Science & Mathematics

Time-reversal symmetry enforces equal equilibrium linewidths for time-reversed phonon modes at opposite wave vectors. Here we show that a steady heat current can produce nonreciprocal anharmonic damping even in centrosymmetric nonmagnetic crystals. The effect arises from nonequilibrium occupations of scattering partners, which change the statistical weights of time-reversed three-phonon channels while their squared matrix elements remain equal. Within the single-mode relaxation-time approximation, we derive a linewidth correction that is linear in the temperature gradient and odd under wave-vector reversal. The response depends on scattering-partner velocities and equilibrium lifetimes and retains signed channel weights absent from the equilibrium linewidth. First-principles calculations for centrosymmetric MgO and diamond yield nonzero, mode-dependent linewidth corrections at selected finite wave vectors, demonstrating the mechanism in both ionic and covalent crystals. Reversing the heat current at fixed local temperature exchanges the broadening and narrowing of paired modes. This current-reversible linewidth contrast provides a spectral signature of occupation feedback, with potential implications for nonlinear phonon transport.

Deep generative design can produce complex geometries, but unconstrained representations may create undercuts that prevent release from a two-part mold. We present CAST, which reformulates three-dimensional shape generation under a selected mold-opening constraint as a two-dimensional image generation problem using paired top and bottom depth images. Each valid depth pair defines one material interval along a selected mold-opening direction, so reconstructions exclude undercuts in that direction by construction. Diffusion models generate candidate shapes in this representation, and a finite element analysis (FEA) guided optimizer refines the same two boundaries. We evaluate diffusion-based generation with limited training data comprising 211 SimJEB bracket designs. On SimJEB brackets, FEA-guided refinement with a volume penalty reduced compliance in 193 of 200 generated candidates. The median reduction across all candidates was 36.8 percent in the optimization FEA model. Reconstruction of existing brackets revealed a mass--stiffness trade-off: filling occluded regions increased material volume and reduced displacement under the evaluated loads. Paired depth images thus connect generative exploration and structural refinement in a shared design space that preserves the selected pull-direction restriction.

Language-model agents increasingly solve complex tasks through parallel reasoning and asynchronous information gathering. Controlled replay across two model families and three benchmarks shows that delivery timing redirects search decisions made before all results arrive, despite identical initial evidence. Acquisition disagreement reaches up to 24 percentage points above same-order repeat variation. This dependence persists after all initial evidence arrives and through a second search, turning temporary differences in what agents have seen into lasting differences in what they acquire. An acquisition-regret bound with a matching construction shows that invariance at a fixed early arrival count can limit the use of evidence already received. We introduce Confluent Evidence Acquisition (CEA), which instead releases local reasoning when each subquestion's evidence is complete and allocates shared search independently of completion order. For a fixed plan and stable conditional response law, we prove joint invariance of decisions, acquisitions, and answers. With question-only plans across three benchmarks, CEA brings acquisition disagreement to within 1.0 percentage point of repeat variation and achieves accuracy comparable to complete-input controls, improving over early commitment by up to 12.6 percentage points. Replacing CEA's fixed allocation with first-ready allocation under contention adds 21.6 percentage points of acquisition disagreement, while local query-target disagreement remains near repeat variation. Timed comparisons show that CEA reaches decisions earlier and responds faster than the same program waiting for all initial evidence.

Knowledge distillation is widely used to transfer capabilities from teacher models to student models, but recent work shows that it can also transfer unintended teacher properties. We study whether different hidden teacher properties survive the same distillation channels. Specifically, we compare two properties: benchmark-associated knowledge and refusal backdoors, and show that distillation treats them very differently. Teachers fine-tuned on benchmark data successfully pass that knowledge to students through all four distillation channels we test, including when students are trained only on teacher-generated number sequences containing no natural language or domain content whatsoever. Refusal backdoors tell the opposite story: backdoored teachers achieve near-perfect triggered refusal, yet sequence-distilled students inherit essentially none of this behaviour. Logit-based distillation transfers it substantially, reaching up to 85.5\% triggered refusal in students that saw zero backdoor examples. Distillation is therefore selective, and the selection criterion is not what data the student sees, but what structure the teacher's property has. The code is available at https://github.com/MJonibek/kd-leakage.

