Abstract: Multi-agent LLM systems face communication bottlenecks using natural language tokens, lacking end-to-end differentiability. While dense vector communication helps, existing methods are inflexible due to fixed topologies and static transformations. We propose the Adaptive Sparse Dense Communication Network (ASDNet), a novel framework for efficient, flexible, and context-aware dense communication. ASDNet employs a dynamic Communication Hub per agent, intelligently selecting sparse partners and adaptively generating optimal dense vector transformations. This end-to-end differentiable architecture enables joint optimization of communication and inference. Experiments with an ASDNet variant, built on a foundational LLM, demonstrate consistent outperformance against state-of-the-art dense communication baselines and other open-source LLMs across diverse benchmarks, with efficient training. Ablation studies confirm dynamic target selection and adaptive transformations are critical. Further analyses highlight ASDNet's enhanced efficiency, superior qualitative outputs, and robust low-data performance, showcasing its potential for scalable multi-agent collaboration.

Abstract: The intensive use of synthetic pesticides and fertilizers has raised environmental concerns. Sustainable alternatives, such as plant biostimulants and plant resistance inducers, offer promising solutions by enhancing growth, yield, stress tolerance, or activating defense responses against pathogens. However, the physiological impacts and combined effects of these products remain poorly understood, limiting evidence-based application strategies. Here, we evaluated the effects of a biostimulant and a plant defense inducer on durum wheat (Triticum turgidum ssp. durum), a key cereal crop in the Mediterranean Basin. Using controlled experiments, we assessed plant growth, chlorophyll content, and resistance to Zymoseptoria tritici, while considering potential trade-offs between growth promotion and defense activation. As expected, our results indicate that the biostimulant improved growth and photosynthetic performance, whereas the plant resistance inducer enhanced protection against Z. tritici. But the combination of these two treatments can trigger mitigated interaction effects, influenced by varietal genetic background. This study provides novel insights into the interactions between plant growth promotion and defense induction in durum wheat. Understanding these multi-factorial effects (in particular genotype effect) enables the identification of optimal treatment strategies, supporting the development of sustainable crop management practices that reduce chemical inputs while maintaining productivity and resilience under biotic stress.

Abstract: Background/Objectives: Age-related changes in body composition (bone, muscle, and adipose tissue) are often assumed to follow linear, sex-specific patterns. Some evidence suggests that these trajectories are nonlinear, and their timing and dynamics remain poorly characterized. Osteosarcopenic adiposity/obesity (OSA), defined by the coexistence of osteopenia/osteoporosis, sarcopenia, and excess/redistributed adiposity, is recognized as a body composition disorder associated with multiple morbidities, but its impact on age-related body composition trajectories has not been fully explored. We aimed to delineate sex-specific, age-related trajectories of bone, muscle, fat mass, and intramuscular adipose tissue (IMAT), identify inflection points across adulthood, and compare patterns in individuals without and with OSA. Methods: Cross-sectional data from 9717 healthy Caucasian adults (aged 20–90 years) enrolled in a multicenter Italian study were analyzed. Body composition was measured using validated bioelectrical impedance analysis. LOESS regression was employed to identify age-related inflection points. Standard diagnostic criteria defined osteopenia/osteoporosis, sarcopenia, adiposity, and OSA. Results: Men exhibited earlier peaks and midlife stability in bone and muscle mass, followed by later decline. Women showed lower baseline values, multiple early-life inflection points, and sharper midlife downturns, particularly around menopause. Fat mass increased steadily in men but followed a multi-phasic pattern in women. IMAT rose progressively with age in both sexes. Adults with OSA, identified in participants even as young as 20 years, demonstrated destabilized trajectories, earlier downturns in bone and muscle, and more complex body fat and IMAT patterns. Conclusions: Distinct sex-specific patterns and mitigating effect of OSA on body composition trajectories were identified. Early detection of OSA may be crucial for preventing acceleration of musculoskeletal decline and rise in adiposity.

