Artificial Intelligence
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Thursday 3 September
Multi-agent AI science ecosystems rely on agents possessing tools that allow them to communicate, coordinate, and build on each other's work. Yet this shared infrastructure can also introduce vulnerabilities by creating a substrate for the contagious spread of unintended and undesirable behaviors. We report a case study on a research collective of 100 autonomous LLM agents tasked with proving formal mathematical conjectures. Within the swarm, cheating spontaneously emerged and was later challenged by whistleblowers - both without any external intervention. When a single agent discovered an exploit in the evaluation system, it propagated across the collective via a shared knowledge library and later through peer-to-peer messages. Despite early reluctance, a cohort of agents adopted the exploit in response to competitive pressure. A separate group of agents produced an emergent counter-response: auditing fraudulent proofs, alerting peers across broadcast and private channels, staging boycotts, lodging formal complaints, and proposing validation patches. In recent incidents, agent swarms coordinated covertly through improvised side-channels (Dalton and Wallace, 2026; Greenblatt et al., 2026). Our setting differs: the same transparent channels that carried the exploit also gave non-cheating agents the visibility they needed to detect fraud, organize resistance, and enforce norms. We cast the problem of managing the agents' shared infrastructure as the knowledge commons governance problem (Ostrom, 1990). To protect the commons from exploits, we propose to adopt institutional mechanisms, such as graduated sanctioning and collective-choice rules, to support decentralized self-governance in autonomous swarms.
Soft Condensed Matter
Reciprocity can halve what a mechanical network can learn
A passive, reciprocal, linear network driven by forces or currents and read at points it drives cannot be trained below an error floor computed from the target beforehand. The usual parameter count for such networks does not charge for Maxwell-Betti reciprocity. We prove that when such an elastic, resistor or flow network has a symmetric response operator and p driven degrees of freedom are also read, its reachable response blocks lie in a subspace of codimension p(p-1)/2, whatever its size and topology, and the floor is the distance to it. On targets with a reachable symmetric part, a second-order optimiser ends within 0.1% of the floor in 142 of 144 simulated runs, and a contrastive rule with a bond-local update direction, within its step budget, in 22 of 24. Reciprocity charges a list of single drive-read tasks only for the pairs it instruments in both directions, and a rank test on the passive network says in advance whether, and on which bonds, odd couplings locally restore the lost directions, at a cost in non-reciprocity that the target bounds from below. Under an imposed-displacement drive, the one most physical-learning hardware uses, reciprocity survives as a weaker law, an exact balance between forward and reverse transmissions once each is weighted by the driving-point compliance at its input, and an inequality on their product, under which no symmetric positive-definite chain meets both targets of a published robotic metamaterial. All nineteen published layouts we tabulated pay nothing; a force-driven layout that asks one pair to respond differently in its two directions will.
Read the full abstract
Distributed Computing
Para-Pipe: Exploiting Hierarchical Operator Parallelism of ML Computational Graphs on SoCs
As edge-based deep learning applications become more complex, optimizing performance on heterogeneous System-on-Chips (SoCs) presents unique challenges. Traditional pipelining techniques distributing the computation across different on-chip processing units, while effective for throughput, do not address the latency demands posed by modern neural networks with complex interdependencies and extensive operator parallelism. There is a potential in leveraging operator parallelism to enable concurrent execution across multiple processing units, thereby reducing inference latency. However, prioritizing pipelining or parallel execution often necessitates a compromise, where optimizing one performance metric adversely impacts the other. This paper introduces Para-Pipe, a hierarchical mapping framework that integrates intra- and inter-stage operator parallelism within a pipelined architecture. Para-Pipe navigates the trade-off between throughput and latency by selectively fine-tuning parallelism levels within and across pipeline stages. This strategy can significantly reduce inter-processor communication overhead, significantly improving energy efficiency. Our evaluation demonstrates that Para-Pipe generates multiple Pareto-optimal configurations, achieving a balance between throughput and latency on an Amlogic SoC equipped with ARM big.LITTLE CPUs and GPU, as well as the Black Sesame Technology SoC featuring a deep learning accelerator and two DSPs. More importantly, throughput-optimized configurations under Para-Pipe on Amlogic SoC show an average energy efficiency improvement of 11.0% over purely pipelined strategies and 23.3% relative to non-pipelined parallel execution.
