AUTHOREA
Log in Sign Up Browse Preprints
LOG IN SIGN UP

Preprints

Explore 66,105 preprints on the Authorea Preprint Repository

A preprint on Authorea can be a complete scientific manuscript submitted to a journal, an essay, a whitepaper, or a blog post. Preprints on Authorea can contain datasets, code, figures, interactive visualizations and computational notebooks.
Read more about preprints.

The Holmes Functional Operator for π md : 27-Dimensional Torsional Synthesis and the...
Lee Holmes

Lee Holmes

March 23, 2026
This synthesis establishes a non-Planckian framework for interpreting the 1.33ms Length of Day (LOD) deceleration. We introduce the Holmes Functional Operator (H f o) as a method to reconcile vacuum stiffness with terrestrial rotational drag. By applying the Holmes Unified Scaling Law (HUSL-Matrix), we demonstrate a consistent correlation between the 230 km/s galactic helical move and geodetic nodal snaps. Normalization of terrestrial inputs via the H f o establishes the Holmes Seventh Law (Elastic Limit, E L) as a quantized yield point of the manifold. Empirical results confirm the predictability of secondary flow events at Node 18 (Missouri) and axial peaks at Node 284.
Task-Aware Adaptive Neural Preprocessing for Robust Biometric and Object Recognition...
Li Wei

Li Wei

and 2 more

March 23, 2026
Current computer vision systems deployed on consumer hardware often suffer significant performance degradation when subjected to non-ideal environmental conditions, such as extreme lighting, low-resolution sensors, or high-dynamic-range (HDR) thermal inputs. Conventional pipelines treat Image Signal Processing (ISP) and downstream feature extraction as distinct, disjoint stages, leading to suboptimal information retention during bit-depth reduction (e.g., 16-bit to 8-bit conversion). This paper proposes the Adaptive Signal-to-Feature Network (AS2F-Net), a unified, end-to-end Hybrid Neural Architecture that jointly learns optimal signal tone-mapping and semantic feature extraction. By introducing a differentiable Neural ISP Block, the system adapts raw sensor inputs to maximize the accuracy of downstream tasks-specifically biometric authentication and hazard detection—on resource-constrained edge devices. We validate our approach through extensive experimentation, demonstrating that AS2F-Net outperforms static ISP baselines by substantial margins in low-light and high-contrast scenarios while reducing inference latency by approximately 15% compared to standard ResNet backbones. Furthermore, we provide a rigorous mathematical analysis of the quantization noise suppression and establish a theoretical bound on the gradient variance during Mixed-Precision Training.
Optimizing Mobile Dermatological Analysis via Synthetic Data and Hardware-Aware Neura...
Aizada Almazbekova

Aizada Almazbekova

March 20, 2026
This paper presents a rigorous, mathematically grounded framework for deploying high-fidelity dermatological classification and biometric analysis directly on resource-constrained mobile hardware. Addressing the fundamental trilemma of data scarcity, computational latency, and privacy in mHealth applications, we introduce a novel hybrid learning pipeline. We formulate a robust approach utilizing Generative Adversarial Networks (GANs) with Adaptive Differentiable Augmentation (ADA) for synthetic data generation, explicitly addressing the long-tailed class imbalance inherent in medical datasets and the under-representation of diverse skin tones. Furthermore, we propose a latency-constrained Differentiable Neural Architecture Search (DNAS) utilizing a Lagrangian relaxation method to optimize network topology specifically for mobile inference. We provide an analytical derivation of quantization error bounds for INT8 mobile inference and introduce a Hessian-aware bit-width assignment strategy. Additionally, we model the thermodynamic behavior of mobile SoCs under continuous inference loads, deriving a control-theoretic approach to prevent thermal throttling. Experimental results on a composite dataset of 20,015 images demonstrate that our synthetic augmentation strategy improves mean Average Precision (mAP) by 14.3% in low-data regimes. Crucially, we present a demographic breakdown showing that our balanced synthetic training reduces the False Negative Rate (FNR) gap between Fitzpatrick Skin Types I-II and V-VI from 12.4% to 3.1%. Our architectural optimizations reduce inference latency by 45% (<12ms) on standard mobile DSPs compared to baseline EfficientNet architectures, enabling real-time, privacy-preserving diagnostics.
Adaptive Overcurrent and Reclosing Protection in Renewable-Rich Distribution Networks...
Sinawo Nomandela

