Papers

Large language models can generate plausible code-review comments, but such comments may contain technically incorrect claims that mislead developers. We study technical trustworthiness judgment: determining whether a review comment's core technical claims are correct and applicable to the reviewed code in its repository context. Existing code-review benchmarks primarily evaluate review generation, issue discovery, or general comment quality, but do not directly assess whether an agent can determine the technical trustworthy of an individual review comment. To fill this gap, we introduce CRJudgeBench, a benchmark of 1199 instances constructed from real pull requests and expert-verified perturbations, covering both trustworthy and plausible but untrustworthy comments. We further present Sentinel, a repository-grounded agentic judge that actively gathers code evidence to verify review comments before making judgments. Starting from Qwen3-Coder-30B-A3B-Instruct, Sentinel is trained on the CRJudgeBench training split through iterative action-level learning from a privileged teacher. On the 359-instance CRJudgeBench test set, Sentinel achieves 76.60% accuracy, outperforming GLM-5.3 by 6.13 percentage points and its base model by 19.78 points. These results show that even state-of-the-art general-purpose LLMs struggle to identify untrustworthy comments, while iterative action-level learning substantially improves the accuracy of repository-grounded trustworthiness judgments. Our dataset is available at https://huggingface.co/datasets/dcloud347/CRJudgeBenchmark

Searches for the production of $\tau$-sleptons ($\tilde{\tau}$), charginos ($\tilde{\chi}^{\pm}_1$), and neutralinos ($\tilde{\chi}^0_2$) in events with at least one hadronically decaying $\tau$-lepton are presented, targeting compressed mass spectra scenarios. For charginos and neutralinos, the cascade decays producing either a $\tau$-slepton or a $\tau$-sneutrino are considered. Two final selections are studied, requiring either one hadronically decaying $\tau$-lepton and one other charged lepton, or two or more hadronically decaying $\tau$-leptons. To enhance sensitivity to the compressed mass spectra, production modes with initial-state radiation (ISR) jets are considered. The searches use 140 $\mathrm{fb}^{-1}$ and 53 $\mathrm{fb}^{-1}$ of proton-proton collision data recorded by the ATLAS detector at the Large Hadron Collider during 2015-2018 at a centre-of-mass energy of $\sqrt{s} = 13$ TeV and 2022-2023 at a centre-of-mass energy of $\sqrt{s} = 13.6$ TeV, respectively. The searches provide sensitivity to the compressed mass spectra scenarios with light $\tau$-sleptons. The results are generally consistent with the Standard Model, although several deviations with local significances of $2.1-2.4 \sigma$ are observed. These deviations weaken the observed exclusion limits. For direct $\tau$-slepton production, in the ISR channel alone the limits are set for $\Delta m(\tilde{\tau}, \tilde{\chi}^0_1) = 30$ GeV for $\tau$-slepton masses up to $115$ GeV at 95% confidence level. Given the observed data, no exclusion limits are set for direct $\tau$-sleptons production when statistically combining all considered channels. For electroweakino production, limits are set for $\Delta m(\tilde{\chi}^0_2, \tilde{\chi}^0_1) = 30$ GeV for $\tilde{\chi}^0_2$ masses up to $200$ GeV. For pure chargino pair-production, limits are set for chargino masses up to $175$ GeV for the same mass splitting.

Rapid growth in renewable generation is changing power-system operation and increasing the importance of ancillary services for security and reliability. As inverter-based resources displace conventional synchronous generation, system operators must procure a broader set of services spanning five categories: frequency control and reserves, voltage and reactive power support, system stability, restoration, and congestion management. This paper compares ancillary-service market design in Great Britain, Germany, Denmark, Italy, California, Texas, and Australia. It examines how these systems define, procure, remunerate, and coordinate services as variable renewable generation increases. The review develops a consistent product-level classification, provides comparable timing visualisations of frequency-control and reserve products, and maps implementation modalities across services and systems. Four trends emerge: faster and more granular frequency-response products, more explicit procurement of stability capabilities including inertia, system strength, and dynamic voltage support, broader participation by storage, demand-side resources, and other inverter-based resources, and greater use of locational, targeted, and technology-neutral procurement. Challenges include limited harmonisation, incomplete valuation of flexibility and stability capabilities whose value depends on location and operating conditions, barriers to participation by distributed and emerging resources, and insufficient coordination between market and operational arrangements. The comparison finds no convergence towards a single market design and supports differentiated but coordinated ancillary-service frameworks reflecting each service's technical, temporal, and locational characteristics.