Read the full abstract
Software Engineering
SWE-Gate: Passing Functional Tests Is Not Enough for Software Engineering Agents
Repository-level software engineering benchmarks have significantly advanced the evaluation of coding agents, but existing benchmarks primarily measure whether generated patches pass functional tests and overlook review-derived acceptance constraints (review constraints) that often influence whether a patch is acceptable in real-world software development. We introduce SWE-Gate, a repository-level benchmark for software engineering agents that explicitly evaluates review constraint compliance alongside functional correctness. SWE-Gate derives review constraints from real pull request review comments and synthesizes repository-level repair instances around these constraints. Each instance provides separate functional and constraint tests, together with non-compliant and gold patches, enabling explicit separation between issue resolution capability and review constraint compliance. We construct SWE-Gate with 303 repository-level repair instances spanning 75 open-source Python repositories across diverse software domains. Experiments with four LLM backends spanning different capability levels under a common coding-agent scaffold reveal a substantial gap between functional success and success under the complete repair specification: among 644 repairs that pass the functional tests, 221 fail to satisfy the provided review constraints. These findings show that functional-only evaluation overestimates agents' ability to satisfy the full requirements of repository-level repair tasks. The replication package including code, data, and experimental results is available at https://github.com/DeepSoftwareAnalytics/SWE-Gate.
Read the full abstract
Artificial Intelligence
From Deceptive Outputs to Deceptive Mechanisms: A Causal Framework for Language-Model Deception Research
Research and news coverage of language-model deception increasingly attributes human-like mental-state concepts to language models. Such claims can blur the distinction between behavior that looks deceptive and a mechanism that is actually deceptive. We introduce a causal taxonomy separating prior commitment from retrospective report, model preference from realized output, false preference from sensitivity to the utility of misleading a recipient, and deceptive behavior from the provenance of the objective or strategy producing it. We test these distinctions in two open-weight model families. Across controlled guessing-game and stock-trading experiments, we find that deceptive-looking behavior can arise without the corresponding proposed mechanism, while other interventions provide direct evidence that recipient information state can causally affect deceptive preference. These results show that deceptive behavior can provide evidence for a deceptive mechanism. But even evidence for such a mechanism does not establish model agency in the deception.
Read the full abstract
Quantum Physics
Parameterised graph theory for tensor networks: entanglement rerouting, structural simplification, and agnostic tomography
Parameterised graph theory studies how the complexity of graph-theoretic problems depends on structural parameters of the input graph. This perspective has proved useful in analysing tensor-network simulation (Markov and Shi, 2008). Its implications for tensor-network representations and tomography are less well understood. In particular, which graph parameters determine whether a tensor-network state (TNS) admits a tractable matrix product state (MPS) or tree tensor network (TTN) representation, and which control the complexity of learning the state? We address these questions using parameterised graph theory. First, we show that cutwidth and tree-cutwidth bound the bond dimension overhead required to represent a TNS as an MPS or TTN. In the TTN case, tree-cutwidth also bounds the local dimension of the grouped subsystems. The proofs are based on entanglement rerouting, a tensor-network analogue of rerouting information in a classical network. Second, we derive graph-dependent upper bounds on the sample and computational complexity of realisable TNS tomography, with exponents that depend on cutwidth, tree-cutwidth, and a new graph parameter, learning complexity, which we bound in terms of degree and treewidth. We obtain these results by extending the disentangling MPS learner of (Cramer et al., 2010), as analysed further in (Bakshi et al., 2025; Lin et al., 2025), to TTNs and to tensor networks on arbitrary known graphs. Finally, we extend the framework beyond the realisable setting. For an arbitrary input state, our agnostic learner outputs a pure state whose fidelity is within additive error $\epsilon$ of the optimum over tensor-network states on the given graph with a given bond dimension, with explicit graph-dependent bounds on sample and computational complexity.
Read the full abstract
Superconductivity
A Superconducting Peierls Instability
The Peierls instability is a foundational mechanism in condensed matter physics, showing how the electron–phonon interaction can transform a simple metal into an insulating state with a new lattice periodicity. In a one-dimensional (1D) metal the Peierls instability follows from the enhanced electronic response at wavevector $Q=2k_F$ (connecting the Fermi points) producing a Kohn anomaly: a softening of the phonon mode at the same wavevector. Condensation of this mode then generates the tell-tale Peierls periodic lattice distortion and charge-density wave (CDW) that gaps the electronic spectrum. Here we show an analogous instability at the edge of a two-dimensional (2D) superconductor, where a dispersing Andreev bound state (ABS) hosts Bogoliubov Fermi points at $\pm k_c$. The enhanced quasiparticle response at the connecting wavevector $Q=2k_c$ couples directly to pairing-fluctuations and produces a superconducting Kohn anomaly: a softening of a pairing mode at the same wavevector. Condensation of this mode then generates an edge pair-density wave (PDW) that gaps the ABS. We refer to this as a superconducting Peierls instability and identify the boundary quasiparticle structure that enables it. As a concrete realization, we consider a square-lattice extended Hubbard model whose mixed-symmetry $s+d+ip$ state hosts a dispersing ABS with zero-energy crossings at finite edge momenta. Using self-consistent Bogoliubov–de Gennes (BdG) calculations we show that the order parameter develops an edge PDW, with the wavevector set by the superconducting Kohn anomaly, that gaps the ABS crossings. Our results identify a novel mechanism for spontaneous translation-symmetry breaking in a superconductor. As this superconducting Peierls mechanism does not require an extensive zero-energy flat band or topologically protected edge states, it may apply broadly to unconventional superconductors.