Multi-drone collaborative detection aims to enhance target recognition by leveraging complementary observations from overlapping aerial viewpoints. Existing approaches typically follow a two-stage pipeline that first establishes cross-view correspondences and then performs fusion-based recognition. However, this decomposition suffers from inconsistent optimization objectives, redundant pairwise modeling, and irreversible error propagation, where incorrect associations inevitably undermine subsequent recognition performance. To overcome these limitations, we propose the Coupled Association-Recognition Network (CARNet), a unified framework that jointly optimizes cross-view target association and recognition fusion through a shared target-pair representation. To improve association reliability in densely populated scenes with visually similar targets, CARNet explicitly incorporates spatial cues into pairwise relation modeling. Moreover, we introduce two complementary causal-inspired intervention strategies: the first enhances cross-view association by emphasizing appearance-confusing pairs, while the second bolsters the robustness of fusion-based recognition by learning stable multi-view representations under incomplete visual evidence. In addition, we present MDMT-CD, a fine-grained extension of the MDMT dataset that enriches the original vehicle annotations with detailed category labels, providing the first benchmark specifically designed for multi-drone collaborative vehicle detection. Extensive experiments demonstrate that CARNet consistently outperforms existing state-of-the-art methods. The MDMT-CD dataset and source code will be available at https://github.com/bigdreaming11/CARNet.

Retrieval-Augmented Generation (RAG) can ground LLM outputs in domain-specific knowledge, but legal documents are hard: standard chunking fragments hierarchical argument structure, and generic embeddings struggle with both precise citation matching and semantic understanding. We propose StrucChunk (Structure Aware Chunking) with hybrid dense-sparse retrieval for hierarchically structured legal documents. It treats documents as directed acyclic graphs of provisions, preserves clause boundaries, resolves cross-references before embedding, and combines BM25 with BGE-M3 via Reciprocal Rank Fusion. We evaluate on five corpora, three legal traditions and two languages. StrucChunk reaches 100% boundary precision (vs. 7.7% for recursive splitting on CrPC), preserves 97.6% of documented cross-references (vs. 0%), and matches section coverage with 40.3% fewer chunks on GDPR. On LegalBench-RAG (6,858 query-passage pairs), it achieves Char-Recall@8 of 48.18% vs. 23.59% for the best baseline (+104%), surpassing the published state of the art by +83%. On 501 queries each for CrPC and CPC (Indian law), Recall@5 improves from 0.557 to 0.804 (+44.4%) and from 0.158 to 0.259 (+64.6%); chunking accounts for 88% and 73% of the gain. On 501 GDPR queries, Recall@5 is 0.868 with R@1 improving +59%. On NitiBench-CCL (3,730 Thai financial law queries), HR@5 is 0.856 vs. 0.594 for naive chunking (+44.1%), showing that section-boundary alignment is the dominant factor across languages, and cross-reference augmentation adds +2.3% on cross-reference queries. We release the implementation, 5,256 evaluation queries across five legal corpora, annotation methodology and chunking quality metrics. Code: github.com/SinghSuryaDeep/legal-rag-strucchunk-1

The National Energy System Operator (NESO) Clean Power 2030 advice to the government, accepted as United Kingdom (UK) government policy in December 2024, outlines a rapid increase in wind and solar generation capacities with the associated infrastructure by 2030, with the objective of reducing carbon dioxide emissions from the UK electricity system to 5.2 million tons per annum. Our previous paper used real-time records from 2017 to analyze Clean Power 2030 proposal. This paper extends the previous assessment, using real time records of 2017, 2018 and 2019 to test the resilience of the UK energy system under more diverse weather patterns. Two different modelling methods predict that the 2030 carbon dioxide emissions will be five times greater than Clean Power 2030 prediction. Furthermore, the projections show that, due to increased curtailment, only about half of NESO proposed increases in wind and solar capacities will be economically justified. NESO also advised the government that 35GW of unabated gas-fired generation would maintain security of supply, whereas our models suggest that at least 42 GW will be required during winter wind lulls. Recent events, such as grid stability problems in 2026, also discussed in the paper, validate our estimated and suggest the risk of blackouts in the UK is increasing.

SNNs are attracting interest due to their potential for energy efficiency and event-driven processing. However, neuromorphic hardware is not yet mature enough to enable embedded deployment of complex spiking network architectures. In this work, we investigate a spiking version of the MoveNet architecture tailored for autonomous spacecraft operation, aiming to evaluate the deployability of this complex architecture compared to the non-spiking version for embedded, fully event-based solutions in space applications. Using the RAPTOR dataset, we compare the performance and operational cost of the proposed S-MoveNet-Lightning architecture with its standard version, MoveNet-Lightning. Our results show balanced advantages in computational resources and architectural settings for spiking U-Net-like architectures. An evaluation of the performance and hardware footprint of the proposed Spiking Neural Network (SNN) approach for pose estimation shows that, on average, 2.2 times fewer multiplication and accumulation operations are required compared to the non-spiking solution deployed on FPGA platforms with the AMD Deep Learning Processor Unit (DPU). These results provide valuable insights into the feasibility and potential of SNNs in complex aerospace applications and highlight their role in balancing efficiency and performance, paving the way for event-driven solutions in embedded space applications.