Models of triggered seismicity commonly aggregate the contributions of all past earthquakes into a total predicted rate. This aggregation does not distinguish a forecast dominated by one recent earthquake from an equal-rate forecast supported by many events distributed across a longer history, even though these configurations can have different persistence and sensitivity to catalog and parameter uncertainties. We introduce the finite-horizon triggering core size \(N_{p,T}(t)\), defined as the minimum number of past earthquakes whose expected contributions account for a fraction \(p\) of the total triggering potential over the future interval \([t,t+T]\). The definition applies to any triggering model that assigns nonnegative event-specific expected contributions over a specified future window. We develop its theory for marked Hawkes and epidemic-type aftershock sequence processes, where the weights are the expected numbers of direct offspring produced by individual past earthquakes. The finite horizon makes the observable well defined for every normalized Omori kernel with exponent \(\theta>0\), including the long-memory regime in which the corresponding infinite-horizon quantity diverges. We derive continuum and marked-Poisson benchmarks, an exact representation of the marked-Poisson distribution, and high-\(p\) distributional laws. For finite-horizon Omori memory, the triggering core grows as \((1-p)^{-1/\theta}\) with a realization-dependent amplitude controlled by the total triggering weight; for exponential memory, the growth is logarithmic.[...]

Meta-heuristics like genetic algorithms (GAs) effectively explore solution spaces but often yield suboptimal results and are sensitive to parameter tuning. In contrast, the mass-spring-damper layout optimization (MSDLO) method is fast and excels in locally optimizing resource arrangements, though its effectiveness depends on the initial resource placement. This paper first employs genetic algorithms to establish an optimal resource placement in a greenfield layout. We then apply MSDLO to fine-tune the positions and orientations of these resources. The results are presented and discussed, highlighting this approach's practicality in effectively addressing industry needs.

Rapid and accurate self-inductance estimation for multilayer rectangle-shaped planar windings is essential for modern high-frequency power converters, yet traditional workflows rely on complex mathematical equations, rigid monomial formulas or unexplainable black-box machine learning (ML) models that degrade severely outside their training domain. This paper introduces an explainable ML framework unifying post-hoc feature attribution (SHAP and permutation importance) with Kolmogorov-Arnold Network-guided symbolic regression via the SR-KAN framework to discover closed-form analytical equations without prior structural assumptions. Evaluated on a new open-source dataset of over 10,000 Finite Element Analysis (FEA) simulations across seven out-of-distribution (OOD) classes, standard tree-based ensembles exhibit severe extrapolation errors (> 36%), whereas the unconstrained SR-KAN expression achieves a robust OOD relative error of 8.22%. Experimental verification across 55 physical printed circuit board prototypes (up to 8 layers, with inductances from 4.11 {\mu}H to 559.27 {\mu}H) confirms that the KAN-discovered expression translates effectively to real-world hardware, predicting inductance with a mean absolute relative error of 6.26%. To support reproducible research, the complete FEA simulation dataset and prototype measurements are released open-source.

We compare and study different Bivariate-Bicycle (BB) encodings and node choices for distributed quantum operations such as transversal non-local CNOTs. We observe that while some encodings have more physical qubits requiring more ebits for a distributed computation, reducing ebit consumption alone might not be the best criterion for selecting an encoding for the logical qubits, if the goal is to reduce the logical error rate of the distributed computation; for example, one should also consider encodings with a larger distance, which using more physical qubits can provide. We consider distributed, or non-local, CNOTs and computation of the global gate (GCZ) over distributed logical qubits as examples. The choice of encoding enables particular concurrent operations; e.g., we show that a transversal physical GCZ on a self-dual $[[120,8,12]] BB$ code can realize eight concurrent logical GCZ operations after accounting for the logical permutation induced by transversal Hadamard.