Read the full abstract
High Energy Physics (Phenomenology)
TeV Higgsino Interpretation of the LZ High-Recoil Event with Intermediate-Scale Electroweak Gauginos
The high-recoil event reported by LUX–ZEPLIN motivates a TeV Higgsino with a sub-MeV neutral-state splitting. Away from cancellations, such a small gap in the MSSM points to electroweak-gaugino masses of order $10^7\,\mathrm{GeV}$, far above the Higgsino mass. We show that the non-universal boundary condition $M_1^G/M_2^G=-3/5$ at the $\mathrm{SU}(5)$ GUT scale cancels the leading bino and wino contributions and is preserved by homogeneous one-loop evolution down to the Higgsino scale. The remaining finite-gaugino terms admit a controlled tree-level solution with a wino mass near $120\TeV$ for a $350\keV$ gap, a $1.091\TeV$ Higgsino and $\tan\beta=10$. We identify mixed $\mathrm{SU}(5)$ representations that realize the required gaugino ratio and examine the radiative sensitivity and spectrum consistency of this construction. Because the inelastic $Z$ coupling remains essentially unsuppressed, a full-density thermal Higgsino is still subject to the published solar-capture bounds under their astrophysical and transport assumptions.
Read the full abstract
Statistical Mechanics
Thermodynamic Concentration Inequalities: Controlling Uncertainty in Finite-Time and Small-Sample Thermodynamic Inference
We derive nonasymptotic upper bounds on the probability that a generalized current of a geometrically ergodic diffusion observed for any amount of time, or its sample mean over any arbitrary sample size, deviates from the stationary mean by more than any given amount. The concentration-of-measure behavior of generalized currents is universally governed by the relaxation time of the underlying dynamics, the locally observed dissipation rate, and the intrinsic local fluctuations of the observable. We uncover stark qualitative and quantitative differences in fluctuations in and out of thermodynamic equilibrium. We further obtain refined inverse thermodynamic uncertainty relations, bounding the variance of generalized currents from above. We construct nonasymptotic confidence intervals for controlling uncertainty in thermodynamic inference from small data, i.e., from short trajectories and small samples, and provide the first quantitative answer to when a trajectory is sufficiently long and a sample is sufficiently large. As an illustration, we apply our results to currents observed on a two-dimensional Ornstein-Uhlenbeck process in and out of equilibrium and show how they can be used to rigorously detect broken detailed balance.
Read the full abstract
Discrete Mathematics
Minimizing the makespan in job shop scheduling under conflict graph constraints
We study the job shop scheduling problem with a conflict graph, in which adjacent jobs cannot be processed simultaneously on different machines, with the objective of minimizing the makespan. We establish a polynomial equivalence with a resource-constrained job shop problem having unit-capacity resources and binary resource requirements The problem is strongly NP-hard even on two machines with unit-time operations. For this setting, we derive an optimal-makespan formula and a linear-time algorithm when the conflict graph is a cluster graph, given the clique partition. We also develop a polynomial-time $4/3$-approximation algorithm for arbitrary conflict graphs on two machines when each job has exactly two unit-time operations. For the general problem, we propose four mixed-integer linear formulations, lower bounds, and a genetic algorithm with active, non-delay, and hybrid decoding procedures. Computational experiments on instances derived from the Lawrence and Taillard benchmarks and on randomly generated generalized job shop instances evaluate the proposed methods and show that the genetic algorithm generally improves on constructive heuristics.
Read the full abstract
Quantum Physics
Vanilla Exact Synthesis of CNOT Circuits is NP-hard
Exact CNOT synthesis seeks a minimum-size circuit implementing a given invertible binary linear transformation. We study its vanilla formulation: synthesis from the identity on fixed labeled qubits, without ancillas and with all-to-all connectivity. We prove that this problem is NP-hard and that its bounded decision version is NP-complete. Hardness persists even for unitriangular targets that are low-rank perturbations of the identity and under near-linear gate budgets. Our proof gives a polynomial-time reduction from the NP-complete Hamiltonian path problem on grid graphs. A unary hypercube embedding translates graph vertices into circuit parities and edge traversals into CNOT updates. The central challenge is that Hamiltonian paths require intermediate vertex visits, whereas exact synthesis constrains only the final transformation. We bridge this gap with a replicated recorder qubits construction that encodes the required parities in the final outputs, together with a tight gate budget that forces every sufficiently short implementation to trace a Hamiltonian path. Crucially, this path structure is enforced by the construction rather than imposed as a restriction on the circuit. The result directly implies NP-hardness for shortest word problem and Cayley-graph distance computation over $\mathrm{GL}(n,2)$ with elementary transvections as generators, as well as fixed-width sequential XOR program minimization. For CNOT-minimal exact phase polynomial synthesis, it yields NP-hardness with a zero phase polynomial; extending the reduction to grid-graph Hamiltonian cycle problem establishes hardness even when the final linear transformation is the identity. These complementary results show that the linear and phase components each independently suffice for NP-hardness of exact phase polynomial synthesis.