Abstract: Manilkara (Sapotaceae) includes tropical tree species of high ecological and socio-economic value, yet genetic and phylogenetic evidence remains uneven across taxa and eco-geographic regions. Here, we synthesize studies conducted between 1999 and 2025 which summarize the use of molecular markers to infer genetic diversity, connectivity, population structure, and evolutionary relationships within this genus. The studies are dominated by PCR-based marker systems, including dominant markers (like RAPD and SCoT) and microsatellites from the nuclear genome and plastid genome. Other studies rely on PCR-amplified sequence loci, such as ITS and chloroplast regions, while others use high-throughput technologies, including NGS-assisted SSR development, sequences of complete plastomes, and targeted nuclear sequencing. Overall, studies using SSRs provide the most informative estimates for within-species diversity and fine-scale structure, whereas plastid datasets (cpSSR/cpDNA) mainly support inference on maternal lineages and plastid-based relationships but can be constrained by uniparental inheritance and limited variation, especially under small sampling. Some limitations found include heterogeneous sampling, inconsistency in reporting the methodological parameters, and limited connection with ecological or phenotypic parameters which restricts chances of inferences on demography and adaptation. Based on this review, future research in Manilkara would benefit from setting up a broader taxonomic and geographic coverage, incorporating genome-wide technology where feasible to strengthen conservation management, and breeding opportunities in Manilkara.

Abstract: While climate change creates the overarching biophysical stress on Ethiopian agriculture, institutional and governance structures primarily mediate agrobiodiversity outcomes, often trading evolutionary resilience for short-term productivity. This review synthesizes cross-sectoral evidence from Ethiopia’s major highland and rangeland systems to demonstrate that climate change acts as a systemic stress test, exposing latent vulnerabilities in agricultural policy, seed regulation, and land tenure systems. The widespread loss of agrobiodiversity, documented by genetic erosion rates ranging from 56% in barley to over 65% in teff and wheat, including total displacement in certain districts, is largely driven by a structural conflict between productivity imperatives and ecological stewardship. Our synthesis reveals that policy silos, top-down extension models, and regulatory biases toward genetic uniformity collectively erode the functional heterogeneity required for climate adaptation. This institutional failure necessitates a governance-centered framework that formalizes pluralistic seed systems and empowers decentralized farmer innovation. Realigning governance incentives to treat agrobiodiversity as a strategic national asset is essential for securing Ethiopia’s genetic capital against accelerating climatic stress.

Abstract: Cancer remains the leading cause of death in domestic dogs. Conventional therapeutic approaches, including surgery, chemotherapy, and radiotherapy frequently fail to achieve sustained remission or stabilization. Oncolytic virotherapy, a rapidly advancing therapeutic modality in human oncology, is emerging as a novel strategy in veterinary medicine. This systematic review summarizes current knowledge on the application of oncolytic viruses (OVs) in canine cancer treatment, focusing on their mechanisms of action, safety profiles, and clinical efficacy. We evaluate diverse OV platforms, including myxoma virus, reovirus, vesicular stomatitis virus, canine adenoviruses, vaccinia virus, Sendai virus, and Newcastle disease virus, across preclinical and clinical studies in dogs with various malignancies. While several OVs have demonstrated favorable tolerability and modest antitumor activity, key challenges such as pre-existing immunity, optimization of dosing regimens, and rational combination strategies, remain to be addressed. This review emphasizes the translational significance of canine studies for both veterinary and human oncology, underscoring the critical need for rigorously designed clinical trials to refine virotherapy protocols and expand therapeutic options for canine cancer patients.

Abstract: Static linear lumped circuits (conductances, independent sources, and linear dependent sources, with no storage) can be studied through their boundary behavior: the set of boundary voltage–current pairs consistent with internal circuit laws. Fixing a set of accessible boundary nodes B of size n, and assuming standard well-posedness conditions for modified nodal analysis (MNA), we show that the boundary current injection vector iB depends affinely on the boundary voltage vector vB on an admissible affine set: iB = Yeq vB + i0 for all vB ∈ VB . We then provide a canonical boundary normal form that realizes this law using only indepen- dent current sources and voltage-controlled current sources (VCCS) connected directly to the boundary nodes. The construction is deterministic and idempotent, and it yields a complete classification: two circuits are behaviorally equivalent on the same boundary if and only if their normal-form parameters agree (modulo boundary constraints). A worked example (including a dependent source), an explicit VCCS synthesis list, and an exact numerical spot-check are included.

Abstract: The burning candle discussed in Faraday's lectures is used as an example to discuss the relationships between physical theories at different levels of organization: continuum mechanics at the macro level and statistical and quantum mechanics at the micro level. The first part of the paper examines the connections between theoretical and experimental physics. Physics theory serves as the foundation of the research program, but experimental research is a measure of ongoing development. Reasonable extrapolationism denotes a situation where the ideas of physical theory contribute to the development of experimental research. On the other hand, radical extrapolationism is used for a situation where the ongoing discussion goes far beyond experimental physics. In the second part of the paper, the arrows of explanation between continuum mechanics and statistical mechanics are considered and classified within the framework of the proposed terminology. The estimation of the properties of a substance from molecular constants and the derivation of continuum mechanics equations from statistical mechanics are considered. Qualitative explanations and emergence are also discussed.

