Robotics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Tuesday 8 September
Messages from electronic devices are conventionally received as text, audio, or radio signals. But robots move with rich, articulate motion in the real world, opening up the possibility of transmitting messages through motion itself. In this paper, we consider the problem of motion-based communication, where we seek to modify a robot's movements so as to transmit messages detectable from remote sensing (e.g., video or motion capture), without degrading policy performance. We introduce a method for messaging through motion capable of encoding arbitrary message content over short payloads - such as an agent's current intent - as noise in any pre-trained policy's actions. This brings a new kind of robustness to robot communication: this 'physical' channel complements standard wireless communications channels but does not depend on them, requiring no extra hardware nor the establishment of a direct link to the robot. We systematically characterize the space of encoding schemes and derive design heuristics, then validate them across simulated environments and real-robot deployment; on real robots running at 50 Hz, four robots jointly recover an 8-bit message at an aggregate 0.8 bits/s.
Computation and Language
Experience Funnel: A State-Policy Alternating Loop for Self-Evolving Agents
Autonomous agents powered by large language models (LLMs) continuously accumulate experience through interaction, creating an opportunity to improve future behavior through self-evolution. A fundamental challenge is how to transform abundant, task-specific interaction experience into reusable model competence without sacrificing the ability to adapt rapidly to newly observed evidence. Explicit textual states, such as skills and agent harnesses, provide fast, human-readable and editable adaptation, but incur persistent dependence on external context; parametric policies provide compact and reusable competence, but are substantially slower to update. We present Experience Funnel, a self-evolving framework that couples fast state adaptation with slow policy consolidation in an alternating loop. Interaction trajectories are first distilled into an explicit textual state, where newly acquired experience can be rapidly incorporated and validated. The framework then selectively identifies state-enabled behavior that remains useful across state revisions and consolidates it into the policy through transition-aware distillation. The updated state–policy pair subsequently generates new rollouts, providing fresh evidence for the next round of state adaptation and policy consolidation. Experiments across diverse agent benchmarks show that Experience Funnel consistently improves agent capability over state-only evolution and policy-internalization approaches, while progressively converting useful explicit experience into autonomous policy competence.
Read the full abstract
High Energy Physics (Phenomenology)
Probing the Inert Scalar Sector of the Inert Doublet Model via Vector-Boson Fusion at a Muon Collider
We investigate the sensitivity of a future high-energy muon collider (MuCs) to the inert scalar sector of the Inert Doublet Model (IDM) through a comprehensive set of $2\to4$ electroweak gauge-boson fusion processes. The scalar spectrum consists of two CP-even neutral scalars: $h$ and $H$, one CP-odd neutral scalar $A$, and a pair of charged scalars $H^\pm$. We assume the neutral scalar $H$ to be the lightest inert state and the dark matter (DM) candidate, while the observed $125$~GeV Higgs boson is identified with the scalar of the Standard Model-like doublet. We consider charged Higgs pair production: $\mu^+\mu^-\to H^+H^-\nu\bar{\nu}$ induced by WW fusion, as well as the associated channels: $\mu^+\mu^-\to HH^\pm\mu^\mp\nu$ and $\mu^+\mu^-\to AH^\pm\mu^\mp\nu$, induced by $WZ$ and $W\gamma$ fusion. Owing to their Vector Boson Fusion (VBF) origin, the corresponding cross sections increase significantly with collider energy, making them particularly well suited for exploration at a high-energy MuCs. We first study the $2\to 2$ subprocesses $WW\to H^+ H^-$ and $WV\to H^+ S_i$, $V=Z, \gamma$ and $S_i=H$ or $A$ and establish the destructive interference among the contributing diagrams. After imposing theoretical consistency requirements and current experimental constraints, including those from DM and collider searches, we evaluate the corresponding signal rates at $\sqrt{s}=3$, $10$, and $20$~TeV. We further perform a detector-level Monte Carlo analysis combined with a cut-based signal-background selection. Our results demonstrate that the sensitivity to inert scalar production increases significantly with collision energy, establishing the 10 and 20~TeV MuCs stages as powerful probes of the IDM inert scalar sector.