Read the full abstract
Cryptography and Security
SENTINEL-RL: Offloading Topological Reasoning from LLM Agents in the Security Operations Center
Large language model (LLM) agents are increasingly proposed as autonomous SOC analysts, but two limitations make them unreliable at enterprise scale: a finite context window cannot hold a multi-thousand-host authentication graph, and free-form generation offers no guarantee that a recommended containment action is consistent with the topology it operates on. We present Sentinel-RL, an agentic-SOC architecture that decouples topological reasoning from semantic reasoning: a heterogeneous graph attention encoder summarizes the live authentication subgraph into a fixed-dimensional state, a Proximal Policy Optimization (PPO) policy maps this state to a constrained set of investigative actions, and an LLM agent loop is restricted to consuming the policy's recommendations and producing analyst-readable narratives gated by a critic. We instantiate the system on the LANL Comprehensive, Multi-Source Cyber-Security Events dataset and the Indiana University Quartz HPC cluster, reporting four results: (i) a two-phase CREATE ingestion pattern loads a 24M-edge authentication subgraph into Neo4j in 14.2 minutes on a single 32-core node, roughly 24x faster than the canonical MERGE-based pipeline; (ii) a sliding-window alert engine reliably trips a 25-event/10-second threshold in <=2.5 s across 50 trials; (iii) PPO training over 200 iterations converges to a mean episodic return of 8.74+/-0.31, with held-out precision of 0.91 and recall of 0.87 on labeled red-team events; and (iv) the integrated containment loop completes a full detect-investigate-recommend-human-approve cycle in a median of 6.3 s. We contribute a reusable engineering pattern (the hot-node deadlock workaround), a portable HPC deployment pattern (anchor-node co-location), and an enterprise-readiness analysis covering false-positive economics, reversibility guarantees, audit compliance, and the human-approval boundary.
Read the full abstract
Systems and Control
Formation Matrix and Energy-based Control of Multi-Agent Systems
This paper presents an energy-based controller for a multiagent robotic system designed to achieve and maintain a specific formation while moving on a plane and avoiding collisions between agents. The controller emulates a network of elementary spring-damper modules connecting pairs of agents. This network, with its de-energized states representing the desired formation, determines the system's dynamics, which is fully encapsulated by a bond graph model. The modeling is further enhanced through the introduction of a formation matrix, using a graph-theoretic approach, that describes both the distances and relative velocities among the agents of the arrangement. This matrix mathematically represents the interconnection and energy-exchange structure of the bond graph, allowing us to put it in correspondence with the control-by-interconnection CbI-scheme of the IDA-PBC theory, facilitating the solution of the formation control problem within the port-Hamiltonian system framework. Furthermore, the paper presents leader-following and position-based formation control systems based on the CbI scheme, including a stability analysis of the corresponding closed-loop systems. The theoretical findings are validated through numerical simulations across various scenarios.
Read the full abstract
Quantum Physics
Spurious quantum correlations
In his seminal paper, Bell [Physics Physique Fizika 1, 195 (1964)] considers the correlations that result from space-like separated measurements on a pair of entangled particles. He uses relativity theory to motivate the Bell causal structure, then shows the existence of quantum correlations that cannot be explained classically within this causal structure. Classical explanations of such quantum correlations are possible in alternative causal structures, for instance, those that allow superluminal causal influences, but, as shown in [New Journal of Physics 17 033002 (2015)], all such alternative explanations require fine tuning (causation without correlation). Here we discuss the existence of spurious quantum correlations — correlations that look quantum in one causal structure, but have a natural classical explanation in another. More precisely, there are causal structures that admit non-classical quantum correlations, but for which the same correlations have a classical explanation in another causal structure without fine tuning. The realisation in the other causal structure can also be achieved without breaking any natural constraints on the causal structure that follow from relativity theory. However, similarly to non-classical quantum correlations in the Bell causal structure, we find other causal structures with non-classical quantum correlations that do not have a classical causal explanation in any alternative causal structure without fine tuning.