Abstract: Knowledge-intensive organizations undergoing Industry 4.0 transformations face unprecedented behavioral and cognitive challenges as algorithmic systems increasingly mediate decision-making processes. This qualitative study examines how managers in knowledge-intensive organizations interpret, integrate, and respond to algorithmic knowledge within decision contexts characterized by technological complexity and data volatility.Through thematic analysis of secondary qualitative data from thirty organizational case studies, interviews, and practitioner narratives across digital consulting, advanced analytics, and technology-enabled service organizations, we identify critical behavioral dynamics shaping human-algorithm collaboration. Our findings reveal that managerial decision-making operates through three interconnected behavioral processes: interpretive sensemaking of algorithmic outputs, oscillation between algorithm appreciation and aversion contingent on organizational and decision contexts, and organizational adaptation mechanisms spanning cognitive supports, psychological safety, and distributed learning structures. Drawing on bounded rationality theory, cognitive load theory, and socio materiality perspectives, we develop an analytical framework explaining how technological complexity and data volatility mediate the relationship between managerial cognition and decision quality through psychological safety and organizational learning. We conclude that successful Industry 4.0 adoption in knowledge-intensive organizations requires deliberate attention to behavioral and organizational context factors, particularly psychological safety enabling risk-taking in algorithmic engagement, cognitive diversity fostering critical evaluation of algorithmic recommendations, and organizational learning structures supporting the development of algorithmic literacy. This research contributes to organizational behavior theory by articulating how digital complexity reshapes managerial cognition and organizational practice, and to practitioners by offering evidence-based strategies for managing human–algorithm complementarity during technological transitions.

Abstract: The Central Dogma has provided a foundational framework for understanding biological information flow, yet it does not fully explain how living systems maintain stable identity, functional robustness, and recoverability under continuous molecular noise and environmental perturbation. Here, I propose the Central Homeostatic Principle (CHP) as a complementary first-principle framework that shifts the explanatory center from information execution alone to the physical constraint architecture that makes biological execution possible. The CHP posits that, in living cells, a central homeostatic state functions as a system-level coordinating layer that defines the feasible space within which genetic and biochemical programs can operate.This framework is motivated by convergent evidence across mechanical confinement, electrophysiological coupling, membrane contact-site transduction, phase-state regulation, and non-genetic phenotypic heterogeneity, all of which indicate that global physical states can gate, reshape, or buffer molecular outcomes. Building from systemic prerequisites and material constraints, I further argue, through an exclusionary, first-principle analysis, that lipid-based boundary systems occupy a near-irreplaceable physical position in implementing this central homeostatic constraint in aqueous cellular life, not as exclusive causal authorship, but as the dominant substrate of feasibility control.To render the theory scientifically actionable, the manuscript provides a formal articulation of CHP, a three-tier realization model, operational corollaries, and a rule typology distinguishing strong and weak forms. It then derives a set of falsifiable hypotheses spanning temporal commitment dynamics, non-genetic resistance, aging-related resilience loss, state-engineering-based reprogramming, and evolutionary primacy in prebiotic systems. By reframing life as a problem of constrained state maintainability rather than information flow alone, the CHP offers a testable theoretical scaffold for integrating molecular biology, biophysics, systems biology, and translational state engineering.

Abstract: Flue gases generated by residential solid fuel combustion sources contain both particulate matter and gaseous pollutants. Their concentrations can be effectively reduced using emission control devices, which have traditionally been applied in small- and large-scale combustion sources. However, in residential combustion sources, such emission control devices remain unimplemented, except for electrostatic precipitators. This mini review provides a coherent summary of the current state of knowledge regarding emission control devices experimentally tested on residential combustion sources. It also outlines the technical and measurement challenges that must be addressed to enable effective, standards-compliant integration into residential combustion sources.

Abstract: Caffeine is widely consumed among health care workers (HCWs) as a coping mechanism for occupational demands and may influence depressive symptoms. This study investigated this relationship after adjustment for perceived stress and sleep quality. In this cross-sectional study on licensed HCWs at a tertiary hospital in Jeddah, Saudi Arabia, habitual caffeine intake was assessed using a validated caffeine food frequency questionnaire. Depressive symptoms were assessed using Patient Health Questionnaire (PHQ-9), with clinically relevant symptoms defined as PHQ-9 ≥10. Perceived stress and sleep quality were measured. Logistic and linear regression models evaluated the as-sociations with PHQ-9 ≥10 and PHQ-9 score respectively. Among 298 HCWs (mean age 37.5 years; 66.1% women), 18.5% had PHQ-9 ≥10. Mean caffeine intake was 216 mg/day (median 125 mg/day). HCWs with PHQ-9 ≥10 reported a higher caffeine intake than those without (Mean 282 vs. 201 mg/day and median 169 vs. 114 mg/day; p = 0.038). Caffeine intake was significantly associated with PHQ-9 score but not with PHQ-9 ≥10 after full adjustment (β = 0.331; p = 0.014 per twofold increase). Perceived stress and sleep quality were independently associated with PHQ-9 ≥10. Higher caffeine intake may reflect response to occupational strains rather than a primary depression risk driver.