Sinawo Nomandela

March 20, 2026
The rapid integration of inverter-based renewable energy resources into medium-voltage distribution networks is fundamentally altering fault-current characteristics and challenging the reliability of conventional overcurrent protection and automatic reclosing schemes. Reduced and variable fault contributions, bidirectional power flow, and operating-condition-dependent behavior can lead to protection blinding, miscoordination, delayed fault clearing, and unsuccessful reclosing. To address these challenges, adaptive protection strategies supported by communication-assisted coordination and digital substation technologies have been widely investigated. This paper presents a comprehensive review of adaptive overcurrent and reclosing protection methods for renewable-rich and active distribution systems. Existing approaches are systematically classified into setting-group-based, measurement-assisted, optimization-driven, communication-assisted, and data-driven techniques, and are compared based on operating principles, implementation complexity, and practical applicability. Particular emphasis is placed on IEC 61850-enabled digital substations, including generic object-oriented substation event (GOOSE) messaging and distributed intelligent electronic devices that enable real-time coordination and dynamic setting updates. Adaptive reclosing schemes are examined alongside overcurrent protection to highlight the importance of integrated protection–reclosing design in high-penetration DER networks. The review further surveys validation practices reported in the literature, ranging from offline simulation to hardware-in-the-loop and real-time digital simulation (RTDS) platforms, and assesses their effectiveness for practical deployment. Comparative analysis identifies current limitations, research gaps, and emerging trends, including hybrid protection architectures, cybersecurity considerations, and scalable digital substation implementations. Finally, future research directions and practical recommendations for utilities are discussed. Overall, this review provides a structured technical reference and design guide for researchers and practitioners seeking to develop robust, standards-compliant adaptive protection and reclosing solutions for modern active distribution networks.
Lagrangian Decomposition C. R. Gimarelli (January 10, 2026) \affiliation Indepen...

Farid Ait-Chaalal

and 3 more

March 25, 2026
High-resolution Snow Water Equivalent (SWE) data are essential for hydrological modeling, water resource management, and climate impact assessments in snow-dominated regions. Regional models offer improved resolution but are computationally expensive, motivating the use of deep learning-based emulators. In this work, we introduce the Linear Attention Snow Downscaling Model (LASDM), a hybrid architecture that combines convolutional neural networks (CNNs) for local feature extraction with linear attention blocks for modeling long-range dependencies. LASDM is designed to be highly parameter-efficient using fewer than one million trainable parameters. We evaluate LASDM for downscaling and bias correction against convolutional U‑Net, a Swin Transformer–based model, and a statistical baseline based on empirical quantile mapping over the Great Lakes region of North America. Results show that LASDM consistently outperforms these benchmarks across both typical and extreme SWE regimes, while maintaining substantially lower computational complexity. The case studies of two winter storms further demonstrates model performance. Remaining challenges include correcting topographic and snow melt-related biases. These findings highlight the potential of lightweight hybrid CNN-transformer architectures for efficient downscaling and bias correction tasks.
Metagenomic insights into microbial diversity and multivariate water quality characte...
huimei wang
wen hu

huimei wang

and 5 more

March 20, 2026
This study reveals the interaction between anthropogenic pressures and stratified microbial ecology in Fuxian Lake (China’s largest deep freshwater reservoir with Class I water quality). By integrating stratification indicators (TN, TP, COD, DO) from surface water with metagenomic sequencing data from surface (0–5 m), mid-depth (20–50 m), and deep-water (50–120 m) zones, we demonstrate how nitrogen loads from tourism activities in the northern shoreline zone trigger genomic adaptive changes in benthic microbial communities. In the coastal zone, dissolved nitrogen (accounting for over 53% of total dissolved nitrogen) drove genomic expansion alongside the spread of tetracycline resistance genes. Conversely, the central deep-water basin exhibited archaeal dominance due to total phosphorus enrichment (0.039 mg/L), showing significantly enhanced genomic integrity and surging alkaline phosphatase expression. This accelerated sediment phosphorus cycling rates to 2.7 times those in surface waters. Microbial communities exhibit depth-stratified functional specialization: (1) Tourism-impaired surface layers (0-5 m) display depressed diversity; (2) Metalimnetic zones (20-50 m) demonstrate ’edge-rich, center-cohesive’ structural continuity (elevated N50/N90 contig metrics centrally); (3) The hypolimnion (50-120 m) harbors peak biodiversity (Shannon H↑), biomass (contig counts↑), and co-occurring resistomes (sul1-intI1; Mantel r = 0.85). Metagenomic functional profiling further reveals dominance of carbohydrate (35%) and amino acid metabolism (43%), with central hypolimnetic communities uniquely enriched in CE11 carbohydrate esterases driving carbon mineralization. We propose stratified interventions: (1) enhanced wastewater treatment/wetlands to curtail nitrogen-driven genomic expansion in northern zones; (2) oxygen nano-bubble injection to disrupt archaeal P-regeneration in profundal layers. This study redefines deep lakes as microbial evolutionary theaters and establishes a framework for ecosystem-sensitive restoration.
Cultural Intelligence: Foundations and Key Practices to Work Across Diverse & Glo...
Ana Cristina Vasquez