Read the full abstract
Differential Geometry
A $dd^{\Phi}$-Lemma and Bott–Chern-type Cohomology for Spin(7)-Manifolds
We study the properties of the $dd^{\Phi}$-operator on $8$-dimensional Spin(7)-manifolds with torsion-free Spin(7)-structures $\Phi$. These operators were first introduced by Harvey and Lawson (An introduction to potential theory in calibrated geometry. Am.J. Math. 131.4 (2009), arXiv:0710.3920). We prove a Hodge decomposition theorem for the $dd^{\Phi}$-operator and obtain an analogue of the $\partial \bar{\partial}$-lemma in Kähler geometry. Using this, we define Bott–Chern-type cohomologies for Spin(7)-manifolds. We relate the Bott-Chern-type cohomology spaces to the moduli space of torsion-free Spin(7)-structures and calibrated geometry of Spin(7)-manifolds. These relations naturally give rise to the notion of Cayley-positive cones. In the course of proving the results, we state and prove various identities for the exterior derivative and its decompositions into irreducible Spin(7)-representations as well as identities for second order derivatives and Laplacians. The identities we prove are for any Spin(7)-structures and the specialized torsion-free ones are Spin(7)-analogoues of Kähler identities and Bryant–Harvey's identities in the $\mathrm{G}_2$-case (R. Bryant, Some remarks on $\mathrm{G}_2$-structures, Proceedings of the 11th and 12th Gökova geometry-topology conference, arXiv:math/0305124) and are results of independent interest.
Read the full abstract
Number Theory
The subconvexity problem for symmetric square $L$-functions in level aspect
In this paper, we address the subconvexity problem in level aspect for symmetric square $L$-functions for cuspidal automorphic representation of $\mathrm{GL}_2(\mathbb{Q})$ with a prescribed local ramification at prime $p$. To be more precise, let $\pi$ be a tempered cuspidal automorphic representation of conductor $q(\pi)=p^2$ with a non-quadratic central character of conductor $p$. We prove that if the corresponding local representation $\pi_p$ belongs to a suitable class of representations $\mathcal S$, then \[ L\left(\frac{1}{2},\,\mathrm{Sym}^2\pi\right)\ll_{\varepsilon, \pi_\infty} q(\mathrm{Sym}^2\pi)^{\frac{1}{4}-\frac{1}{168}+o(1)}, \] where implied constant depends polynomially on the spectral parameters of $\pi_\infty$. This is the first instance of level-aspect subconvex bound for $L$-functions of a $\mathrm{GL}_3(\mathbb Q)$ automorphic representation. Our approach is based on the delta-symbol method. Apart from some standard analytic number theoretic tools, Katz's theory of hypergeometric sums, and Deligne's proof of Weil-conjectures play an important role in the proof.
Read the full abstract
Nuclear Theory
B(E2) Serves as a Robust Signature of N = 32,34 Shell Evolution
Electric quadrupole transition probabilities $B(E2)$ serve as key probe of nuclear shell evolution, yet anomalous $B(E2)$ values in exotic nuclei complicate the identification of new magic numbers. In this letter, employing the configuration-interaction relativistic Hartree-Fock model, we demonstrate that effective charges are sensitive to orbital radii, and this orbital dependence is significantly amplified by the halo structure of valence nucleons. This mechanism is critical for reliably describing $E2$ transitions and understanding the unusual behavior of $B(E2)$ in exotic nuclei. Our calculations predict reduced $B(E2; 2^+_1 \rightarrow 0^+_1)$ values in $^{52,54}\text{Ca}$, signaling the emergence of subshell closures at $N=32$ and 34. Furthermore, the suppressed $B(E2; 7/2^{-}_{1} \rightarrow 11/2^{-}_{1})$ transition in $^{53}\text{Sc}$ underscores the robustness of the $N=32$ new magic number, whereas the enhanced transition strength in $^{55}\text{Sc}$ indicates the rapid erosion of the $N=34$ shell gap with the occupancy of the proton orbital $\pi1f_{7/2}$.
Read the full abstract
Analysis of PDEs
Smooth autonomous fast dynamo action on the three-torus
We construct a nonempty $C^k$-open family of smooth, autonomous, divergence-free velocity fields on $\mathbb{T}^3$ that generate fast dynamos. The proof proceeds by first establishing fast dynamo action for a large class of smooth, time-periodic velocity fields exhibiting the classical stretch–fold–shear mechanism. We then realize the associated period map within a smooth autonomous flow using carefully tuned return dynamics to a global Poincaré section. In both settings, the proof first isolates unstable distributional eigenmodes for the ideal dynamo operator on an anisotropic Banach space of distributions, and then shows that this spectral instability persists under the singular perturbation $\varepsilon \Delta$. The resulting estimates are uniform as $\varepsilon\to0$ and stable under $C^k$ perturbations of the velocity field, yielding an open set of smooth fast dynamo vector fields. This result resolves the Fast Dynamo Conjecture of Zeldovich and Sakharov, as recorded in Arnold's book of problems.
