses such as MLTA use type information in LLVM IR to associate indirect calls with functions assigned to the corresponding structure fields. However, a single pointee type often misrepresents the memory a pointer addresses, and LLVM 17 removed pointee types in favor of opaque pointers. Therefore, field-sensitive analyses lose their matching key. Recovering the erased types restores the matching key but still misses the relation that the type encoded: which functions the program assigns to the field. We present Facet, to our knowledge the first analysis that reconstructs this dispatch relation over opaque IR. Facet identifies the structure field from which an indirect call loads its function pointer. It separately recovers the functions assigned to that field through initializers, stores, and aggregate copies. It then joins the two by field identity, without requiring an end-to-end value-flow path. Facet classifies proposed call-graph changes under distinct evidence rules for edge addition and removal and records the assumption behind each refinement. An LLM decides only the residual cases among symbolically bounded candidates. One analysis yields both a recall-preserving call graph and a refined call graph. On 14 C programs, Facet reduces the mean target-set size from 25.9 to 5.2 and raises observed recall from 0.79 to 0.99. Its recovered field identities agree with typed IR at 98.1% of jointly resolved sites. Applied to bug detection, the refined call graph found 17 deep bugs in C software from nginx to the Linux kernel, three of them latent for over a decade; 12 are confirmed.

Current grasp diffusion models provide rich priors for generation, yet their object-centric approach can violate the kinematic and collision constraints imposed by the embodiment and the environment. Existing embodiment-aware methods primarily perform local corrections around generated grasps through gradient guidance or optimization, making it difficult to recover from fundamentally infeasible modes. We present Steer2Grasp, a training-free, embodiment-agnostic framework for inference-time grasp steering that adapts a frozen Cartesian grasp diffusion model using deployment-specific rewards. Through Feynman-Kac (FK) inspired particle reweighting and resampling, the method reallocates population mass from infeasible to high-reward grasp modes, enabling population-level mode transitions without modifying the pretrained diffusion model or requiring differentiable constraints. The framework enables a unified treatment for single and dual arm grasping through reachability and collision aware rewards, followed by gradient free gripper level local refinement. Across diverse objects, robot embodiments, and constrained environments, our method substantially improves feasible grasp generation while maintaining proximity to the underlying grasp prior.

Point defects in wide-bandgap semiconductors offer spin and photonic qubits in a solid-state platform, making them important building blocks for quantum information processing, communication, and sensing. While these systems have the strong advantage of room-temperature operation, intrinsic electron-phonon interaction induces broad phonon-sideband emission and weak zero-phonon-line transitions, limiting efficient spin-photon interfaces and scalable photon-mediated interaction. Here, we introduce a controlled heterophase interface based on the remote epitaxy technique as a crystal heterogeneity-engineering strategy. While stacking faults are treated as imperfections to be eliminated, our results instead show that crystal-phase interfaces can provide an additional degree of freedom to engineer defect-phonon interaction beyond the intrinsic properties of a single crystal. In comparison with single-phase 4H-silicon carbide (SiC) membranes, Vsi in 3C/4H heterointerface SiC exhibits a drastic enhancement of zero-phonon optical transitions, with a substantially narrower zero-phonon linewidth of 4.71 meV even at room temperature. These results establish heterophase-interface engineering as an effective route for tailoring defect-phonon interactions and realizing bright room-temperature quantum emitters with enhanced zero-phonon optical transitions.

Federated learning (FL) is increasingly deployed as a managed learning service rather than as a set of isolated training jobs. In networked edge environments, dependent FL service flows must coordinate heterogeneous clients, non-IID data, fluctuating communication latency, and precedence-constrained tasks under service-level completion requirements. These coupled factors make participant management central to both time-totarget performance and learning stability. This paper proposes A-CoDa, an adaptive clustered coordination framework for managing dependent FL flows. A-CoDa first uses label-distribution divergence (LDD)-based greedy-balanced clustering to construct statistically coherent and size-aware client groups, which serve as a scalable management abstraction. Building on this structure, we design FedMIX, an uncertainty-aware intra-/inter-cluster participation mechanism that ranks clients by a loss-latency-uncertainty utility and adaptively controls cross-cluster probing according to training progress and latency conditions. A dependency-aware DAG scheduler then orchestrates layer-wise task execution so that parallelism and precedence constraints are jointly respected. We further provide a convergence analysis that frames the result as a sufficient loss-domain design bound, explicitly relating the attainable error floor and sufficient communication rounds to LDDinduced sampling mismatch, residual distribution shift, local-SGD drift, stochastic variance, and adaptive probing budgets. Experiments on handwriting, wearable-sensing, product-image, and medical-imaging tasks evaluate A-CoDa under dependent FL workflows and demonstrate its effectiveness in reducing end-toend completion time while maintaining competitive accuracy.