Ana Cristina Vasquez

and 1 more

March 23, 2026
Have you ever felt that a technical idea was misunderstood because of cultural differences? Have you found it difficult to collaborate with colleagues from other countries or professional backgrounds? In this hands-on seminar, we will explore how cultural intelligence can make a difference in global technical and scientific teams. We will move beyond the concept of cultural intelligence as a “soft skill” and frame it instead as a key capability for effective collaboration. Through real-world examples and practical tools, you will learn how to Enhance intercultural communication by focusing on intentions and attitudes beyond words, Avoid misunderstandings in technical contexts by adopting a learner mindset and seeking clarification, andFoster more inclusive work environments through self-awareness, respect, and a practical toolkit for navigating cultural differences. Here we offer a practical toolkit centered on curiosity, humility, and self-awareness, advising attendees to research cultural contexts beforehand, adopt a learner mindset, and reframe judgments from “this is illogical” to “this is different.” Specific tips for academia include over-communicating with concrete examples, focusing on intentions over exact wording, observing before concluding, and avoiding personal interactions. Ultimately, cultural intelligence is framed not as mastering every culture but as approaching differences with respect, curiosity, and adaptability through small, human moments that shape global collaboration. Participants will gain actionable strategies in academic and professional settings, including how to overcommunicate examples, navigate hierarchy and social norms, and transform potential friction into opportunities for mutual understanding.
Pharmacological targeting of cholesterol biosynthesis in cancer: sterol intermediates...

Philippe De Médina

and 4 more

March 20, 2026
A document by Marc Poirot. Click on the document to view its contents.
Quantum-Resilient Decentralized Identity: A Framework for Future-Proof Digital Trust
Mohammad Tawrid Hyder

Mohammad Tawrid Hyder

March 20, 2026
Proliferation of digital interactions has underscored the critical need for robust, privacy-preserving, and secure identity management systems. Traditional centralized identity models, while prevalent, are increasingly vulnerable to data breaches and identity theft, and face an existential threat from the advent of quantum computing. This paper proposes a Quantum-Resilient Decentralized Identity (QR-DID) platform, an innovative framework that integrates Decentralized Identifiers (DIDs) with Post-Quantum Cryptography (PQC), blockchain technology, and AI-driven behavioral analytics. The QR-DID platform empowers users with self-sovereign control over their digital identities and verifiable credentials, ensuring long-term security against quantum attacks, enhancing privacy through selective disclosure, and providing adaptive, real-time threat detection. We detail the architectural design, the synergistic integration of these advanced technologies, and the end-to-end operational flow, demonstrating a pathway to a more secure, private, and resilient digital trust ecosystem.
Interfacial Engineering of COF/MXene Heterostructures: Towards Synergistic Multifunct...
Kunli Xu
AKang Dan

Kunli Xu

and 5 more

March 20, 2026
With the escalating global energy crisis and environmental pollution, the development of advanced materials combining high activity and stability has emerged as a critical research frontier. Two-dimensional transition metal carbides/nitrides (MXenes) and covalent organic frameworks (COFs) have attracted immense attention for their unique physicochemical properties; however, their practical implementation as individual components is hindered by the restacking and oxidation tendencies of MXenes, and the intrinsic low conductivity and poor processability of COFs. This review provides a comprehensive overview of recent advancements in COF/MXene heterostructures, highlighting how precise interface engineering strategies-such as electrostatic self-assembly, in-situ growth, and covalent bonding-achieve synergistic structural and functional enhancements. The resulting conductive skeleton-porous sieving architecture not only effectively resolves the trade-off between conductivity and active site accessibility but also significantly improves mechanical flexibility and environmental stability. Furthermore, the article systematically summarizes the broad applications of these composites across emerging fields, including electrocatalytic energy conversion, electrochemical energy storage, environmental remediation, intelligent biosensing, precision membrane separation, and biomedicine. Finally, the challenges regarding scalable fabrication and long-term stability are critically analyzed, offering perspectives on future research directions.
The chemokine receptor-like fourth extracellular loop of the apelin receptor differen...
Chloé Bonef
Hela Bouhsine