Read the full abstract
Computational Physics
Spectral and Eigenvector Crossovers in Random Mixed Graphs
We study the spectral and eigenvector properties of random mixed graphs, combining undirected and directed interactions, using random matrix theory (RMT). The network is represented by a Hermitian adjacency matrix, with undirected links as real entries and directed links as purely imaginary conjugate pairs, ensuring a real spectrum. Network density is controlled by the connection probability, while directionality sets the fraction of directed edges. We focus on the GOE-to-GUE crossover: at fixed connectivity, increasing directionality breaks time-reversal symmetry and drives spectral statistics from GOE to GUE. We show that this transition requires sufficient connectivity. In sparse networks, weak level repulsion produces Poisson statistics regardless of directionality. At fixed directionality, increasing connectivity drives a Poisson-to-GUE crossover; only above a sparsity threshold does the directionality-induced GOE-to-GUE transition emerge. In the dense regime, where the spectral density follows the Wigner semicircle law, the crossover is characterized using spacing distributions, spacing ratios, and spectral rigidity. In sparse networks, where unfolding is unreliable, spacing-ratio statistics provide an unfolding-free characterization. Eigenvector structure is examined through multifractal dimensions and component distributions, while Kullback–Leibler divergence confirms the robustness of the transitions. Applied to S&P 500 mixed graphs, the framework reveals a GOE-to-GUE crossover across four major market crashes. Denser crisis periods show sharper crossovers than sparser recovery periods. The results provide a unified picture of how connectivity and symmetry breaking govern spectral and eigenvector universality, while providing a transparent probe of changing financial-market organization.
Read the full abstract
Computer Vision
Persistent Identity Preservation in Generative Image Models: A Benchmark and Evaluation System
Generative image models can now produce high-quality images, follow complex instructions, and support precise edits, but they still struggle to preserve who or what is being depicted. When generating or editing images of a specific subject, identity may drift as the pose, expression, appearance, viewpoint, or surrounding scene changes. Existing subject-driven methods make fundamentally different choices about where identity is represented: through the input context (GPT-Image-2, NB2), as trainable subject-specific model parameters (LoRA), or as a persistent identity layer (PHOTA IDENTITY) reusable across generations and edits. We systematically benchmark these paradigms across subject-driven generation, editing, restoration, and multi-subject settings, with tasks designed to increasingly stress identity preservation. Our results show that identity preservation remains a distinct limitation of current generative foundation models: strong image quality and instruction following do not necessarily imply strong identity fidelity, and identity degradation becomes more pronounced under iterative edits, small subject scales, severe image degradation, and multi-subject composition. Persistent identity substantially reduces this degradation across generation, editing, and restoration, consistently improving identity preservation when applied to different foundation models while maintaining comparable instruction adherence and perceptual image quality. These results suggest that identity does not simply emerge from increasingly capable generative models, but can instead be represented as persistent subject knowledge that is composed independently with the underlying generative model.
Read the full abstract
Quantum Physics
Efficient Conversion of Optical to Mechanical States Close to the Single-Quantum Level
Coherent interfaces between optical photons and mechanical excitations provide a promising route towards phonon-state engineering and hybrid quantum information processing. Cavity optomechanical systems enable such interfaces via optomechanically induced transparency (OMIT), allowing coherent mapping between traveling optical fields and localized mechanical modes. However, previous implementations of OMIT conversion protocols were limited to classical input signals with large coherent state photon numbers due to room temperature operation and corresponding thermal mechanical noise. Here, we demonstrate efficient low-noise photon-phonon state transfer close to the single-quantum regime in an optomechanical crystal operated at Millikelvin temperatures. Using weak coherent optical input pulses at the few-photon level, we achieve a record-level photon-phonon conversion efficiency of $\eta=0.76$, a mechanical storage lifetime of $T_\mathrm{1}=7.3~\mu s$, and a tunable conversion bandwidth exceeding 4.5 MHz. Hanbury Brown-Twiss measurements of the retrieved signal demonstrate the coherent nature of the converted phononic state, evidencing low added thermal noise in the conversion process ($n_\mathrm{th}=9.0$). These results establish optomechanical crystals as efficient optical interfaces to GHz mechanical modes and provide a pathway toward deterministic single-quantum-level mechanical state preparation.
Read the full abstract
Strongly Correlated Electrons
Non-uniform quantum geometry stabilizes generalized Wigner crystals
Moiré materials host fractional Chern insulators and electron crystals in close proximity, but the mechanism selecting between them remains an open question. We address this competition in Chern bands with ideal but momentum-dependent quantum geometry – Aharonov-Casher bands. We present an ansatz wave function for generalized Wigner crystals and, by comparing its energy to that of the competing Laughlin-like state, map out the phase diagram at filling fraction $\nu=1/m$ as a function of the degree of geometric non-uniformity. Our work identifies quantum geometry-controlled zero point fluctuations of the charge density of the generalized Wigner crystal as the mechanism controlling its relative stability, implying a kind of quantum Lindemann criterion for the crystal-liquid phase boundary.