Read the full abstract
Algebraic Geometry
Hirzebruch signature theorem on Hochschild homology, with applications to derived invariance of Hodge numbers
We prove a refinement of Hirzebruch's signature formula on the individual Hochschild diagonals of a smooth projective complex variety. Natural Hermitian forms, obtained by symmetrizing a Todd-corrected pairing, have explicitly computable positive and negative indices and are preserved by derived equivalences. On diagonals of the same parity as the dimension, their signatures separate the even and odd contributions to the Hochschild-Kostant-Rosenberg decomposition; on the other diagonals, their radicals and signatures are governed by the first Chern class. We deduce the derived invariance of $h^{n-2,0}$, $h^{n-1,1}$ and $h^{n-3,1}$ in any dimension, and of $h^{2,0}$ when the canonical bundle is trivial. Combining the invariance of $h^{2,0}$ with the multiplicative structure of Hochschild cohomology and Verbitsky's orthogonal action, we give a new short proof of the theorem of Huybrechts and Nieper-Wißkirchen that a derived partner of an irreducible holomorphic symplectic variety is again irreducible holomorphic symplectic. In dimension five with trivial canonical bundle, the signature invariants and the Libgober-Wood identity together establish the derived invariance of all Hodge numbers
Read the full abstract
Classical Analysis and ODEs
Estimates for $L^p$ variants of Gowers norms
Motivated by a log-convexity question of Bennett and Tao, we consider Gowers-type functionals defined by $L^p$ norms of multiple autocorrelations. We prove sharp bounds in terms of $L^q$ norms and characterise the near-extremisers on Euclidean spaces as well as locally compact abelian groups. We also establish two families of degree-lowering inequalities and study their near-extremisers. As a byproduct of this broader theory, we show that the constant in the log-convexity estimate for Gowers norms is strictly less than unity, confirming the aforementioned conjecture of Bennett and Tao. Finally, we characterise the $p$ for which the Gowers-type functionals necessarily satisfy the triangle inequality on nonnegative measurable functions.
Read the full abstract
Number Theory
Sharp order-preserving integer models for short additive equalities
We ask how small an increasing integer model of a finite real set can be while preserving all equalities between equal-length sums of at most $q$ elements, with repetitions allowed. The increasing correspondence must preserve exactly which sums are equal; the signs of unequal comparisons may change. Let $H_m(q)$ be the least diameter sufficient for every ordered real set of size $m$. For every integer $q\ge2$ we prove $H_4(q)=q(q+1)$ and $H_5(q)=q^2(q+1)$, and we determine $H_6(2)=24$. The proofs use the linear space determined by the additive equalities and the inequalities prescribing the order of the elements. They also give bounds for larger sets, determine the first two asymptotic terms for each fixed $m\ge4$, and yield further families of sharp examples. An application bounds the integer alphabets needed to realize finite additive-square-free spectra. The supplementary data support the finite six-element classification.
Read the full abstract
Computer Vision
Data-Efficient Crosswalk Segmentation from Overhead CCTV via Confidence- and Geometry-Guided Pseudo-Labeling
Pixel-level annotation of fixed traffic-camera imagery is expensive, while crosswalk models trained from street-level imagery face a substantial viewpoint and appearance shift when applied to elevated CCTV. We investigate a data-efficient target-domain pipeline using 241 manually annotated CCTV images and 5,926 unlabeled CCTV frames. A source-domain experiment trains a 31.0M-parameter custom U-Net on 3,300 first-person-view (FPV) images and obtains 93.05% IoU on its 330-image FPV test split. This result is a source baseline, not transferred performance: the released CCTV notebook instantiates a 42.0M-parameter DeepLabV3-ResNet50 from torchvision weights, and no compatible mapping from the U-Net checkpoint is implemented. Training on 201 manual CCTV images and selecting on 40 held-out manual masks yields 88.91% IoU. The model then predicts all unlabeled frames; image-level certainty and a largest-component area prior rank the candidates, and the top 1,000 attain mean certainty 0.976 and mean combined score 0.988. A repository audit shows that the reported second-stage 98.52% IoU was measured on a 150-image split containing only teacher-generated pseudo-masks. Because of a directory-layout mismatch, the executed combined-data loader found zero manual samples and split 1,000 pseudo-labeled samples into 850 training and 150 evaluation samples. We therefore report 98.52% as internal pseudo-label agreement rather than human-ground-truth accuracy. The defensible target-domain result is 88.91% IoU on the 40 manual validation images. Batch-one FP32 inference at 512 x 512 requires 12.98 ms, corresponding to 77.03 FPS, on an NVIDIA RTX A6000 48 GB GPU. These findings support the practicality of confidence-and-geometry filtering while also showing why pseudo-label evaluation must remain isolated from the labels used for self-training.
