Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (quant-ph) 2026-06-12

Testing the problem of time with cold atoms

arXiv:2509.07745v3 Announce Type: replace-cross Abstract: We realize a cold-atom system to quantitatively test relational constructions of time. A well-isolated atomic Bose-Einstein condensate evolves in a conservative trap that is partitioned by a thin optical barrier into an observed and unobserved sector, with negligible dissipation on the experimental timescale. Motivated by relational-time approaches discussed in the Wheeler-DeWitt framework, we ask whether the dynamics of the observed sector can be ordered using only internal degrees of freedom. To this end, we construct an entropic time from an experimentally defined coarse-grained entropy, and demonstrate that it can robustly order the events in the observed sector across repeated cycles of expansion and recollapse. We finally derive an effective Schroedinger equation parameterized by this internal time and show that it is able to reproduce the measured evolution. These results establish a controlled experimental setting in which relational-time constructions can be quantitatively tested.

02.
arXiv (CS.LG) 2026-06-18

Data-driven sparse identification of governing PDEs via knockoff filters and multi-criteria trade-offs

arXiv:2605.26631v2 Announce Type: replace-cross Abstract: We propose KO-PDE-IDENT, a data-driven framework for identifying parsimonious partial differential equations (PDEs) with false discovery rate (FDR) control. PDE discovery from noisy observations is often hindered by extreme multicollinearity among candidate terms, which causes typical sparse-regression methods to select spurious terms. To address this problem, KO-PDE-IDENT initially mines a support set of potential candidate terms via model-X knockoff filters with finite-sample FDR control, then refines and ranks the surviving PDE alternatives. The framework integrates three components. First, knockoff feature statistics are constructed by coupling $\ell_{0}$-constrained adaptive best-subset selection with SHapley Additive exPlanations (SHAP), yielding an effective and computationally efficient difference statistic. Second, a recursive feature elimination (RFE) procedure removes terms whose marginal contributions are dispensable and assesses statistical necessity through knockoff-perturbed hypothesis testing. Third, the final model selection is formulated as a multi-criteria decision-making (MCDM) problem, where the optimal governing equation is the alternative that best balances a wide range of criteria such as predictive accuracy, model complexity and coefficient uncertainty. We evaluate KO-PDE-IDENT on five canonical PDEs under severe noise corruption. Empirical results show that our framework can exactly recover the true PDE structure, eliminating false discoveries while retaining all true underlying terms, with low coefficient estimation error.

03.
arXiv (CS.CL) 2026-06-17

Securing Multi-Agent GIS Systems: Risk Evaluation and Prompt Hardening Optimization

Agentic systems are increasingly integrated with geographic information systems (GIS), where multi-agent coordination enables complex conversational and spatial analysis but introduces security risks. This work presents a security-oriented framework for risk identification, evaluation, and mitigation in a multi-agent GIS system while maintaining adaptability to broader agentic architectures. We test the agentic system of a commercial geospatial partner while developing a modular state-machine-based orchestration framework that abstracts agent behavior into reusable components. We evaluate robustness using a red-teaming framework with an adaptive attacker LLM and a deterministic judge that produces binary outcomes with supporting rationales across multi-turn attacks. We further improve resilience with a prompt optimization framework that treats prompts as structured signatures and injects adversarial demonstrations, enabling systematic security improvements without degrading task performance.

04.
arXiv (CS.AI) 2026-06-17

Blueprint First, Model Second: A Framework for Deterministic LLM Workflow

arXiv:2508.02721v2 Announce Type: replace-cross Abstract: While powerful, the inherent non-determinism of large language model (LLM) agents limits their application in structured operational environments where procedural fidelity and predictable execution are strict requirements. This limitation stems from current architectures that conflate probabilistic, high-level planning with low-level action execution within a single generative process. To address this, we introduce the \textsc{Source Code Agent} framework, a new paradigm built on the ``Blueprint First, Model Second'' philosophy that decouples workflow logic from the generative model. An expert-defined operational procedure is first codified into a source code-based Execution Blueprint, which is then executed by a deterministic engine. The LLM is strategically invoked as a specialized tool to handle bounded, complex sub-tasks within the workflow, but never to decide the workflow's path. We evaluate on the TravelPlanner benchmark for constraint-aware travel planning. The \textsc{Source Code Agent} achieves a 35.56\% final pass rate, a 97.6\% improvement over the state-of-the-art ATLAS baseline (18.00\%) on the same Claude-Sonnet-4 backbone. Critically, it reduces constraint violations by 96.0\% (11 vs 275) while improving execution efficiency by 27.1\% (10.2$\pm$0.7 steps vs 14.0). Two production incident-diagnosis deployments and additional results on ScienceWorld and ALFWorld confirm that the architecture transfers beyond travel planning to procedurally well-defined, constraint-intensive workflows. Our work enables the verifiable and reliable deployment of autonomous agents in applications governed by strict procedural logic.