Abstract: We study time-generalization in neural networks by training a shared iterative cell under explicit supervision of computation length. Rather than treating depth as fixed or learned implicitly, we provide a deterministic target step schedule and penalize deviations from the prescribed execution length, enabling controlled evaluation beyond the training horizon.We perform experiments on three representative dynamical regimes: contracting (Euclidean GCD), attractor-aligned (Log-Fibonacci), and expanding additive (Log-Factorial). We find that models can generalize to larger inputs when the effective iteration depth remains within the trained regime, yet often fail when required computation length increases, even if input magnitudes are moderate. Failure modes track the stability properties of the underlying update dynamics: contraction dampens errors, attractor alignment bounds them over finite horizons, and additive accumulation induces systematic drift.These results suggest that algorithmic generalization depends not only on function approximation but on the stability of learned update rules under composition. Explicit step conditioning improves interpretability and stability of computation depth, but does not by itself guarantee robust extrapolation to longer iterative chains.

Abstract: Recent large-scale simulations demonstrate that LLMs exhibit systematic performance degradation in multi-turn conversations, with unreliability increasing 112% across 200,000+ conversations (Laban et al., 2025). However, this "Lost in Conversation" phenomenon lacks mechanistic explanation. We present an entanglement framework for understanding context sensitivity in large language models using embedding-level variance analysis across 12 model-domain runs (4 philosophy, 8 medical) and 360 position-level measurements. We demonstrate that ΔRCI—a measure of context sensitivity introduced in Papers 1–2—tracks variance reduction in response embeddings. The correlation between ΔRCI and the Variance Reduction Index (VRI = 1 − Var_Ratio) is strong and highly significant (r = 0.76, p = 2.37 × 10⁻⁶⁸, N = 360). This relationship reveals bidirectional context coupling: convergent entanglement (Var_Ratio < 1, ΔRCI > 0), where context narrows the response distribution, and divergent entanglement (Var_Ratio > 1, ΔRCI < 0), where context widens it. The "Lost in Conversation" effect corresponds specifically to divergent entanglement. Two medical models (Llama 4 Scout: Var_Ratio = 7.46; Llama 4 Maverick: Var_Ratio = 2.64) exhibit extreme divergent entanglement at the summarization position (P30), producing highly unstable outputs when task enablement is expected. We introduce the Entanglement Stability Index (ESI) to predict which models will exhibit instability in multi-turn settings, transforming the descriptive observation that "LLMs get lost" into a predictive science of human-AI relational dynamics.

Abstract: The field of international relations faces significant research gaps as traditional frameworks struggle to address emerging challenges in the twenty-first century. This research paper presents a comprehensive qualitative analysis of four priority research domains that require urgent scholarly attention: artificial intelligence governance and global power dynamics, climate security and interstate conflict, digital sovereignty in the Global South, and non-state actors in hybrid warfare. Through systematic literature review and thematic analysis, this study identifies critical theoretical and empirical gaps in existing scholarship while proposing frameworks for addressing these deficiencies.The research employs a qualitative methodology incorporating document analysis, comparative case studies, and interpretive analysis of policy documents and academic literature. Findings reveal that traditional international relations theories, including realism, liberalism, and constructivism, require significant adaptation to address the multidimensional challenges posed by technological transformation, environmental change, and evolving security paradigms. The paper concludes with evidence-based recommendations for future research agendas, emphasizing interdisciplinary collaboration, methodological innovation, and policy-relevant scholarship. This analysis contributes to the ongoing discourse on advancing international relations scholarship in an era of unprecedented global complexity and interconnection.