Read the full abstract
Materials Science
Zirconium Carbide as a High-Temperature Benchmark for the Beyond Quasi-Harmonic Method
Molar heat capacity at constant pressure, C_p (T), is a central thermodynamic quantity in materials science, but it remains difficult to calculate accurately from first principles when anharmonic vibrational effects become important. In this work, rocksalt ZrC is used as a high-temperature benchmark for the Beyond Quasi-Harmonic (BQH) method, a prepared-supercell first-principles approach that extracts anharmonic vibrational energy directly from density-functional-theory energy evaluations. The calculated BQH molar heat capacity is compared with CALPHAD reference values, a fully anharmonic thermodynamic-integration calculation, a calculated C_v (T) curve from the present work, and a quasi-harmonic calculation. The BQH curve gives a substantial improvement beyond C_v (T) and beyond the quasi-harmonic approximation. Because ZrC is electronically conductive, an electronic heat-capacity correction was also estimated from the density of states at the Fermi level and added to the vibrational BQH result. This correction brings the calculated molar heat capacity into close agreement with the fully anharmonic theoretical benchmark through approximately 1200 K. These results show that the BQH method fully captures important anharmonic phonon contributions missing from ordinary quasi-harmonic calculations.
Read the full abstract
Numerical Analysis
On the well-posedness and efficient approximation for the classical Melan equation in suspension bridges
The classical Melan equation modeling suspension bridges is considered. We first study the explicit expression and global properties of the analytical solution for the simplified “less stiff” model, based on which we derive an a priori estimate for the original classical Melan equation and establish, under an explicit load condition, that it has a unique solution with nonnegative integral under downward live loads, thereby showing the uniqueness of the corresponding deflection curve in the engineering setting. We also develop an efficient iterative approximation method by taking the solution of the simplified “less stiff” model as the first iterate, and prove its geometric convergence with explicit error estimates. The applicability and computational efficiency of the method are demonstrated through calculations for two actual bridges, which also quantify the influence of the nonlinear nonlocal term on the solution and clarify the relationship between the simplified and original models. Several engineering observations are verified and explained, and some related open problems are suggested.
Read the full abstract
Group Theory
Finiteness Properties of Fibre Products over Virtually Nilpotent Quotients
The $n$-$(n+1)$-$(n+2)$ theorem, recently established by Cohen and Shusterman, says that if two groups of type $\mathrm{F}_{n+1}$ map onto a common quotient $Q$ of type $\mathrm{F}_{n+2}$ and one of the two kernels is of type $\mathrm{F}_n$, then the associated fibre product is of type $\mathrm{F}_{n+1}$. We prove a stronger, symmetric variant of this theorem when $Q$ is virtually nilpotent. More precisely, for $i\in\{1,2\}$, let $1\to N_i\to \Gamma_i\overset{\small\pi_i}{\to} Q\to 1$ be short exact sequences of groups and let $P=\{(\gamma_1,\gamma_2)\in\Gamma_1\times\Gamma_2\colon \pi_1(\gamma_1)=\pi_2(\gamma_2)\}$ be their fibre product. Let $k,l,m,n\in\mathbb N_0$, assume that $N_1$ and $N_2$ are of type $\mathrm{FP}_k$ and $\mathrm{FP}_l$, respectively, and that $\Gamma_1$ and $\Gamma_2$ are of type $\mathrm{F}_m$ and $\mathrm{F}_n$, respectively. We prove that $P$ is of type $\mathrm{F}_{\min\{k+l+1,m,n\}}$. The same formula holds with the homological finiteness properties $\mathrm{FP}_r$ in place of $\mathrm{F}_r$. This provides an alternative proof of the homotopical and homological Virtual Surjections Theorem, which gives a finiteness criterion for subgroups of direct products in terms of their embedding in the ambient product.