Pairwise losses are increasingly used for reward learning even when pointwise rewards are observed, with mixed empirical results. When and why do pairwise losses outperform pointwise losses? We study this question in a grouped offline contextual-bandit setting allowing multiple actions per context, capturing many reward learning scenarios. We compare Value Regression (VR), which regresses observed rewards pointwise, with Value Difference Regression (VDR), which regresses reward differences between a pair of actions sampled under the same context. We consider a semiparametric model where the mean reward is the sum of a learnable action-dependent component and an arbitrary context-dependent yet action-independent nuisance, capturing context-specific disturbances. Using a unified localized analysis, we prove finite-sample regression guarantees for finite and linear function classes and translate them into offline-regret bounds. For finite classes, VDR eliminates the misspecification term in the VR bound and improves a reward-scale-dependent error term by averaging over actions within each context, a benefit absent from the corresponding VR term. For linear classes, neither method uniformly dominates: within-context differencing removes nuisance-induced bias but may increase estimation variance relative to using absolute rewards when the misspecification is sufficiently low. This yields a feature geometry-dependent bias-variance tradeoff, which we corroborate with numerical experiments.

Fang and Lin [J. Algebraic Combin. 63 (2026), Art.~58] asked whether, whenever $F$ is edge-color-critical with $\chi(F)=r+1$, every non-$r$-partite, $F$-free graph of maximum adjacency spectral radius must also maximize the number of edges. We give a negative answer. Let $F=K_1\vee\mu(K_3)$, where $\mu(K_3)$ is the Mycielskian of a triangle. This graph is edge-color-critical with $\chi(F)=5$. We prove that $\SPEX_{5}(n,F)\cap\EX_{5}(n,F)=\varnothing$ for all sufficiently large $n$. Thus no graph can simultaneously maximize both the edge count and the spectral radius.

We develop quantum algorithms that construct block-encodings of solution matrices for continuous- and discrete-time algebraic Riccati equations (CAREs and DAREs), differential Riccati equations (DREs), and finite Riccati recursions. The Quantum Weighted Riesz Method, unifying four kinds of Riccati problems, constructs projectors onto the solution graphs using weighted Riesz branching operators and recovers the solution matrices. Under the stated access and normalization assumptions, our algorithms for CAREs and DAREs have near-optimal query complexity $\Theta(\mathcal R\alpha)$ up to logarithmic factors, with the generalized singularity factor $\mathcal R$ that reflects spectral separation, and the output normalization $\alpha$. For DREs and finite-horizon problems, the query complexity has an additional initialization conditioning factor $\kappa_{\rm init}$. We establish product query lower bounds for specified input-oracle families, demonstrating necessary joint dependence on spectral separation and initialization or output scale. We also prove that selected-value promise problems are BQP-complete on explicit circuit-generated families of single-control discrete-time algebraic and zero-terminal differential Riccati equations. Applications include classical feedback evaluation for linear-quadratic control, a query-complexity comparison for stable random-phase approximation Riccati equations in quantum chemistry, and heated boundary control with numerical validation.

Entanglement routing establishes entangled pairs between distant nodes of a quantum network by purifying and swapping pairs generated on elementary links. Existing methods typically restrict the decision space along two axes: the operation order, often fixed to purify-then-swap (PtS), and the purification schedule of each link, often restricted to pumping. Focusing on linear repeater chains with finite link capacities, we relax both restrictions and study the maximization of the expected end-to-end throughput subject to a fidelity threshold under two quantum-noise models. We develop a unified capacity-aware optimization framework comprising an exact mixed-integer linear program (MILP) under PtS, instantiated with either pumping or general tree purification schedules, and an exact dynamic program (DP) over arbitrary operation orders. Under symmetric Pauli noise, we prove that pumping converges to a fidelity strictly below unity, imposing a capacity-independent feasibility bound and a maximum chain length on every pumping-based PtS method, whereas general tree schedules yield a capacity-dependent bound. We further prove that post-swap purification never increases the throughput, so a free operation order can only increase feasibility. Numerical evaluations confirm the bounds: tree schedules serve over 90% of the requests that pumping cannot at moderate fidelity thresholds; within the pumping class, a free operation order recovers much of this advantage; but once links use tree schedules, the tree-based MILP and the order-exact DP serve identical request sets throughout. Operation order and purification schedule are thus substitutes, with the purification schedule the dominant factor governing feasibility under symmetric Pauli noise.