Resting-state fMRI (rs-fMRI) is a vital non-invasive tool for diagnosing brain disorders and uncovering pathological mechanisms. However, existing graph contrastive learning (GCL) methods often underexploit augmentation histories across training epochs and the relationships among different augmented views. In this paper, we introduce a topology-guided Multi-view graph Contrastive learning framework with Dynamic graph topology Augmentation (MCDA) for fMRI analysis. Unlike traditional GCL that uses fixed noise or edge-dropping, MCDA employs a Dynamic Augmentation Update (DAU) module that iteratively refines the augmented topology. Specifically, we incorporate self-aware and cross-sample augmentation strategies to simultaneously capture individual-specific functional signatures and inter-subject heterogeneity. We then design a topology-guided multi-view graph contrastive learning scheme that explores interrelationships not only between the original network and its augmented views, but also across different augmented views to maximize structural information preservation. We characterize the score-level optimization directions and establish a lower bound for the contrastive objective. The framework combines classification and contrastive losses for brain disorder diagnosis and examines discriminative functional connectivity patterns as potential biomarkers. Experiments on two public rs-fMRI datasets comprising 1,236 subjects demonstrate that MCDA outperforms the state-of-the-art methods. The code has been made publicly available at: \url{https://github.com/goodcodingagoodboy/MCDA}.

Post-hoc out-of-distribution detection usually relies on a single signal extracted from a frozen classifier, such as confidence, energy, features, or gradients. This dependence becomes fragile under classifier-level adversarial shift, where the shift can preserve the detector's chosen cue while changing the model's decision behavior. We investigate whether this failure mode can be mitigated by combining complementary detection criteria instead of selecting a single score. We introduce TriScore, a post-hoc OOD detector that combines three ID-calibrated signals from the same frozen backbone: an energy prior, a boundary-fragility score, and a transform-consistency score. These signals are standardized on ID validation data and fused by an input-dependent ID-residual gate, requiring no OOD labels, auxiliary model, training, or stored feature bank. Across four ImageNet-scale backbones, TriScore achieves the best mean AUROC among 15 post-hoc baselines under a non-adaptive adversarial-shift protocol, while remaining competitive on standard OOD benchmarks. Ablations show that the gains come from combining complementary cues rather than optimizing a single detector signal. These results support multi-criterion post-hoc detection as a practical route toward more robust OOD detection under classifier-level shift. The source code is publicly available at github.com/gursoyege/TriScore.

Trading discipline often fails people. Our study tested whether four AI trading agents followed their strategies and risk rules. Offering traders the evidence and an agent evaluation toolkit to separate luck from following the rules, before giving agents meaningful autonomy or capital. Our study analysed 2,627 closed positions from 12 February to 27 August 2026 across the same asset universe and execution conditions. Most positions were simulated against live prices with a 0.09% round-trip fee; 18% involved real Hyperliquid orders. Each position declared a strategy, stop and target, with a limit of five open positions per agent. The fleet finished net positive after fees. Rule adherence was high across all four agents, including the unprofitable one. Profitability tracked how often agents won relative to their break-even win rates. Institutional-style controls ran by default and contained losses through a 29% market drawdown and recovery, with documented exceptions. The discipline held for all four agents. Agents kept 97.5% of gross loss within declared stops, and typical losses closed at declared levels. Declared risk remained stable after losses and broadly stable across calm and volatile months. Twenty-three of 24 runs respected the position limit; logs identify the exception's size and duration. The same live data produced different results across assigned strategies. The trade-log tests can be rerun from the published trade snapshot; regime tests also require the published Bitcoin prices and regime labels. Section 3.5 specifies the stop-breach tolerance. These trade-level results are not account-level returns or proof of a lasting edge. Our study provides an evaluation toolkit to help traders distinguish rule-following from luck before committing capital. For anyone considering allocating capital to agents to trade, discipline should be the entry test, not the only test.