Read the full abstract
Mathematical Physics
On spatial decoherence in many-body systems
We study the dynamics of a many-body quantum system strongly interacting with a bosonic reservoir. The coupling is given by a potential operator for the system and it is linear in the field. We show that in the limit of infinite coupling strength the system undergoes instantaneous spatial decoherence. To resolve the decoherence process in time we consider large coupling strengths $\lambda$ and short times scaling as $t\propto \lambda^{-\alpha}$, with $\alpha\ge0$, in the limit $\lambda\rightarrow\infty$. On short time scales $\alpha>1$ the dynamics is trivial while on longer ones $0\le\alpha<1$ the decoherence is instantaneous. We show that the decoherence process is resolved exactly for $\alpha=1$, defining the fine-grained time scale $\tau=\lambda t$. We construct an approximate effective evolution map of the many-body system with controlled error estimates. Generically, the effective dynamics is markovian, but not given by a dynamical semigroup. As a physical application we give a rigorous description of the phenomenon of localization of macroscopic quantum objects in position space.
Read the full abstract
Human-Computer Interaction
To Stop or Not to Stop: Exploring the Intention-Behavior Gaps in Smartphone Usage
As smartphones become integral to daily life, researchers have sought to identify when the use becomes problematic. Previous studies have operationalized problematic smartphone usage (PSU) from either an intention or a behavior perspective. Both risk delivering interventions not welcomed by users. We propose a novel approach to operationalizing PSU as the intention-behavior gap (IBG). We collected self-reported data on intentions to stop phone usage, alongside usage behavior data, from 37 participants over two weeks. We calculated IBG, examined effects of demographic and contextual variables, and developed machine learning models to predict IBG in real time. We found that IBG was explained by gender, time, app, and input interactions, among other factors. Intention was predicted most accurately with only personal data, whereas behavior and IBG were predicted most accurately with both personal and global data. Our findings can inform the design of future intervention tools optimized for timing and adaptive intensity.
Read the full abstract
Quantum Physics
A Platform-aware Compilation Framework for Fault-tolerant Quantum Computation
The compilation of an algorithm can vary significantly with the choice of physical hardware platform and error correction model. Yet, current compilation frameworks typically commit to a single architecture-hardware configuration, making it difficult to assess resource estimates across platforms. We present a platform-aware compilation framework that re-compiles a quantum circuit into a hardware-compatible instruction set as well as fault-tolerant operations and provides end-to-end resource estimates in terms of physical-qubit count, time-to-solution, and classical processing time. We benchmark the framework by obtaining end-to-end resource estimates for different compilers, each tailored to the functionalities of specific hardware modalities: connectivity, clock speed, and noise model. As part of this framework, we introduce a transversal active volume (t-AV) compilation architecture designed for the efficient execution of fault-tolerant operations in platforms supporting long-range logical connectivity. We benchmark the framework for Hamiltonian simulation of the 2D Fermi Hubbard model as well as for eigenenergy estimation of a small molecule (trimethylenemethane) as a candidate for early fault-tolerant demonstration of quantum chemistry. For the latter, we show that end-to-end quantum simulations can be achieved with $\sim10^4$ physical qubits and runtimes ranging from $10^2$ ms (photonics, superconducting) to $10^5$ ms (neutral atoms).
Read the full abstract
Emerging Technologies
Sparse Walsh-Pauli Correlation Encoding for MaxCut: A Quantum-Inspired Formulation with Quantum-Hardware Validation
We introduce a sparse Walsh-Pauli correlation encoding for MaxCut, in which graph variables are represented by diagonal Pauli-string expectation values computed classically as normalized XOR autocorrelations of Walsh coefficients. The complete Walsh basis can reproduce the diagonal correlators of arbitrary quantum states, while a structured sparse support reduces the number of trainable coefficients to $m+1$ for $m$ graph variables. We evaluate the method on $1{,}400$ held-out random graphs with up to $7{,}000$ vertices, obtaining empirical approximation ratios near $0.98$ relative to MQLib heuristic references. On Gset instances G14, G23, G60, and G64, the sparse model achieves mean refined approximation ratios of $0.9828$, $0.9926$, $0.9781$, and $0.9369$, respectively, with $80.4\%$-$89.3\%$ fewer parameters than the complete Walsh representation. The trained coefficients are transferred without retraining to quantum circuits. Statevector simulations of the four Gset instances achieve fidelities above $0.99994$ and preserve all correlator signs. Separately, a seven-qubit experiment on a $15$-vertex graph executed on IBM quantum hardware achieves a correlator-sign agreement of $0.927\pm0.049$ relative to the corresponding noiseless circuit. These results demonstrate the potential of sparse Walsh-PCE for large-instance quantum-inspired MaxCut optimization while validating the transfer of classically trained parameters to quantum hardware.