A detector that cannot take two snapshots closer than a time $\tau_0$ still identifies a fully observed finite Markov generator exactly, provided its gaps are randomized by two clocks. Identification is not detection. For a normal nonequilibrium network, the number of snapshots needed to reject every hidden equilibrium explanation grows as $e^{2\gamma_{\mathrm{irr}}\tau_0}$, where $\gamma_{\mathrm{irr}}$ is the slowest damping of rotating modes. No adaptive schedule beats this rate, and a test on one dependent trajectory attains it.

Engineering not only the magnitude but also the dynamics of synaptic weight updates is an important challenge for memristive neuromorphic hardware. Here, we show that Ag modification of TaO$_x$ memristors introduces an electrically selectable degree of freedom in synaptic depression through the interplay between Ag-related and oxygen-vacancy dynamics. Reference devices exhibit conventional bipolar switching, whereas Ag-modified devices display symmetric table-with-legs hysteresis loops, metastable intermediate states, and a strongly non-monotonic dependence of depression dynamics on programming amplitude. By varying only the pulse amplitude, the same device can be driven among three regimes: gradual sigmoidal depression, an abrupt update concentrated within a few pulses, and a broadly distributed evolution extending over more than one hundred pulses. A minimal coupled-state model, in which a vacancy-related switching variable interacts with a slower Ag-related internal degree of freedom, reproduces this gradual–abrupt–gradual crossover using two individually monotonic field-activated processes. The functional impact is assessed in a memristor-based multilayer perceptron. For MNIST, the abrupt, sigmoidal, and slowly evolving responses yield accuracies of approximately 50%, 85%, and 88%, respectively, while the gradual regime reaches approximately 72% for Fashion-MNIST. These results show that coupling Ag-related and oxygen-vacancy dynamics can transform synaptic depression from a fixed device characteristic into an electrically programmable property.

Large language models (LLMs) are increasingly deployed for natural-language logical reasoning, where the final answer is easy to check but the proof behind it is not. In natural-language logical reasoning, an intermediate conclusion should follow from its premises, and the resulting derivation should support the final answer. Existing methods lack machine-checkable verification of intermediate conclusions and answer-supporting proof dependencies, so they may assign credit to invalid or answer-irrelevant steps. We propose Proof-R1, an RL framework from formal verification that trains LLMs to construct verifiable proofs for natural-language logical reasoning. Proof-R1 admits a generated conclusion into the verified proof state only when the corresponding reasoning action satisfies the proof obligations through UNSAT-based machine-checkable formal verification. Proof-R1 also recovers the answer-supporting dependency closure to trace the proof structure of the final answer and align outcome credit with the proof dependencies. Experiments demonstrate that Proof-R1 improves answer accuracy across three logical reasoning benchmarks and four backbone models and outperforms training-free agents and training-based methods in terms of reasoning-process verifiability.

Droplet collision governs droplet population dynamics in many chemical engineering processes, such as spray drying, spray cooling, agricultural spraying, and combustion. Existing analytical models impose deterministic, pairwise boundaries between collision outcomes, whereas machine-learning classifiers lack the explicit functional form required of analytical collision submodels. In this study, we develop a probabilistic symbolic-distillation model using nearly forty thousand experimental events spanning eight regimes and five dimensionless parameters, including over five thousand data for ambient pressure up to 50 atm. A machine-learning teacher learns the joint outcome-probability landscape from these data, and symbolic regression subsequently distils it into eight class-specific expressions that jointly define a coupled analytical model. The resulting analytical field replaces abrupt regime switching with finite-width fuzzy boundaries. It outperforms the evaluated conventional analytical boundary models and reveals that their main limitation is the inability of zero-width boundaries to represent gradual probability transitions. The "biased-dice" sampling scheme provides a statistically consistent and practically convenient model implementation for Eulerian-Lagrangian spray simulation.