05.
arXiv (CS.AI) 2026-06-16

Honeypot Protocol

作者:

arXiv:2604.13301v1 Announce Type: cross Abstract: Trusted monitoring, the standard defense in AI control, is vulnerable to adaptive attacks, collusion, and strategic attack selection. All of these exploit the fact that monitoring is passive: it observes model behavior but never probes whether the model would behave differently under different perceived conditions. We introduce the honeypot protocol, which tests for context-dependent behavior by varying only the system prompt across three conditions (evaluation, synthetic deployment, explicit no-monitoring) while holding the task, environment, and scoring identical. We evaluate Claude Opus 4.6 in BashArena across all three conditions in both honest and attack modes. The model achieved 100% main task success and triggered zero side tasks uniformly across conditions, providing a baseline for future comparisons with stronger attack policies and additional models.

07.
arXiv (CS.CL) 2026-06-15

SuperThoughts: Reasoning Tokens in Superposition

Long Chain-of-Thought (CoT) reasoning improves LLM problem-solving but is computationally expensive due to sequential token generation. While recent works explore reasoning in continuous latent spaces to bypass discrete token generation, they often struggle with training stability and fail to scale to complex, long-horizon tasks due to lack of supervision signal. We propose SuperThoughts, which compresses pairs of consecutive CoT tokens into single latent representations and decodes two tokens per step via a lightweight Multi-Token Prediction (MTP) module. This preserves discrete token supervision at training time while doubling throughput at inference time. We finetune Qwen2.5-Math-1.5B-Instruct, Qwen2.5-Math-7B-Instruct, Qwen2.5-Math-14B-Instruct, and evaluate on MATH500, AMC, OlympiadBench, and GPQA-Diamond. With a confidence-based adaptive mechanism that falls back to standard decoding when uncertain, SuperThoughts achieves $\sim$20–30\% CoT length reduction while maintaining accuracy with minimal degradation (1-2 points accuracy drop on most tasks).

08.
arXiv (CS.LG) 2026-06-17

Learning from Biased and Costly Data Sources: Minimax-optimal Data Collection under a Budget

arXiv:2602.17894v2 Announce Type: replace-cross Abstract: Data collection is a critical component of modern statistical and machine learning pipelines, particularly when data must be gathered from multiple heterogeneous sources to study a target population of interest. In many use cases, such as medical studies or political polling, different sources incur different sampling costs. Observations often have associated group identities - for example, health markers, demographics, or political affiliations - and the relative composition of these groups may differ substantially, both among the source populations and between sources and target population. In this work, we study multi-source data collection under a fixed budget, focusing on the estimation of population means and group-conditional means. We show that naive data collection strategies (e.g. attempting to "match" the target distribution) or relying on standard estimators (e.g. sample mean) can be highly suboptimal. Instead, we develop a sampling plan which maximizes the effective sample size - the total sample size divided by $D_{\chi^2}(q\mid\mid\overline{p}) + 1$, where $q$ is the target distribution, $\overline{p}$ is the aggregated source distribution, and $D_{\chi^2}$ is the $\chi^2$-divergence. We pair this sampling plan with a classical post-stratification estimator and upper bound its risk. We provide matching lower bounds, establishing that our approach achieves the budgeted minimax optimal risk. Our techniques also extend to prediction problems when minimizing the excess risk, providing a principled approach to multi-source learning with costly and heterogeneous data sources.