Read the full abstract
Analysis of PDEs
Planar Gross–Pitaevskii traveling waves at every subsonic speed
For every subsonic speed $c\in(0,\sqrt2)$, we prove the existence of a finite-energy traveling wave for the planar Gross–Pitaevskii equation. This resolves the longstanding problem of the existence of prescribed-speed traveling-wave solutions in two dimensions, explicitly stated as open by Mariş (Ann. of Math., 2013) and Bellazzini and Ruiz (Amer. J. Math., 2023). The proof relies essentially on the energy estimate \begin{equation*} E(\psi)\le C_J\bigl(I_c(\psi)+\ind(\psi)\bigr), \qquad c\in J, \end{equation*} where $E$ is the energy, $I_c$ the action at speed $c$, $\ind$ the real Morse index, $J$ is any compact interval contained in $(0,\sqrt2)$, and $C_J$ is a positive constant depending only on $J$. We also prove finite-bubble compactness, including splitting of the energy, action, potential energy, and momentum, and attainment of the action among nonconstant waves of Morse index at most one.
Read the full abstract
Robotics
Visible-Reachable Workspace for Perception-Aware Humanoid Design
Workspace analysis measures where a robot can place its end effector. For visually guided manipulation, reachability alone is insufficient: a kinematically reachable target may not be visible in the specific pose required to reach it. The robot must then redirect its sensing or move its body to acquire a view, turning a perception limitation into additional motion. Existing humanoids largely inherit this limitation when copying human form factors. We introduce the visible-reachable workspace (VRW), a design-stage measure that conditions visibility on feasible reaching configurations and extends it to concurrent visibility of spatially separated work regions. We apply VRW by building a 31-DoF humanoid with independently actuated RGB-D cameras. On the same robot, camera articulation increases visible-reachable coverage from 38% to 97%. With actuated camera layouts, a second camera raises pairwise coverage from 0.45 to 0.95, while a third changes it only to 0.97. In a controlled two-target reach-and-grasp benchmark, our dual-actuated design reduces mean completion time by 17% and mechanical energy by 19% relative to the same robot with its cameras fixed. Hardware experiments demonstrate simultaneous observation and manipulation of front/back and left/right target pairs without torso reorientation. The results suggest that reachability becomes a more informative design quantity for perception-driven humanoid manipulation when it is evaluated together with the sensing configurations that make the reachable space observable. We will open-source all software and the humanoid hardware design. Our website is https://generalroboticslab.com/DukeHumanoidv2
Read the full abstract
High Energy Astrophysical Phenomena
Resolving structure within the iron line profile of Serpens X-1 with XRISM and NuSTAR
We present the first simultaneous XRISM and NuSTAR observations of the neutron star low mass X-ray binary Serpens X-1. We perform spectral modeling on the joint observations of XRISM/Resolve, XRISM/Xtend, and NuSTAR, testing Comptonization and double thermal continuum model prescriptions. We find that a hybrid double thermal model adequately describes the underlying continuum of the source while the Comptonization models predict largely unphysical parameter values. We perform reflection modeling and confirm with tight constraints an inclination of 5^{\circ} \pm 1^{\circ} and inner disk radius of 6.6 \pm 0.6 Rg for the source. We show that the spectral resolution of XRISM/Resolve allows for the determination of a unique inner disk radius, which has the potential to constrain neutron star spin in other systems. We discuss discrepancies in spectral shape between XRISM/Resolve and NuSTAR above \sim 8 keV. We also present lightcurve and spectral analyses of 8 Type-I X-ray bursts that occurred during the observations.
Read the full abstract
Combinatorics
On the classification of regular graphs with positive Lin-Lu-Yau curvature
We prove several new structural and classification results about $d$-regular graphs with positive Lin–Lu–Yau (LLY) curvature. We show that any positively curved $d$-regular graph has diameter at most $2d-2$, which improves the previously best known diameter bound obtained from the Bonnet-Myers-type theorem for positively curved graphs. We further show that every positively curved $d$-regular graph with $d\ge 3$ is $3$-connected. We classify all positively curved $3$-regular graphs, as well as all positively curved regular planar graphs.