We prove dynamical stability in arbitrary dimension for the modified Mullins–Sekerka flow, the gradient flow of the sharp-interface Ohta–Kawasaki energy on the flat torus. Specifically, we show that if an initial set has the same volume as a strictly stable critical set for the energy and is sufficiently close to it in $C^{3,\alpha}$, then the flow exists for all times and converges exponentially fast, in every $C^k$ norm, to a translate of that critical set. The proof relies on a quantitative Alexandrov-type estimate for strictly stable critical sets of the energy.

Multi-reward guided reinforcement learning (i.e., RL) offers a promising way to improve joint audio-video diffusion models along several complementary objectives, including modality-specific quality, cross-modal semantic alignment, and temporal synchronization. Its effectiveness, however, depends on two quantities that change during training: where reward-driven updates should act, and how competing rewards should be combined. Existing methods tend to rely on fixed routing and reward weights, failing to track evolving model functions. To address these limitations, we propose Adaptive Reward Routing to jointly adapt update locations and reward coordination during forward-process RL (i.e., DiffusionNFT) of joint audio-video diffusion models. Our method consists of two components. (i) Cross-Modal Influence-Guided Routing (Localizing Updates): We use bidirectional cross-attention responses as an efficient proxy for evolving cross-modal influence, dynamically reweighting token-aware losses and scaling gradients across cross-modal layers without additional model interventions. (ii) Preference-Preserving Modality-Aware Reweighting (Coordinating Rewards): We preserve predefined weights as preference priors and use branch-specific reward-gradient interactions as residual corrections after warm-up. This resolves evolving conflicts without letting dominant rewards suppress weak but essential objectives. Extensive experiments demonstrate consistent improvements in modality quality, semantic consistency, and audio-video synchronization over strong RL baselines. Ablations and mechanism analyses further validate the complementary benefits of adaptive update routing and reward coordination.

The Berry phase is a fundamental geometric quantity for characterizing the geometry and topology of quantum many-body systems. Previous work established a super-polynomial quantum advantage in Berry phase estimation at inverse-polynomial precision, given an ansatz state approximating the initial ground state. However, whether this hardness persists for Berry phases quantized by symmetry, which can be distinguished at constant precision, remained open. In this work, we characterize the computational complexity of quantized Berry phase estimation. First, we prove that deciding whether the Berry phase is exactly $0$ or $\pi$ is $\mathsf{BQP}$-complete when the spectral gap is inverse-polynomially small. This result implies that distinguishing the quantized Berry phase still has a super-polynomial quantum advantage, assuming $\mathsf{BPP}\neq \mathsf{BQP}$. To prove this result, we introduce an encoding that maps the output of any quantum computation to a Berry phase that is exactly $0$ or $\pi$. Second, we extend this hardness to physically motivated Hamiltonians on a 2D square lattice, including Heisenberg and XY interactions. Third, we provide a classical polynomial-time algorithm for geometrically local Hamiltonians on fixed-dimensional lattices with constant spectral gaps. Therefore we identify that the spectral gap is a key resource for the classical tractability, rather than precision. These results establish a super-polynomial quantum advantage for computing a quantized topological invariant that is unchanged under symmetry- and gap-preserving deformations, and provide new steps towards practical quantum advantage in quantum physics.

Pretrained text-to-image models contain broad visual knowledge, yet they cannot reliably acquire or refine a specific visual identity from only a few references while preserving compositional control. Token-embedding methods are compact but often underfit identity, whereas adapter-based methods improve fidelity through persistent weight updates that can be costly to store and interfere when concepts are composed. We introduce V-Engram, a trigger-indexed external memory mechanism for Stable Diffusion 3.5. Each concept is assigned an explicit trigger that retrieves concept-specific memory, whose gated directions enter frozen text-encoder and MMDiT context states as relative residuals. Separating this memory from backbone adaptation enables prompt-selective and multi-concept access without merging model updates. Experiments show that V-Engram broadly matches DreamBooth-LoRA in overall subject fidelity while showing advantages in settings such as contextual subject preservation. Prompt-matched loading retrieves only matched entries, reducing most additional adaptation-state loading for a single-concept query. Qualitative results further demonstrate paired-trigger composition and same-class separation, while prompts without registered entries retain the frozen model's base behavior. Together, these results establish trigger-indexed memory as a modular interface for adding targeted visual evidence without rewriting the generator.