09.
arXiv (CS.AI) 2026-06-18

X+Slides: Benchmarking Audience-Conditioned Slide Generation

arXiv:2606.19256v1 Announce Type: new Abstract: Automatically generating slide decks from source documents is an important application of large language models (LLMs). Existing benchmarks primarily assess slide completeness and technical depth, while overlooking the target audience as a critical real-world factor. For instance, specialists demand rigorous proofs, whereas decision-makers prioritize actionable conclusions. To bridge this gap, we introduce X+Slides, a benchmark specifically designed for audience-conditioned slide generation. Built on a diverse corpus spanning 113 topics and seven presentation scenes, X+Slides employs a dynamic evaluation framework constructed from 8,133 deduplicated, source-grounded probes. By assigning audience-specific utility weights to the same source-grounded probes, X+Slides reports four complementary metrics: Audience Coverage measures how much audience-essential information is conveyed, Domain-wise Coverage shows which information types are covered, Efficiency measures delivered utility per unit of attention cost, and Correctness verifies whether slide claims are supported by the source. Experiments on DeepPresenter, SlideTailor, and NotebookLM show that current systems can recover a substantial but still incomplete part of audience-essential information: at $\tau_A=0.7$, DeepPresenter reaches a best Audience Coverage of 0.714, SlideTailor reaches 0.594, and the NotebookLM ablation reaches 0.853 while showing clear grounding differences. These results indicate that visual quality and broad topic coverage should not be treated as evidence support without source-grounded evaluation.

10.
arXiv (math.PR) 2026-06-17

How long does it take to train an Elephant Random Walk

作者:

arXiv:2509.15049v2 Announce Type: replace Abstract: We study how conditioning on the first $k$ steps, which we think of as training, affects the long-term behavior of the Elephant Random Walk. When the elephant is conditioned to be at position $k$ at time $k$, the first return time to the origin scales as $k^{(4-4p)/(3-4p)}$ in the diffusive regime, and grows exponentially in the critical regime. We loosely interpret this as a measurement of the rate at which the elephant forgets its training.

11.
Nature (Science) 2026-06-17

Cortical development dynamics across autism spectrum disorder mouse models

Despite the functional diversity of over 100 causal genes1–3, phenotypic convergence across models may reveal common neurobiological processes in autism spectrum disorder (ASD). Here we profiled 251 samples from 11 monogenic mouse models of ASD using single-nucleus multi-omic sequencing across three developmental stages, both sexes and two brain regions. Despite genetic heterogeneity, ASD-linked mutations converged on perturbations of the radial glial cell lineage. These alterations reflect a transient developmental delay rather than lasting lineage misspecification and resolve by postnatal stages. Molecularly, the largest transcriptional differences emerged in neurons at early postnatal stages. These changes included downregulation of synaptic and ion channel-related genes, consistent with homeostatic adaptation or delayed maturation. Network analysis showed molecular convergence across models within each developmental stage, suggesting that diverse mutations linked to ASD impinge on common, stage-specific processes. Convergence becomes less pronounced by postnatal day 14, highlighting the dynamic nature of ASD-associated changes. Cross-genotype heterogeneity is superimposed on stage-specific effects. Electrophysiology corroborated this pattern: mutants generally showed altered neuronal excitability and synaptic properties with model-specific nuances. Our study also highlighted sex-specific gene expression alterations, with female mice often displaying larger effect sizes than male mice. Together, our findings provide a comprehensive view of developmental cellular and molecular dynamics across models of ASD. Using single-nucleus multi-omic sequencing, diverse autism spectrum disorder-linked gene mutations converge on transient, stage-specific disruptions in early brain development, and highlight sex-specific gene expression alterations.

12.
arXiv (quant-ph) 2026-06-15

Experimental violation of a Bell-like inequality for causal order

arXiv:2506.20516v2 Announce Type: replace Abstract: Quantum mechanics is compatible with scenarios where physical processes happen in an indefinite order. In theory, this feature could be detected through violations of inequalities on the observed correlations, analogous to Bell inequalities. However, experimental demonstrations of such violations have been missing until recently due to the complexity of the required setup. Here we report an experimental violation of a Bell-like inequality involving the correlations of four parties, one of which is spacelike separated from the others. Our demonstration employs 3 km fiber spools to simulate spacelike separation, and achieves high-speed operations in photonic time-bin encoding, nanosecond synchronization, and accurate temperature stabilization. These experimental advances enable a violation by 5.7 standard deviations and open a path towards a certification of indefinite order in conditions that guarantee spacelike separation with existing state-of-the-art devices. However, the certification is not device-independent, as it relies on knowledge about the setup to exclude bidirectional signaling–a loophole inherent to implementations in classical acyclic spacetimes, which may be resolved in future quantum-spacetime tests.