Read the full abstract
High Energy Physics (Phenomenology)
Relativistic spatial distributions of transverse angular momentum
In our previous work [C. Lorcé et al., Phys. Lett. B 868 (2025) 139792], we investigated the 2D spatial distributions of transverse total angular momentum, including orbital angular momentum and intrinsic spin, relative to the canonical center (or center of spin). In the present work, we extend this analysis in two directions. First, we study the corresponding transverse boost distributions relative to the canonical center. Second, since the definition of generalized angular momentum density depends crucially on the choice of pivot, we analyze how the spatial distributions of transverse total angular momentum and boost are modified when they are defined relative to different relativistic centers, viz.~the relativistic centers of mass, energy, and spin. Considering spin-1/2 targets, we derive the corresponding 2D spatial distributions in the transverse plane, and further investigate how the spatial patterns evolve under longitudinal Lorentz boosts. Additionally, we provide the corresponding light-front distributions in the transverse plane, establishing a clear connection between the instant-form and light-front descriptions of transverse angular momentum and boost.
Read the full abstract
Machine Learning
The BatchNorm Illusion: Diagnosing Normalization Artifacts in Machine Unlearning Evaluation
Approximate machine unlearning aims to remove the influence of specific training data from a trained model without retraining from scratch. We identify a previously undocumented confound in how unlearning is evaluated on BatchNorm-based architectures: a single forward pass over retain data, an operation that modifies no weight, can deterministically rewrite the model's normalization state and reverse the apparent surface-metric forgetting. We formalize this operation as a weight-preserving fixed-point operator and prove that any pre-versus-post gap it induces is provably attributable to BN running statistics rather than to any modification the unlearning method made to the weights. This attribution claim cleanly separates measurement failure (BN artifact) from encoder failure (residual weight-encoded information, recently documented in concurrent work), and the same operator framework yields a unique decomposition of linear-probe elevation into BN-measurement-bias and encoder-geometry components. Empirically, the artifact reverses headline forget accuracy by up to 78 pp across nine evaluated methods on standard benchmarks; an attacker with as few as 10 unlabeled images recovers most of the masked accuracy; and a strict GroupNorm control reduces the artifact to zero across all methods. The tested membership-inference attacks change little under recalibration, locating the observed evaluation failure in forget accuracy and linear probing.
Read the full abstract
High Energy Physics (Experiment)
NOvA Dual-Baseline Search for Active-to-Sterile Neutrino Oscillations using Neutrino- and Antineutrino-Enriched Samples
We report a search for neutrino oscillations to sterile neutrinos in the NOvA detectors under a model with three active and one sterile neutrinos. This search simultaneously fits data in the two NOvA detectors and is the first from NOvA to use both neutrino- and antineutrino-mode beams, with exposures of $26.61\times10^{20}$ and $12.50\times10^{20}$ protons on target, respectively. There is no evidence for sterile neutrinos in the data and we are able to exclude regions of parameter space that were allowed by previous experiments, including most of the allowed region reported by IceCube.
Read the full abstract
Sound
From Scores to Evidence: Auditable Decisions Can Improve Speech Deepfake Detection
Speech deepfake detectors usually emit one score per utterance, but a borderline score does not reveal why two examples differ during retrospective error analysis. We ask whether a final score can be calibrated from component fields while keeping those fields visible for inspection. We build a decision record with a passive detector score and a score from a probe applied to a marked copy. It also includes retrieval support held out of the evaluated family, a margin from a support-set profile, and raw neighbor closeness. A cross-fit calibrator combines these fields and two differences between raw scores into one final score. On matched ASVspoof development data, the calibrated record reduces equal error rate (EER) by 3.48 percentage points relative to the fixed retrieval-augmented rule. It reaches 8.43% EER, whereas a passive WavLM baseline reaches 6.71% on the same subset. The record is therefore not the strongest detector in this comparison. Its value is to retain inspectable component fields while producing one scalar score for retrospective diagnosis.