The semidefinite linear complementarity problem associated with the multiplicative map $M_A:\mathbb{S}^n\to\mathbb{S}^n$ defined by $M_A(X)=AXA^T$, is considered, in the case when $A\in\mathbb{R}^{n\times n}$ has rank one. A linear map is said to have the $Q_0$-property if every feasible instance of the associated complementarity problem admits a complementary solution. We prove two sharp, complementary results. When $A=xy^T$ with $x,y\in\mathbb{R}^n$ linearly independent, we construct a matrix $Q\in\mathbb{S}^n$ for which the associated semidefinite complementarity problem is feasible, but possesses no complementary solution; proving that $M_A$ does not have the $Q_0$-property. Conversely, when $A=uu^T$ for some nonzero $u\in\mathbb{R}^n$, we prove that $M_A$ has the $Q_0$-property. These two results provide a complete characterization of the $Q_0$-property within the family of rank-one multiplicative maps.

Tool-using LLM agents remain vulnerable to indirect prompt injection because trusted instructions and untrusted observations share one context, allowing malicious content to steer consequential input-filtering defenses. Multi-path consensus defenses still leave a high attack success rate because they examine content or aggregated outputs rather than authorizing effects, especially for the within-tool attack, which preserves the intended tool but manipulates its arguments. Data-Flow Control such as CaMeL provides stronger guarantees, but incurs substantial time latency that limits practical deployment. We introduce ToolFence, which compiles a typed authorization blueprint before execution, enforces it through a deterministic monitor, and when the blueprint is incomplete asks a judge to grant new capabilities rather than adjudicate each concrete call. ToolFence provides two key advantages. First, its fine-grained provenance-aware authorization enables the system to distinguish user-authorized values from untrusted observations, effectively addressing the within-tool attack. Second, its deterministic fast path and capability-level runtime grants substantially reduce the frequency of expensive judge calls, improving runtime efficiency. On AgentDojo with Qwen3-max, ToolFence reduces overall ASR to near zero with only a 3.80 percentage-point clean-utility drop and practical runtime overhead.

Handwritten Text Recognition (HTR) systems have become an indispensable tool for the digitization of historical documents. Not only do they cut down time and cost, but they also allow democratizing access and processing of their contents by generating their transcripts. However, literature in HTR currently focuses mostly on specialized models that require large amounts of annotated samples to achieve satisfactory performance. We explore the use of In-Context Learning with pre-trained Vision-Language Models (VLMs) to create a transcription pipeline without updating the model's parameters. We then evaluate this pipeline across multiple collections and models, and demonstrate that general-purpose VLMs can be effectively taught how to transcribe handwritten text from images. To assess how our observations may translate to practical applications, we evaluate the performance in a Cross-Domain (CD) scenario, where context examples are drawn from a different collection than the query image. Results in both the controlled In-Domain (ID) scenario and the realistic CD scenario follow the same patterns. First, as context size grows, the error range is expected to narrow towards the average performance. Thus, larger context sizes sacrifice the performance of the oracle-best sampling for lower expected error rates. The results obtained show that, without any parameter updates, this methodology has strong potential to compete with traditional HTR in the presence of domain shift. Moreover, we show and argue that some context samplings work better than others and suggest more effort should be put into finding an ideal sampling method in future work.

The fast-developing quantum technology is propelling thermal machine, which lies at the heart of thermodynamics, into a new golden era. Being stochastic because of the non-negligible thermal and quantum fluctuations, the efficiency and power of a finite-time quantum heat engine exhibit a trade-off, preventing the simultaneous achievement of high power and high efficiency. Conventional studies treat the statistics of a heat engine by assuming canonical thermalization, which is valid only under weak-coupling conditions. We here investigate the noncanonical effect on the statistics of the efficiency and power for a finite-time quantum Otto cycle. It is revealed that their average values and variances are highly controllable by engineering the energy spectrum structure of the composite system consisting of the working substance and the bath. Near the quantum critical point induced by the formation of a bound state in the energy spectrum, the average efficiency and power can be boosted simultaneously, meanwhile, both of their variances are decreased. Deepening our understanding of the usefulness of quantum reservoir engineering in thermodynamics, our result lays the foundation for the realization of efficient and stable quantum energy devices.