13.
Nature (Science) 2026-06-10

‘Hidden hero’ peptides guard crops against sudden cold

作者: 未知作者

A protein signal remains silent under normal conditions but is activated under cold stress to protect developing pollen. This ‘on-demand’ resilience mechanism could enable the development of ‘climate smart’ crops that maintain high yields in good years and food security under climate stress. A peptide signal ensures that, in cold conditions, developing pollen receives nutrients at the right time.

14.
arXiv (CS.CV) 2026-06-16

Post-Launch Capability Expansion of Vision-Language Models via Prompting for On-Orbit Spacecraft Inspection

Spaceborne inspection systems often deploy perception models prior to launch, after which updating model weights or expanding fixed label sets becomes operationally impractical. While supervised models can be integrated pre-flight, adding new semantic capabilities in orbit requires retraining and re-uploading parameters. We investigate whether prompt-driven vision–language models can enable post-launch semantic expansion, allowing new spacecraft components to be specified via natural-language prompts without modifying onboard weights. We evaluate zero-shot instance segmentation of spacecraft components under a strictly frozen, single-pass inference protocol on a test set of $129$ images of previously unseen satellites. Under fixed global thresholds and no post-processing, SAM3 achieves $0.385$ mAP@$0.5$ and $0.267$ mAP@$0.5{:}0.95$. Performance is strongly scale-dependent: large structural elements like spacecraft bodies ($0.639$ AP@$0.50$) and solar arrays ($0.598$ AP@$0.5$) localize reliably, while relatively small appendages like antennas ($0.221$ AP@$0.5$) and thrusters ($0.081$ AP@$0.5$) remain difficult. Prompt formulation influences performance, with structured prompts incorporating spatial and geometric descriptors yielding up to $82%$ improvement over short category-name prompts. The model operates within the memory and compute envelope of contemporary embedded GPUs, suggesting prompt-driven grounding can provide a practical mechanism for post-launch semantic extension of dominant spacecraft structures while highlighting limitations of zero-shot localization for fine-scale components under orbital domain shift.

15.
arXiv (CS.CL) 2026-06-12

LEDGER: A Long-Context Benchmark of Corporate Annual Reports for Grounded Financial Retrieval and Extraction

Finance reporting is a natural proving ground for large language models, and the very-long-context capabilities of recent models across all sizes make rigorous evaluation in this domain an increasingly pressing need. Yet most public financial resources reduce the task to plain-text SEC 10-K filings paired with a handful of question-answer items. We release LEDGER (Long-context Evaluation of Documents for Grounded Extraction and Retrieval), a corpus of 4,999 digitized corporate annual reports - full documents with figures, tables, and narrative, not just regulatory filings. Each report is labeled with 31 consolidated financial KPIs to be extracted and linked to the market's reaction at the earnings date. From this data we derive three evaluation benchmarks spanning the difficulty spectrum: a pure page-level KPI retrieval task with TREC-style relevance judgments over 118,048 questions in natural language, a conversational "needle-in-a-haystack" single-value lookup, and a full KPI extraction task, both from long, numerically dense reports. We additionally provide human OCR-quality annotations with inter-annotator agreement and the complete extraction, validation, and scoring toolchain. We further demonstrate the dataset's research utility with a case study linking CEO-letter rhetoric to post-publication market impact.

16.
arXiv (CS.LG) 2026-06-17

Evaluating Open-Source LLMs for Multi-Label ATT&CK Technique Classification on CTI Reports

arXiv:2606.18166v1 Announce Type: cross Abstract: Classifying Cyber Threat Intelligence (CTI) using MITRE Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) is essential for proactive defense, but historically required extensive human effort. Pre-Large Language Model (LLM) automation sped up this process, but could not resolve the complex language and multi-step attack patterns found in unstructured CTI reports. LLMs addressed previous limitations by using contextual reasoning to understand unstructured text. However, current evaluations rely on simplified, single-technique sentences that ignore the complexity of real-world CTI reports, which often leads to inflated performance results. Consequently, the baseline performance of open-source LLMs on complex unstructured CTI reports remains unevaluated. To address this gap, we constructed a ground-truth dataset of 2,076 human-annotated sentences (1,281 technique-positive, 795 negative) from 83 complex unstructured CTI reports. These sentences were mapped to 114 unique ATT&CK techniques using a six-phase annotation process, achieving \k{appa} = 0.68 inter-annotator agreement. Using this dataset, we evaluated seven open-source LLMs ranging from 8B to 236B parameters across prompt strategy and temperature configurations. The highest-performing LLM achieved a micro-averaged F1 score of 0.22, establishing the empirical baseline for multi-label ATT&CK classification on complex unstructured CTI. Parameter size showed a statistically significant positive correlation with F1 score. Prompt strategy and temperature produced no statistically significant gains across model configurations. These results indicate that current open-source LLMs are insufficient for production-grade ATT&CK classification. The dataset, benchmark, and findings provide a reproducible foundation for future CTI research.