Read the full abstract
Emerging Technologies
OTTER - Two Transistor - One RRAM Architecture for Reliable In-Memory-Computing in 28 nm CMOS Technology
This work presents OTTER, a 28 nm CMOS platform co-integrated with TaOx-based valence-change mechanism (VCM) RRAM, demonstrating a two-transistor-one-memristive-device (2T1R) architecture for reliable in-memory computing. The 2T1R cell combines a low-drive-current (LD) transistor and a high-drive-current (HD) transistor in parallel, providing dedicated bias paths for SET programming and RESET operation, respectively. Through systematic experimental and simulated comparison of various transistor-pairing configurations using the physical compact model JART VCM Rth, design guidelines for transistor sizing are derived, establishing the minimum RESET transistor W/L required for complete RESET as a function of the SET current compliance. The 2T1R cell is further characterized under pulse-based programming, demonstrating multilevel analog conductance tuning with narrow, well separated conductance states across six programmable levels. An analog content-addressable memory (aCAM) design based on the same 2T1R cell is additionally analyzed at the circuit level, evaluating trade-offs between top- and bottom-connected RRAM comparator configurations. A hardware implementation of compute-in-memory (CIM) multiply-and-accumulate (MAC) operations is further demonstrated on a 15 x 15 2T1R crossbar array.
Read the full abstract
Combinatorics
On the Stability of the Independence Number in Random Distance Graphs
We consider a random subgraph $G_p(n,r,<s)$ of the complete distance graph $G(n,r,<s)$ whose vertices are the $r$-element subsets of the set $\{1,\dots,n\}$ and whose edges join pairs of subsets that intersect in fewer than $s$ elements; each edge survives independently of the others with probability $p$. The independence number of the graph $G(n,r,<s)$ equals $C_{n-s}^{r-s}$ – this is the classical Erdos-Ko-Rado theorem. We prove that, for $r=r(n)\to\infty$, $s=s(n)\to\infty$, $s=o(r)$, $r^2=o(n)$ and $p\ge 16\,sr^2\ln(n/r)/n$, with probability tending to 1 the independence number of the random graph $G_p(n,r,<s)$ also equals $C_{n-s}^{r-s}$, i.e., the Erdos-Ko-Rado result is stable under random sparsification of the graph. Thereby, in the range of parameters $s\to\infty$, $s=o(r)$, a recent result of Raigorodskii and Karas is strengthened: the lower bound on the probability $p$ that guarantees stability is lowered by a factor of about $r/s$.
Read the full abstract
Computation and Language
Evolution of Multimodal Question Answering: From Modality-Adaptive Extraction to Unified Language Representation
The rapid growth of multimodal data has intensified the need for question answering (QA) systems capable of reasoning across heterogeneous sources such as text, tables, and images. In this paper, we present a comprehensive methodological comparison of three influential frameworks, namely Multimodal Adaptive Extraction (MAE), Solar, and UniMMQA, tracing the evolution of multimodal question answering from modality-adaptive pipelines to fully unified architectures. We examine how each approach models cross-modal interactions, transforms heterogeneous inputs, and performs reasoning, highlighting key design differences in modality representation, reasoning, and answer generation. Our analysis demonstrates a clear shift from explicit modality-specific processing toward unified text-centric formulations enabled by pre-trained language models (PLMs). Empirical comparisons across benchmark datasets show that this transition leads to substantial improvements in both Exact Match (EM) and F1-Scores, with UniMMQA achieving the most consistent and scalable performance. Despite these advances, we identify persistent challenges, including information loss during modality transformation, error propagation in multi-stage pipelines, and limitations in capturing fine-grained cross-modal dependencies. Overall, this study provides a deeper understanding of current design trends and offers insights into the future direction of unified multimodal reasoning systems.
Read the full abstract
High Energy Astrophysical Phenomena
Strong Polarization Signatures from Magnetically Stabilized Luminous Thin Accretion Disks
We utilize the Monte Carlo radiation transport capabilities of Cosmos++ to explore the polarization and Faraday rotation of radiation emitted from a set of general relativistic radiation magnetohydrodynamic simulations of magnetically stabilized, thin, black hole accretion disks. The guiding question is whether or not the Faraday rotation depolarizes the radiation to such a degree as to be inconsistent with the relatively high polarization measurements coming from the Imaging X-ray Polarimetry Explorer (IXPE). After first confirming that our code reproduces expected polarization results for electron clouds, planar scattering atmospheres, and AGN disks, we demonstrate the polarization and Faraday rotation capabilities using a Novikov-Thorne disk threaded with a purely toroidal magnetic field. We then analyze temporal snapshots from three different simulations of luminous, thin accretion disks threaded with different magnetic field configurations. We find that the effects of Faraday rotation are generally modest over the energy range of interest, since the strongest fields present in these simulations ($\gtrsim 10^8$ G) are mostly hidden beneath the photosphere. We can easily produce polarization degrees $\ge 4$% for sources seen mostly edge-on. As one example of a relevant target, we compare our results to an IXPE observation of Cyg X-1, finding that we can match the polarization degree quite well, though our polarization angle is rotated approximately $90^\circ$ with respect to the observed data. We speculate on a few possible explanations for this.