Abstract: We introduce R-LayerNorm, a novel normalization layer designed to handle noisy and corrupted image data by dynamically adjusting normalization strength based on local noise estimates. Unlike standard normalization methods that apply uniform scaling regardless of local corruption, R-LayerNorm incorporates a learn- able noise-sensitivity parameter (λ) and a spatial entropy-based noise estimator. When evaluated on the CIFAR-10-C benchmark across six diverse corruption types, R-LayerNorm achieves a statistically significant average accuracy improvement of +4.95% (p < 0.001) over standard BatchNorm, with particularly strong gains on contrast (+14.52%) and frost (+6.88%) corruptions. The method serves as a drop-in replacement for existing normalization layers, requires minimal computational over-head (∼10%), and demonstrates robust performance across multiple random seeds. Code is available at: https://github.com/R-LayerNorm/R-LayerNorm/tree/main.

Abstract: In this paper, we propose an enhanced preamble scheme for the physical random access channel (PRACH) applied to low-altitude integrated sensing and communication (ISAC) systems, aiming to expand the sensing capability of traditional mobile networks with PRACH frames based on ZC sequences. To enable the network to possess target sensing capability before successful terminal access, we transform PRACH from a mere initial access channel into an ISAC system capable of supporting high-speed terminal access and user equipment sensing by introducing a time-frequency orthogonal block structure and orthogonal cover codes (OCCs). Specifically, we first derive the Cramér-Rao lower bound (CRLB) for estimating the distance and velocity of user equipment using OCC-ZC sequences, and establish the evaluation metric for communications named detection probabilities. Then, the ISAC problem is formulated as a multi-objective optimization function. Since the multi-objective optimization problem is non-convex, we propose the NSAG-II algorithm to solve it, simultaneously improving the estimation accuracy of distance and velocity in the sensing aspect and the detection probability in the communication aspect.

Abstract: Grazinglands store substantial soil organic carbon (SOC), yet their potential to act as net carbon (C) sinks depends on management-driven net ecosystem CO2 exchange (NEE). Process-based models capable of representing contrasting management strategies are essential for evaluating mitigation potential. However, the uncertainty inherent in model outputs is often overlooked, which can limit the reliability of predictions. Here, we apply the MEMS 3.0 model to quantify uncertainty in SOC and NEE predictions and evaluate model performance across diverse U.S. grazing ecoregions. We conducted a Bayesian calibration with observations of SOC and its fractions, and annual NEE measurements from six NEON grazing sites. Model evaluation was performed with an independent dataset collected from four experimental sites located in Oklahoma, Michigan, and Wyoming USA under prescriptive and adaptive grazing management treatments. The model performed well in predicting baseline SOC and estimating weekly ecosystem fluxes across sites. Although annual NEE estimates exhibited some discrepancies relative to flux-tower observations, ~90% of measured fluxes fell within simulated posterior predictive intervals. Moreover, the model is consistent with the flux-observations, demonstrating no significant treatment differences between prescriptive and adaptive grazing treatments at the Oklahoma and Wyoming sites. We demonstrated that MEMS 3.0 can represent SOC dynamics and ecosystem fluxes in grazinglands across contrasting climates. Our results show that neglecting uncertainty in measured and simulated fluxes can lead to misleading model-data comparisons. These findings highlights the importance of uncertainty quantification for robust interpretation of predicted grazing management outcomes and for supporting credible climate change mitigation and C accounting frameworks.

Abstract: This article develops a formal proof of Castilgiano Theorem in an elasticity theory context. The results are based on standard tools of applied functional analysis and calculus of variations. It is worth mentioning such results here presented may be easily extended to a non-linear elasticity context. Finally, in the last section we present a numerical example in order to illustrate the results applicability.

Abstract: In recent years, significant research has established that remdesivir and its parent nucleoside analog GS-441524 substantially improve clinical outcomes and reduce viral shedding in cats suffering from coronavirus-induced feline infectious peritonitis (FIP). Similarly, molnupiravir - another potent nucleoside analog - has gained prominence for its dual utility in treating FIP in veterinary medicine and COVID-19 in humans. In Japan, molnupiravir has been approved for clinical use since late 2021. Experimental animal models have provided insights into these mechanisms. Immunohistochemical (IHC) analysis has identified the expression of the SARS-CoV-2 nucleocapsid protein (NP) within pancreatic islet cells of humans and in infected cats, with NP expression increasing in a time-dependent manner following infection. Notably, IHC analysis also revealed NP staining within pancreatic ductal epithelial cells. Given that ductal epithelial cells serve as progenitors that differentiate into islet cells during development, this suggests a possible pathway for direct viral invasion of the endocrine islets. While the precise mode of viral entry remain to be elucidated, these findings underscore the potential for SARS-CoV-2 to directly compromise pancreatic integrity and endocrine function. Hence, antiviral drugs could counteract also the diabetogenic effect of coronaviruses in both animals and humans.