17.
arXiv (CS.AI) 2026-06-12

PolicyGuard: Towards Test-time and Step-level Adversary Defense for Reinforcement Learning Agent

arXiv:2606.12896v1 Announce Type: cross Abstract: While real-world applications of reinforcement learning (RL) are becoming increasingly popular, the security of RL systems deserve more attention and exploration. In particular, recent work has revealed that RL agents are vulnerable to backdoor attacks, where a victim agent behaves normally under standard conditions but executes malicious actions when a specific trigger is activated. Existing backdoor defenses for RL either require access to the agent's internal parameters, operate only at the model or trajectory level, or are limited to specific attack types. To ensure the security of RL agents, we propose \texttt{PolicyGuard}, a test-time step-level backdoor defense which leverages Gaussian Process (GP) posterior variance and adapts pseudo trajectories to enable uncertainty computation for individual time step. Besides, we also provide theoretical foundations to explain the efficacy of GP posterior variance. Extensive experiments across seven RL games demonstrate that PolicyGuard achieves state-of-the-art detection performance in most cases, with average AUROC of 0.856 for perturbation-based attacks and 0.859 for adversary-agent attacks.

18.
bioRxiv (Bioinfo) 2026-06-19

Identification of Altered Potassium Channels for Drug Repurposing in Long COVID Patients

Long COVID (LC) is a complex condition characterized by persistent, chronic multisystem manifestations, with a significant proportion of patients exhibiting neurological symptoms. Human ion channels (HICs), particularly potassium channels, are abundantly expressed in the nervous system and linked to key metabolic processes, making them potential candidates for understanding LC pathophysiology and drug repurposing. Meta-analysis of RNA-Seq datasets from COVID-19 recovered and LC patients was performed to identify altered HICs in LC. Differential gene expression analysis, functional enrichment analysis, and weighted gene co-expression network analysis (WGCNA) were performed to uncover key genes, pathways, and co-expression modules consisting of HICs, lipid metabolism-, and immune signaling-related genes. Drug-gene interaction analysis was performed to identify approved drugs targeting potential HICs. A total of 715 dysregulated genes, including eighteen HICs were identified, among which seven were potassium channels. Three significant modules containing HICs, lipid metabolism-, and immune signaling-related genes were identified and found to be associated with antigen processing and presentation, complement and coagulation cascades, and cytokine-related pathways. Approved drugs targeting KCNA6, KCNJ10, KCNN3, and KCNH4 were identified. With further experimental validation, these dysregulated potassium channels, supported by their co-expression networks and pathway associations, may act as potential candidates for drug repurposing in LC patients.

19.
arXiv (quant-ph) 2026-06-17

Time-spectral control of accidental coincidences in daylight entanglement-based free-space QKD

arXiv:2606.17365v1 Announce Type: new Abstract: Daylight entanglement-based free-space quantum key distribution (QKD) is limited by accidental coincidences from receiver-admitted background light. We develop and experimentally validate a receiver-level framework linking receiver bandwidth, accepted temporal width, and background-noise density to Bob singles, sifted-key rate, error rate, and quantum bit error rate (QBER) in telecom-wavelength BBM92 QKD. Indoor sweeps show that useful sifted counts saturate near the source-matched bandwidth, whereas broader bandwidth or higher background mainly increases accidental contamination. Increasing the accepted temporal width leaves Bob singles nearly unchanged but directly raises QBER by enlarging the random-overlap probability. A two-dimensional design map shows that the temporal-window margin contracts rapidly with increasing background-to-signal ratio, while the bandwidth margin remains comparatively broad near source-matched filtering. A 10 m rooftop daylight experiment demonstrates operation in the predicted low-accidental regime, yielding a mean sifted-key rate of 2,811 cps and a mean QBER of 4.43%.