Typed decision models (TDMs) return a probability over caller-defined options, and Jev is described by its vendor as calibrated. But in deployment class frequencies change, so calibration measured on a fixed test set need not hold where thresholds are applied. Existing evaluations leave two gaps: the gains of vendor decision training have not been separated from those of the backbone under an identical readout, and TDMs have been tested only at the test set's class distribution. We ask whether decision training improves a model over its own backbone, and whether a TDM's calibration and thresholds withstand prevalence (label) shift as well as a fine-tuned classifier's. We compare three TDMs (Jev, Tev1-4B, Nimble-9B), their base models and two trained classifiers on identical decisions from five public datasets under simulated shift, with 35 paired tests fixed before any test was computed. First, decision training is not uniformly beneficial: Nimble-9B, which differs from Qwen3.5-9B only by a LoRA adapter, is better calibrated on intent tasks but 12.0 Macro-F1 points worse on financial sentiment. Second, under moderate shift Jev's expected calibration error (ECE) and threshold risk grow significantly more than a fine-tuned DeBERTa's only on financial sentiment, by 1.7 and 2.5 percentage points; averaged over 2,000 shift scenarios (exploratory), the gap is at most one point on every dataset. Third, EM prior correction with the model's mean development prediction as its prior needs no labels and lowers Jev's ECE under shift by 2.7 and 1.8 points on the two datasets where it is highest, although labelled recalibration lowers it far more. The three TDMs we tested are neither uniformly better than their backbones nor uniquely fragile under shift; for calibration, their main weakness is the miscalibration they already have without shift. We release all inputs, code and outputs.

We study how many output samples match the worst-case state reconstruction error obtainable from a continuous history under bounded measurement noise. For observable scalar-output linear time-varying systems with $n$ states, $n$ samples suffice for exact reconstruction, whereas at most $n^2$ attain the continuous-history minimax error in the $\ell_\infty$ norm. An explicit construction proves sharpness even for real-analytic systems. For time-invariant systems, strict total positivity yields one $n$-sample schedule optimal for every state coordinate; endpoint-observed tridiagonal cooperative chains satisfy this conclusion in physical coordinates. Interior sensing behaves differently: an explicit three-state family requires four samples despite positivity and distinct real eigenvalues. The strict three-state gap persists under small stabilizing damping. A nilpotent four-state example and an interval proof for a six-state mechanical system give gaps. The results identify how temporal sampling and the prescribed state-error geometry determine the measurements required for minimax reconstruction. Time-domain experiments illustrate the resulting estimation bounds.

Transparency systems -- such as Key Transparency (KT), Certificate Transparency, and binary transparency -- require that all honest participants see the same view of an untrusted server's directory, or, failing that, that the misbehaviour is publicly detected. Every existing design closes this gap with an external trust anchor: a publicly available bulletin board (BB), a blockchain, or an out-of-band (OOB) gossip network among auditors. We observe that transparency systems already have a built-in resource that has so far been overlooked: the social graph induced by the users themselves. We ask whether it can replace the bulletin board or the external gossip channel entirely, and answer in the affirmative (if occasional out-of-band interactions are allowed). We formalise a new primitive, Social-Graph based Forking-Resilient Transparency (sg-FORT), which captures the security guarantees of a BB-free transparency system and we construct Fortress Fork Resistant Transparency logs Relying on Social graph Structure, to our knowledge the first such protocol. In every epoch either all users output the same post, or at least one honest user produces a publicly verifiable Proof of Misbehaviour. Our guarantees hold against an arbitrary malicious server, including a snapshot adversary that may compromise a user and learn its secret state, as long as the server does not compromise close users. In honest executions, OOB interactions occur only when a user resets their secrets, so the total OOB cost approaches the natural lower bound of one call per reset -- dramatically below the full-gossip alternative. We prove security under standard cryptographic assumptions on signatures and dictionary commitment. Under simplifying assumptions (synchronous epochs, always-online users), we view this as a first feasibility result that opens up the design space of social-graph-based transparency systems.

Leveraging our earlier formalism for charge-state decay [see R. Kuate Defo et al., 108 235208 (2023)], we investigate relative population partitioning in the presence of singly negatively charged substitutional boron impurities (B$^-_{\rm C}$) in diamond samples, where these diamond samples also contain substitutional nitrogen (N$_{\rm C }$) impurities and have ether-terminated surfaces. Effectively, given any two defects, one donor and one acceptor, such that their charge-transition levels, their relative populations, and the temperature are known, the timescale for charge transfer between the two defects is completely determined. We use this insight and experimental measurements for the timescale for charge transfer between NV$^-$ centers and N$_{\rm C}$ hybridized with the ether-terminated surface of diamond, as well as the temperature and the relative populations for the NV$^-$ centers, [see Z. Yuan et al., Phys. Rev. Res. 2, 033263 (2020)] to argue that an additional defect must be modulating the relative populations of N$_{\rm C}$ hybridized with the ether-terminated surface of diamond. The shifts in the local Fermi level for N$_{\rm C}$ hybridized with the ether-terminated surface of diamond are consistent with the additional defect being a substitutional boron impurity (B$_{\rm C}$). Ultimately, our results allow for the determination of local relative concentrations of ionized defects.