Read the full abstract
Astrophysics of Galaxies
Turbulence Cascade in Cygnus X Revealed by Multi-point VDF Method
Turbulence plays a crucial role in regulating star formation activities within molecular clouds, yet few methods can directly reveal its properties and underlying processes. We use molecular line data from the Nobeyama 45m Cygnus X CO Survey to study the turbulence properties and their relationships with star-forming activities and/or other non-thermal motions. In this work, we apply the multi-point velocity dispersion function (VDF), rather than direct linewidth measurements, to investigate the non-thermal properties of molecular cloud motions. We filter out the large-scale ordered structure and isolate a relatively small-scale turbulence component. Through the Friends In Velocity (FIVe) algorithm, we identify 10 substructures of the clouds and derive the turbulent properties of each cloud using the VDF method. We find that both the cloud-complex regions and the 10 velocity substructures exhibit turbulence correlation lengths of $\sim 2$–5 pc. This plateau scale suggests a parsec-scale turbulence correlation or driving scale in Cygnus X. Below this scale, the rising VDFs trace the velocity scaling of the turbulent cascade, whereas larger-scale VDF variations likely reflect cloud-scale motions. The comparison between cloud complexes and substructures further suggests that, in observational data, the VDF may constrain the turbulence correlation scale more robustly than the turbulence velocity dispersion.
Read the full abstract
High Energy Physics (Phenomenology)
ALP-mediated inelastic dark matter and the LUX-ZEPLIN high-recoil candidate event LZ230616
The LUX-ZEPLIN (LZ) Collaboration has reported a high-energy candidate event LZ230616 with a reconstructed nuclear recoil energy $E_R=248\pm23_{\rm stat}\pm23_{\rm sys}~{\rm keV}$. We investigate a possible interpretation in terms of inelastic scattering between two Majorana dark matter states mediated by an axionlike particle coupled to gluons. The positive mass splitting suppresses low-energy recoils, while the momentum dependence of the interaction reshapes the high-energy spectrum. We treat the dark-sector and gluonic couplings independently and retain the momentum-dependent nucleon form factors and xenon nuclear responses. Using an approximate single-event likelihood, we find that, for $m_a=0.3~{\rm GeV}$, a narrow spectrum near the candidate energy arises at $m_\chi\simeq0.35~{\rm TeV}$ and $\delta\simeq330~{\rm keV}$, although this configuration requires a large coupling product and is highly sensitive to the Galactic halo speed cutoff. Our analysis establishes the kinematic and coupling requirements for subsequent tests using the thermal relic abundance and laboratory constraints on the mediator.
Read the full abstract
Operator Algebras
The hyperfinite II_1 factor is not quasidiagonal
We construct an MF C*-algebra A such that the spatial tensor product of A and the hyperfinite II_1 factor R contains a proper isometry. Consequently, stable finiteness of C*-algebras is not stable under tensor product and the hyperfinite II_1 factor is not a quasidiagonal C*-algebra. The C*-algebra A is non-simple and has only non-faithful tracial states. This result was obtained using OpenAI's Chat GPT Pro 6.0.
Read the full abstract
Combinatorics
Four triangle-free intrinsically knotted graphs with 22 edges
An intrinsically knotted graph is one for which every spatial embedding contains a nontrivially knotted cycle. Classifying such graphs is a central problem in spatial graph theory. It is known that every intrinsically knotted graph has at least 21 edges, and the case of 21 edges has been completely resolved. For 22 edges, however, the classification remains incomplete. In particular, exactly eight triangle-free examples with a vertex of degree at least 5 are known, leaving only the case in which all vertices have degree 3 or 4. In this paper, we introduce a method for detecting intrinsic knottedness based on induced subgraphs obtained by deleting pairs of vertices. Using this method, we classify all triangle-free intrinsically knotted graphs with 22 edges having eight vertices of degree~4 and four of degree~3. We prove that there are exactly four: Cousins 43, 105, and 109 in the $E_9\!+\!e$ family and the graph $H_{12}\! +\! e$ in the $H_9\!+\!e$ family.