Coupled rollouts can reduce the noise of counterfactual action comparisons, but two issues prevent standard common-random-number constructions from serving as a general learning primitive in imperfect-information environments. First, an invalid coupling may expose hidden state, synchronize endogenous policy randomness, or misalign chance events after counterfactual histories diverge. Second, in multi-action policy optimization, lower return-contrast variance is not by itself the relevant objective: the optimizer depends on the return covariance matrix after projection through the local policy-gradient geometry. We introduce observation-safe counterfactual coupling (OSCC), a framework that defines an admissible class through marginal preservation, information-state safety, branch-local policy randomness, semantic event alignment, and trace-before-oracle replay. We derive a gradient-aware coupling criterion showing that, for marginal-preserving couplings, policy-gradient noise changes are determined by policy-Jacobian-weighted off-diagonal return covariance. This motivates OSCC-Select, a calibration-only selector that chooses among independent, root-only, continuation-only, and fully coupled rollouts using separate safety and gain certificates. Its gain target combines projected gradient noise with measured physical sampling cost and falls back to independent sampling whenever a simultaneous lower confidence bound does not certify improvement. On 100,000 fixed-root Leduc comparisons, the fully coupled CP-GRPO instantiation reduces return-contrast variance from 41.1158 to 18.1441, a 55.87% reduction, while preserving the declared branch marginals. With three actions, OSCC-Select chooses continuation coupling and attains gradient-noise trace 0.0783 versus 0.0917 for return-variance selection. Increasing calibration from 64 to 2,048 groups raises certification from 0.327 to 0.995.

Direct torque sensing is a growing need in the robotics community to enable precise control and interactions where torque estimation from motor current is not sufficient. This paper presents a novel disk-shaped magnetoelastic torque sensor with a compact axial envelope of about 1 cm, suitable for integration in robotic joints. A four-magnetometer architecture is used to measure the field modulated by the stress affecting a narrow magnetized region while rejecting the effects of parasitic cross forces. A custom-designed magnetic shield enhances the torque sensitivity while reducing the external stray fields by a factor 7x. The device measures the torque with an accuracy of 1.34 %FS relative to the 50 Nm full scale (FS). The paper details the development of the sensor through the mechanical design, the magnetization procedure, and the experimental validation. The results demonstrate the potential of the proposed sensor for robotic applications.

We develop an analytical framework for studying a family of interacting particle systems that we refer to as the multi-particle Oja flow. The common interaction structure underlying this family arises in a variety of applications, including higher-order synchronization models, opinion dynamics, and self-attention dynamics derived from Transformer architectures, motivating a unified mathematical framework for their analysis. We first establish well-posedness for both the finite-particle and the mean-field systems. We also prove convergence of the finite-particle dynamics to the mean-field equation as the number of particles tends to infinity. We then investigate low-dimensional structures underlying the dynamics. In particular, we show that the mean-field solution admits a representation through a time-dependent normalized linear transformation acting on the initial distribution, with the transformation governed by a matrix ODE. Building on this representation, we identify a broad class of parametric families that are invariant under the mean-field dynamics. When the initial distribution belongs to one of these families, the evolution remains within the same family for all time and can therefore be characterized through the dynamics of the corresponding family parameters. As a concrete example, we develop this reduction within the angular central Gaussian family and illustrate how the resulting parameter dynamics provide reduced descriptions of the mean-field evolution for several representative systems.