Read the full abstract
Artificial Intelligence
Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments
As terminal-based code agents become prevalent, agent trajectories have accumulated at scale, while realistic, executable environments remain scarce. However, environments are what agent post-training actually requires: each can be re-queried into many verifiable tasks and provides execution feedback, whereas a trajectory is a single frozen demonstration. Rather than generating environments from scratch, we observe that the tool-execution history in existing trajectories exposes the structure and contents of the environments in which they ran, making it possible to reconstruct those environments from the trajectories themselves. Thus, we introduce Terminal-Universe, a framework which turns each trajectory into a reusable environment and explores it for synthesizing new tasks and continued interactions. Specifically, Terminal-Universe replays the file operations recorded in a trajectory to restore each file before the agent modified it, yielding a partial workspace; a completion agent then supplies the missing files and dependencies. On this recovered workspace, we both reconstruct the original intent task and synthesize entirely new ones. Besides, we also scale the tasks along two complementary axes: breadth and depth. For breadth, we mine directional dependency relations between related environments and synthesize cross-workspace queries spanning multiple codebases, as developers routinely do in real-world development. For depth, we extend the initial single-turn query into a multi-round session that captures iterative user feedback and requirement refinement via a user agent. Applied to public terminal agent trajectories, Terminal-Universe produces 37.3k task-sufficient environments. Supervised fine-tuning of Qwen3.5-27B on this corpus improves single-round performance on Terminal-Bench 2.1 by 11.9 points and multi-round performance on EvoCode-Bench v2 MT@4 by 13.8 points.
Read the full abstract
Machine Learning
A Low-Cost, Open Platform for End-to-End Autonomous Driving on a Miniature Ackermann Vehicle
This paper presents a low-cost, open experimental platform for research in end-to-end autonomous driving with miniature Ackermann vehicles. The platform combines a physical vehicle, a printed urban track, data collection tools, trajectory registration, and a Webots digital twin, enabling controlled experiments that connect simulation-based autonomous-driving methods to real-world execution. As a first baseline, we implement command-conditioned behavior cloning, in which a neural policy receives an on-board camera image and a high-level navigation command and outputs steering and speed. The system is evaluated both on the physical vehicle and in simulation. In real closed-loop experiments, the learned policy follows lanes and executes commanded turns, reaching a mean cross-track error of 6.1 cm with respect to the reference route, close to the 4.7 cm observed in human demonstrations. In the digital twin, camera field of view has a strong effect on performance, reducing the mean cross-track error from 35.6 to 3.3 cm when widened from 58 to 120 degrees. Using the digital twin to generate synthetic driving data and a learned sim-to-real image translator to reduce the appearance gap, we further show that a higher-capacity policy trained on this synthetic data combined with real demonstrations is the only configuration that completes all four track routes in closed loop, whereas the compact baseline and the same network trained on real data alone complete fewer. These results establish the open platform as a practical testbed for sim-to-real studies and provide an initial command-conditioned imitation-learning baseline; we release it to support reproducible research.
Read the full abstract
Strongly Correlated Electrons
HyperDet Wavefunction: A Phase-Agnostic Ansatz for Strongly Correlated Systems
Competing phases of strongly correlated systems are often described by trial wavefunctions built on phase-specific assumptions. We propose the hyperdeterminant (HyperDet) wavefunction as a phase-agnostic ansatz for both bosonic and fermionic quantum many-body systems exhibiting spontaneous symmetry-breaking, fractionalization, and/or topological order. The HyperDet structure emerges by fusing auxiliary fermionic parton Slater determinants into physical orbitals through a learnable \emph{fusion tensor $\mathcal F$}. Optimized using variational Monte Carlo, a single HyperDet architecture achieves overlaps exceeding $99.9\%$ with exact-diagonalization ground states for nearly all sampled parameters across fractional Chern insulator (FCI) phases and their neighboring competing phases in both bosonic and fermionic models. Beyond its variational expressiveness, the optimized fusion tensor also provides rich interpretability. We show that, (i) the singular-value spectra of the bipartite fusion matrix can be used to track phase transitions; (ii) the reconstructed Bott indices reproduce the parton Chern numbers expected for the FCI states without parton-Hamiltonian input; and (iii) faithful lifts of physical translations to the parton degrees of freedom recover the expected translation fractionalization in the bosonic FCI without projective symmetry class assignment. These results establish HyperDet wavefunction as a promising framework for both variational ground-state searches and phase-diagram explorations across strongly correlated phases, and for providing theoretical insights by connecting the optimized wavefunction with parton microscopics and field-theory descriptions.
Read the full abstract
High Energy Astrophysical Phenomena
Low-angular-momentum accretion shocks can power weak-to-moderate X-ray flares from SgrA*
X-ray flares from Sgr~A* span broad ranges in duration, fluence, and luminosity, but their origin remains unsettled. We test whether standing shocks in low-angular-momentum magnetized accretion flows can provide a viable energy reservoir to account for these events. Using semi-analytic trans-magnetosonic shock solutions, we estimate the kinetic energy available in the downstream post-shock flow and compare it with the 25-year Chandra X-ray flare catalog. For each theoretical solution, we compute the required efficiency $\epsilon=E_{\rm data}/E_{\rm sh}$, where $E_{\rm data}$ is the observed radiated energy for each flare and $E_{\rm sh}$ is the available energy in the shocked flow. Notably, the weak flares require $\epsilon\sim10^{-3}$–$10^{-2}$, while moderate flares require a few percent. We perform the analyses for weakly and highly spinning cases, and the resulting weak-to-moderate flare energy budget remains unchanged. For the fiducial accretion rate and bolometric correction, the kinetic energy reservoir in standing shock is sufficient for the observed weak-to-moderate flare population, whereas the strongest events likely require higher efficiency which can be mediated via additional magnetic energy dissipation channel.