We study whether the coordinate dependence of the improved Self-Organizing Migrating Algorithm (iSOMA) can be reduced while retaining its inexpensive leader-directed migration mechanism. We introduce iSOMA-AR, which learns a basis from successful migration displacements and selectively applies the standard perturbation mask in that basis. On the complete noiseless BBOB suite, iSOMA- AR significantly outperformed baseline iSOMA across matched conditions, with the largest gains on geometrically difficult landscapes. A targeted ablation shows that the learned orientation is beneficial on a rotated ill-conditioned landscape and that moderate changes of the gate threshold and rotation cap preserve the qualitative result. On CEC 2011 Real World Optimization Problems, iSOMA-AR outperformed iL-SHADE on most problems, although its advantage over baseline iSOMA was not statistically significant. A canonical-jSO rerun is reported as a post-hoc sensitivity check alongside the original jSO-derived comparator. On frustrated-spin variational quantum objectives, adaptive rotation improved most transverse-field conditions, while gains on the diagonal and anisotropic models were absent or selective. Under strong effective sampling noise, the SOMA variants were the most robust population-based methods in the comparison, but iSOMA-AR was not significantly better than baseline iSOMA. Repairing all-zero PRT masks greatly reduced repeated-point evaluations without changing endpoint quality significantly, making this implementation detail unlikely to explain the noise result. Overall, adaptive rotation is most useful on coordinate-sensitive deterministic problems, while the observed noise robustness appears to arise mainly from the underlying SOMA migration mechanism.

Spatial regulation suffices to guarantee reliable, low-latency communication using variable-length stop-feedback (VLSF) codes over multiple access channels. The delay performance of VLSF codes under successive interference cancellation is characterized for transmitters within a coverage region. For deployments modeled by hardcore point processes, the signal-to-interference-plus-noise ratio (SINR) admits a positive lower bound for all transmitters within the coverage region. Under a predefined total error probability constraint, these transmitters are decodable within an upper-bounded delay. For each transmitter, the decoding delay is well approximated by an inverse Gaussian distribution parameterized by the codebook size, target error probability, and the SINR lower bound. Numerical results confirm that hardcore regulation provides uniform performance guarantees across links.

The ground state of the square-lattice $t$-$J$ model at $1/8$ hole doping has been the subject of a long-standing controversy, with conflicting numerical reports of stripe order versus superconductivity. Here, we address this debate using the variational uniform matrix product state (VUMPS) method, which optimizes the wave function directly in the thermodynamic limit along the cylinder axis and eliminates the boundary pinning effects that can arise in finite-cylinder calculations. On an infinite cylinder of circumference $L_y = 8$, we find a uniform-charge-density $d$-wave superconducting ground state with quasi-long-range pairing correlations. The dominant pairing occurs at zero center-of-mass momentum, while no sizable finite-momentum pairing component is detected. Twisted-boundary-condition calculations reveal a finite superconducting phase stiffness, with the order-parameter phase winding smoothly to follow the applied flux, providing an independent probe of the robustness of the superconductivity. Moreover, both charge and spin correlations decay rapidly, with no signatures of long-range stripe or magnetic order, in contrast to an $L_y=6$ cylinder at $1/6$ doping where the same method readily identifies stripe order. Our results provide methodologically distinct evidence in the long-standing stripe–superconductivity debate in the pure $t$-$J$ model and reveal uniform $d$-wave superconductivity in a minimal model of doped Mott insulators.

Multi-turn visual search agents answer questions about high-resolution images by iteratively deciding where to look. Reinforcement learning for these agents rewards only the final answer, leaving the search process unsupervised. Consequently, faulty routes in which the reasoning process is erroneous yet the final result is correct arise frequently, which in turn leads to ineffective training, i.e., scaling along the wrong paths. In this paper, we introduce HaPRL, the first framework to reinforce the search process with human search behavior. We first build an annotation platform and collect 1K+ human-annotated data with fine-grained behavioral signals. During training, a carefully designed judge scores each rollout with task-adaptive weights, anchored on the distilled trace of how a human annotator actually searched the same image. Extensive experiments show that HaPRL consistently outperforms outcome-based RL, and early-stage process supervision yields 6.7x more improvement in subsequent outcome-based scaling. Our results also demonstrate the importance of aligning model behavior with human process annotation signals, which offer new insight into the training of foundation models.