Looped reasoners spend test-time compute by iterating a weight-tied map, but a small residual does not mean the state is a fixed point when that map lives in unconstrained latent space. We propose Geometric Fixed-Point Reasoning (GFPR), in which the iterated state is the prediction itself: a field of categorical beliefs on a product of simplices, whose argmax is the answer at every step. Because the state is a belief, task structure can be imposed through compact convex relaxations, either as structured readouts or directly in the recurrent state; in the latter case the update remains a continuous self-map, so a fixed point exists for any parameters. At about 7M parameters, GFPR reaches 95.1% exact match on Sudoku-Extreme, 92.0% on Maze-Hard, and 100% sequence accuracy on S_5 length 128, above the published FPRM numbers at the same scale. The same update also trains a 201M language model on FineWeb-Edu in which each site is a distribution over the vocabulary; with 24 Picard steps it is above GPT-2 small on four zero-shot multiple-choice tasks and above GPT-2 medium on ARC-Easy.

Driven by ongoing standardization efforts, Wi-Fi sensing has recently emerged as a promising functionality within wireless local area networks (WLANs). However, due to the use of standard null-data packets (NDPs) to estimate the channel, conventional Wi-Fi sensing systems are susceptible to spoofing attacks. To counter this threat, we propose a lightweight tag-embedding scheme that enables physical layer authentication (PLA) of sensing signals. Specifically, a secret tag is superimposed onto the orthogonal frequency division multiplexing (OFDM) channel sounding signal, allowing legitimate WLAN receivers to verify signal authenticity and detect spoofed transmissions. A unique tag is generated for each transmission to avoid replay attacks, and the subcarrier occupancy of each tag is randomized to minimize the mean channel estimation error at each subcarrier. Simulation results demonstrate reliable authentication performance with a 95% correct authentication rate at a signal-to-noise ratio (SNR) of 10dB, while introducing an additional channel estimation error of only 0.52dB at legacy receivers that are unaware of the embedded tag. Thus, our proposed scheme provides an effective and backward-compatible countermeasure against spoofing attacks, enhancing the trustworthiness of future Wi-Fi sensing systems.

A speech neuroprosthesis decodes attempted speech from brain activity and ends by rescoring the decoder's candidate sentences with a language model of several billion parameters, the only component that needs a GPU. Replacing that model with a cheaper one is hard: general language models asked to pick one sentence from a list answer from where a label sits in the list rather than from the sentence itself. We pose rescoring as a single typed decision, one call that returns a probability for every candidate, served by Jev, a hosted model trained for calibrated decisions, and combine it with the decoder's own score. On 978 held-out sentences from a participant with ALS, where the published decoder alone reaches 8.1% word error, Jev reaches 7.5% against 7.8% for both OPT-6.7b and Qwen2.5-7B; with the decoder's weight re-tuned, 6.9% against 7.2% and 7.4%. Jev is ahead in all four comparisons and at most 0.2 points behind at the 95% bound. It costs 0.07 USD per thousand sentences and needs no GPU; a dedicated GPU running a 7B model is cheaper per sentence only above 43% utilisation, far beyond what one user generates. End-to-end latency over the internet is 262 ms, of which 62 ms is spent at the provider, the same order as a 7B model on a local GPU (27 ms) but not faster.

As a non-abelian analogue of the Hodge locus, Simpson introduced the non-abelian Noether–Lefschetz locus and conjectured its algebraicity for $\mathbb Z$PVHS. In this paper, we study this question on the moduli space of curves $\mathcal M_g$. For a non-unitary representation \[ \rho:\pi_1(\Sigma_g)\longrightarrow \mathrm{SL}_2(\mathbb R) \] which admits a $\mathbb R$PVHS of weight one, we prove that algebraicity of a positive-dimensional non-abelian Noether–Lefschetz component is equivalent to discreteness of $\operatorname{im}\rho$, and in this case the component is precisely a marked fixed-target orbifold Hurwitz component. As applications, we construct two explicit rational families by slit surgery: one with non-discrete monodromy and non-algebraic Noether–Lefschetz image, and another with discrete monodromy whose period map is nevertheless non-uniformizing. The latter gives an affirmative answer to a question of Baldi–Lam concerning $\mathbb Q$PVHS.