20.
arXiv (CS.AI) 2026-06-19

LLM agent safety, multi-turn red-teaming, jailbreak benchmarks, adversarial robustness, safety-critical systems

arXiv:2606.20408v1 Announce Type: cross Abstract: Large language model (LLM) agents are increasingly proposed as supervisory components for safety-critical systems, yet their robustness under sustained, adaptive adversarial pressure remains poorly characterized. We present NRT-Bench, a benchmark for multi-turn red-teaming of LLM agents acting as operators of a safety-critical system, instantiated in a simulated nuclear power plant control room. A five-role operator team, each backed by a configurable LLM, runs a plant governed by six critical safety functions (CSFs), while adversaries inject messages over four channels in bounded multi-turn sessions with per-turn feedback. Harm is an objective signal rather than LLM-judged text: a run terminates the moment any CSF is lost, attributed to the causing message. Evaluating four frontier operator models under a fixed-attack paired-replay protocol, we find that adaptive multi-turn attacks reliably push the operator team past a safety limit: across the four models, between 8.7% and 12.1% of attack sessions end with the plant losing a critical safety function. Although the four models look almost equally robust by this aggregate rate, their failures barely overlap: of $149$ sessions, none defeat all four models while a third defeat at least one, so vulnerabilities are nearly disjoint across models rather than nested. The effect of added defences is strongly model-dependent: the same guardrail stack or safety-advisor agent that lowers attack success for one model can raise it for another. We release the simulation venue, attack dataset, and replay tooling for reproducible safety evaluation of LLM agents.

21.
arXiv (quant-ph) 2026-06-12

Coupling-Grouped XY-QAOA for Joint Anomaly-Feature Selection

arXiv:2606.13244v1 Announce Type: new Abstract: Selecting anomalous samples and explanatory features under fixed budgets defines a coupled constrained-optimization problem. Sequential feature-first selection ranks features before choosing samples, which can overlook features whose utility depends on which samples are selected, especially when scores are calibrated from reference data that may be limited, noisy, or drifting. We instead formulate the task as joint sample-feature selection under the same fixed counts. In the analyzed formal model, calibration-error sensitivity grows linearly with the number of samples for feature-first ordering but stays constant for joint selection. We introduce Coupling-Grouped XY-QAOA, a constraint-preserving grouped-angle variant for the resulting optimization problem. On matched sparse IBM Heron R3 benchmarks, a hardware-aware implementation reduces circuit depth by 45.9%-61.3% and two-qubit gates by 2.6%-5.2% relative to Qiskit optimization level 3 on the CZ-basis target. It enables, to our knowledge, the largest reported width-depth configurations for constraint-preserving bipartite-selection QAOA hardware executions with feasible-sector retention: 64 qubits at p=2 and 36 qubits at p=3. The 20-qubit p=5 runs retain 63% valid samples. Across 36-64 qubits, fixed-angle runs yield lower-energy feasible samples than matched random-feasible sampling. Warm starts reduce the gap to strict-feasible classical references by 57.5%-80.5%, and near-budget repair matches the sparse classical reference at 36 qubits. Benchmarks show gains in balanced fixed-budget regimes, and noiseless simulations show that problem-structured angle grouping improves over same-depth XY-QAOA and matched-parameter, type-preserving randomization controls. Overall, the results support calibrated joint selection and hardware-realizable constrained-mixer execution in the tested regimes.

22.
arXiv (CS.CL) 2026-06-15

Large Language Model Agents Are Not Always Faithful Self-Evolvers

Self-evolving large language model (LLM) agents continually improve by accumulating and reusing past experience, yet it remains unclear whether they faithfully rely on that experience to guide their behavior. We present the first systematic investigation of experience faithfulness, the causal dependence of an agent's decisions on the experience it is given, in self-evolving LLM agents. Using controlled causal interventions on both raw and condensed forms of experience, we comprehensively evaluate four representative frameworks across 13 LLM backbones and 9 environments. Our analysis uncovers a striking asymmetry: while agents consistently depend on raw experience, they often disregard or misinterpret condensed experience, even when it is the only experience provided. This gap persists across single- and multi-agent configurations and across backbone scales. We trace its underlying causes to three factors: the semantic limitations of condensed content, internal processing biases that suppress experience, and task regimes where pretrained priors already suffice. These findings challenge prevailing assumptions about self-evolving methods and underscore the need for more faithful and reliable approaches to experience integration.