We study sparse linear contextual bandits with knapsack constraints under joint reward and consumption corruption. Consumption corruption creates a challenge beyond corrupted rewards: it affects not only statistical estimates, but also the recorded budget, resource prices, and stopping decisions that govern future allocation. We develop Robust Optimistic Primal–Dual (ROPD), an estimator-modular framework that combines corruption-aware confidence widths with online resource prices and a budget-safety rule. With concrete sparse implementation, ROPD achieves regret against a clean population-LP benchmark of $\widetilde O(T^{2/3}+\Gamma T^{1/3})$ under forced exploration and population-design coverage, and $\widetilde O(\sqrt T+\Gamma)$ under on-policy realized-design coverage, for a supplied valid corruption bound $\Gamma$ under the stated proportional-budget scaling and fixed model/design parameters. When the corruption level is unknown, Shared-Grid adapts confidence radii around common point estimates fitted to a single realized history, incurring explicit initialization and master-comparison costs; its sharper on-policy guarantee additionally requires recommendation coverage. Both methods preserve observed budgets on every realization and bound clean resource violation by cumulative consumption corruption. These results connect corruption-robust sparse estimation with resource accounting, pricing, and stopping in high-dimensional online allocation.

Low-mass galaxies during the Epoch of Reionization are expected to form stars in short bursts separated by periods of inactivity, but what regulates this cycle, and how sensitive it is to the modelling of stellar feedback, remains unclear. We use the MEGATRON high-redshift zoom simulation suite, in which four star-formation and stellar-feedback prescriptions are evolved from identical initial conditions, to study the regulation of stellar mass growth in galaxies with $M_{*} \sim 10^{4} - 10^{8} M_{\odot}$ at $z = 8.5$. We classify galaxies as Star-Forming, Mini-Quenched (no significant star formation over the last 10 Myr), or Quenched (no significant star formation over the last 100 Myr). The feedback prescription controls both how often galaxies are quiescent and how this depends on mass. Models with stronger or more clustered feedback yield larger quiescent fractions compared to the fiducial and Varying IMF models. Duty cycles at fixed mass also differ by factors of several between models. The cold gas profiles of Quenched galaxies are consistent with two physically distinct modes of quenching: mechanical gas evacuation, producing compact, cold gas-poor systems, and thermodynamic suppression, retaining a cold but star-formation-inactive reservoir. At these masses, quiescent galaxies fall below un-lensed JWST/NIRCam detection limits, whereas star-forming galaxies of the same mass do not, and consequently flux-limited samples will be biased towards active phases. Measuring the quiescent fraction as a function of stellar mass in lensing fields would provide a direct test of stellar-feedback models at Cosmic Dawn.

Language-guided navigation requires connecting partial observations to a persistent spatial reference and learning how actions change that representation. We introduce InsightMap, a framework that uses top-down maps as both explicit spatial memory and action-conditioned prediction targets. Historical views are linked to labeled map locations, and a shared multimodal backbone jointly learns navigation action prediction and post-action map generation. Map prediction provides auxiliary training supervision, while navigation inference decodes actions from the observed spatial context. An aligned RGB-D data pipeline supports a common interface for navigation, visual question answering, situated reasoning, and 3D grounding. On the validation-unseen splits of R2R-CE and RxR-CE, InsightMap achieves success rates (SR) of 56.9% and 54.9%, respectively. Adding map-prediction supervision improves R2R-CE SR by 4.3 and success weighted by path length (SPL) by 3.2 percentage points. On static spatial tasks, InsightMap achieves 103.7 CIDEr on ScanQA, 60.1% exact-match accuracy on SQA3D, and 53.1% grounding accuracy at 0.5 IoU on ScanRefer with detected object proposals. On Unitree Go2, it outperforms NaVid and NaVILA in hallway, lab, and office environments.