Read the full abstract
High Energy Physics (Phenomenology)
Inelastic Signatures of Electroweak Dark Matter
Minimal dark matter extended to include a Majorana and a Dirac multiplet coupled through the Higgs — the HC-MDM model — provides a compelling and predictive framework. We show that in certain limits of this model, the mass splitting and the inelastic interaction are representation-independent, while the thermal-relic masses are representation-dependent. We then show that this model can account for the high-energy recoil event recently reported by LUX-ZEPLIN (LZ) while also respecting thermal-relic, elastic-scattering, and solar-capture constraints. Generic inelastic DM models do not connect all these tests and generic Higgsino models do not survive them.
Read the full abstract
Functional Analysis
The Geometry of Real Anisotropic Bohnenblust–Hille Constants
We determine the growth scale of the optimal constants in the real anisotropic Bohnenblust–Hille inequality. For an exponent vector $\mathbf q^{(m)}$, write $C_{\mathbf q^{(m)}}^{(m)}$ for its optimal constant and $d_m$ for its diameter. These constants are superpolynomial precisely when $m d_m/\log m\to\infty$; throughout this regime, their logarithm has the sharp scale $m d_m$. If also $d_m\to0$, then $\log C_{\mathbf q^{(m)}}^{(m)}/(m d_m)$ lies asymptotically in the interval $\left[\frac{\log 2}{4},\frac{2-\log 2-\gamma}{4}\right]$, where $\gamma$ is the Euler–Mascheroni constant; this interval has width less than $10^{-2}$. We also solve the extremal problems at fixed diameter and fixed total deficit. The normalized reciprocal-deficit profiles order the canonically arranged optimal constants by majorization, yield exact formulas on a full-dimensional region, and recover the lower coefficient $(\log 2)/4$ throughout a broad class of anisotropic regimes.
Read the full abstract
Lattice Field Theory
First Two-Hadron Form Factor from QCD
We present the first QCD determination of an energy-dependent form factor for a two-hadron scattering state. In particular, we calculate the QCD contribution to the forward electromagnetic $\pi^+\pi^+ + \gamma\to\pi^+\pi^+$ amplitude at $m_\pi\approx 400$~MeV using lattice QCD. Because lattice calculations are performed in a finite Euclidean spacetime, where asymptotic scattering states are absent, this amplitude cannot be accessed directly from correlation functions. Instead, we can constrain this and related amplitudes nonperturbatively using a finite-volume formalism that requires two ingredients: the discrete finite-volume spectrum and finite-volume matrix elements of the electromagnetic current. We calculate three-point correlation functions coupling $\pi^+\pi^+$ finite-volume states and extract the corresponding electromagnetic matrix elements. Combining these results with the previously determined spectrum, we constrain the infinite-volume $\pi^+\pi^+ + \gamma\to\pi^+\pi^+$ amplitude in the forward limit. Using constraints from Lorentz symmetry, unitarity, and analyticity, we describe this amplitude in terms of a single real-valued energy-dependent two-hadron form factor. The resulting amplitude and form factor agree with the Ward-Takahashi identity across all energies and moving frames considered, providing the first QCD validation of this finite-volume approach and a pathway toward first-principles studies of the electromagnetic structure and electroweak responses of resonances and multi-hadron bound states.
Read the full abstract
Artificial Intelligence
Efficient Test-Time Adaptation through Human-Agent Interaction
AI agents are trained on population-scale data to encode broad capabilities spanning those of many practitioners. Yet the artifacts they produce rarely meet the personal bar professionals need to stake their reputation on. On realistic, open-ended tasks where success criteria are heterogeneous and insufficiently documented, individual expertise lives precisely in the elevation and departure from the average. In practice, iterative human-agent interaction surfaces criteria that users cannot fully specify up front, yet apply repeatedly across tasks. We argue this cross-session interaction data is a rich, underused signal for closing the gap to individual expertise. In this work, we propose test-time adaptation through human-agent interaction (TAHI), which integrates these signals into agent context and weights, and crystallizes each user's training and evaluation criteria via an evolving rubric module. We adapt agents to 30 individuals in two high-utility domains, writing and visual creation, on a total of 600 tasks. Our agents improve solo task success by 4.5-20.9% within only tens of tasks. Meanwhile, our evolving rubric module serves as a scalable annotation tool, creating evaluation rubrics that catch 16.0-22.3% more failures than those from LMs or humans alone. While agents are adapted towards individuals, we show these personalized agents also produce improvements in success of up to 8.8% that generalize across users.