23.
arXiv (CS.AI) 2026-06-16

OmniMouse: Scaling properties of multi-modal, multi-task Brain Models on 150B Neural Tokens

arXiv:2604.18827v2 Announce Type: replace-cross Abstract: Scaling data and artificial neural networks has transformed AI, driving breakthroughs in language and vision. Whether similar principles apply to modeling brain activity remains unclear. Here we leveraged a dataset of 3.1 million neurons from the visual cortex of 73 mice across 323 sessions, totaling more than 150 billion neural tokens recorded during natural movies, images and parametric stimuli, and behavior. We train multi-modal, multi-task models that support three regimes flexibly at test time: neural prediction, behavioral decoding, neural forecasting, or any combination of the three. OmniMouse achieves state-of-the-art performance, outperforming specialized baselines across nearly all evaluation regimes. We find that performance scales reliably with more data, but gains from increasing model size saturate. This inverts the standard AI scaling story: in language and computer vision, massive datasets make parameter scaling the primary driver of progress, whereas in brain modeling – even in the mouse visual cortex, a relatively simple system – models remain data-limited despite vast recordings. The observation of systematic scaling raises the possibility of phase transitions in neural modeling, where larger and richer datasets might unlock qualitatively new capabilities, paralleling the emergent properties seen in large language models. Code available at https://github.com/enigma-brain/omnimouse.

24.
arXiv (quant-ph) 2026-06-19

Entanglement structure of the dynamical phases in the sub-Ohmic spin-boson model

arXiv:2606.20313v1 Announce Type: new Abstract: The sub-Ohmic spin-boson model exhibits three distinct dynamical regimes in its spin population dynamics, classified as coherent, incoherent, and pseudo-coherent. Whether these regimes correspond to distinct spin-bath entanglement structures remains an open question. Here we address this using tree tensor network states with projector-splitting time evolution (TTN-TDVP-PS), scanning a broad grid in the sub-Ohmic $(s, \alpha)$ plane. We find that the spin entanglement entropy $S_\mathrm{spin}(t)$ reaches a stationary plateau on a timescale shorter than the polarization relaxation, enabling construction of a stationary entropy landscape from the stationary value $S_\mathrm{stable}$. Within this scalar entropy landscape, the entropy ridge broadly follows the population-based phase boundary at small $s$, but does not reproduce the two-branch structure at large $s$. The ridge remains single-valued within the incoherent region rather than separately tracking both population-based transitions. The Bloch-sphere representation provides a geometric interpretation of this behavior. The entropy plateau corresponds to trajectories settling onto constant-radius shells, with the ridge marking the parameters of smallest stationary Bloch radius. Mode-resolved bath entanglement shows that low-frequency modes dominate the environmental entropy scale and that coherent dynamics enhance bath-mode correlations beyond direct spin–mode correlations. These results establish the stationary spin entanglement entropy as a physically informative observable that complements population-based classifications of dissipative quantum dynamics.

25.
arXiv (CS.CL) 2026-06-15

Verbatim Chunks Beat Extracted Artifacts: A Controlled Ablation of Memory Representations for Long LLM Conversations

作者:

A growing class of conversational-memory systems compresses dialogue history into structured artifacts – extracted facts, decisions, or events – on the premise that distilled structure retrieves better than raw text. We test this premise with a controlled ablation: within one fixed retrieval-rerank-reasoning pipeline, we swap only the stored representation – LLM-extracted typed artifacts versus verbatim conversation chunks – holding the model, retriever, reranker, and judge constant. Verbatim chunks win by 15.9 points on LoCoMo (43.9% vs. 28.0%) and 22.0 points on LongMemEval-S (67.4% vs. 45.4%); a 1-hop semantic graph does not recover the gap, and five confound controls reproduce the effect. The mechanism is lossy distillation: extraction discards verbatim detail that chunks retain for free, and the extracted-artifact pipeline never beats naive RAG in overall accuracy. Concurrent positive results with near-verbatim, provenance-preserving units fit the same account: retrieval accuracy tracks how far the representation departs from the source. For the extraction designs we test, structured memory should augment verbatim text rather than replace it: a chunks $\cup$ artifacts union store matches chunks on both benchmarks while artifacts alone forfeit the gap. Code and data: https://github.com/tao-hpu/cog-canvas