Chloé Bonef

and 11 more

March 20, 2026
Background and Purpose Apelin and Elabela are endogenous ligands of the apelin receptor (ApelinR/APJ), a GPCR involved in cardiovascular regulation and body fluid homeostasis. Human contains a conserved disulfide bridge linking the N-terminal domain to extracellular loop 3 (ECL3) via Cys19 and Cys281, forming a structural constraint analogous to a fourth extracellular loop (“ECL4”). In related GPCRs of the chemokine receptor family, this motif plays a critical role in ligand engagement and receptor activation. Experimental Approach The functional role of this disulfide bridge was investigated using site-directed mutagenesis (Cys19Ala, Cys281Ala), plasmon waveguide resonance binding assays, and BRET-based biosensors to monitor Gαi activation and β-arrestin-2 recruitment. The impact on endogenous ligands and antibodies targeting ApelinR was evaluated. Key Results Disruption of the disulfide bridge did not affect receptor surface expression but markedly impaired binding of apelin fragments (pE13F, K17F), leading to reduced Gαi signalling and β-arrestin-2 recruitment. In contrast, binding of the Elabela fragment K22P and the subsequent ApelinR signalling activation were largely preserved. Antagonist antibodies (JN241, JN241-Fc) retained activity, whereas the agonist antibody JN241-9-Fc showed strongly reduced efficacy for the mutated receptors. Conclusion and Implications The N-terminal/ECL3 disulfide bridge is a critical structural determinant for apelin binding and ApelinR activation but is less critical for Elabela signalling, supporting distinct modes of ligand engagement. Antibody-mediated agonism is particularly sensitive to alterations of « ECL4 » without affecting its binding, suggesting that this structural constraint is required for optimal ApelinR activation. Our results further demonstrate that ApelinR shares functional similarities with the chemokine receptors with respect to the role of this « ECL4 » domain.
Adaptive strategies for biodiversity monitoring integrating Indigenous ecological cal...
Orlando Acevedo-Charry
Gloria Rivera-Velasco

Orlando Acevedo-Charry

and 16 more

March 20, 2026
Integrating Indigenous and Western ecological knowledge can strengthen understanding of phenological patterns, yet this integration is often constrained by epistemological differences, power asymmetries, and histories of exclusion. We evaluated the potential of the Two-Eye Seeing guiding principle, which integrates the strengths of both knowledge systems, in ongoing biocultural conservation dialogues among local ethnographic leaders, Indigenous scholars, and Western researchers in ecology and anthropology. To identify similarities and differences between the two knowledge systems for success integration, we compared temporal patterns of bird species richness, composition, and relative abundance derived from a culturally embedded phenological device – an ecological calendar developed by the Pamiwã people – and from a community science initiative (eBird data contributed primarily by visiting avitourists). Our comparison revealed both convergence and divergence: phenological patterns in species richness and relative abundance for culturally important birds were broadly similar, whereas species composition varied across the year. These differences reflect distinct experiential approaches to observation. Indigenous observers ground their knowledge in culturally embedded experience and attention to environmental change, resource use, and bird behavior, whereas avitourists focus on personal encounters with more and rarer birds. Biocultural conservation requires integrative methods that are subject to interpretation and uncertainty. Acting as researchers and translators across disciplines, we demonstrate the opportunity to apply the Two-Eye Seeing guiding principle to integrate Indigenous and Western knowledges. This integration provides a tool for long-term monitoring that aligns global platforms such as eBird with local priorities for safeguarding biodiversity and cultural, ancestral, and spiritual values. Our findings highlight the value of iterative ecological calendars, framed by the Two-Eye Seeing guiding principle, as adaptive strategies for local communities to monitor biodiversity and manage their territories.
Spectral Universality Classes in One-Dimensional Fractal Geometries: From Hierarchica...
Enrique Vidal Silvente

Enrique Vidal Silvente

March 20, 2026
We investigate the spectral properties of fractional Laplacians H = L 0.9 defined on several one-dimensional fractal point sets, including Cantor-type hierarchical geometries and subdivision-based fractals such as the Sierpiński and Koch constructions. Using a unified numerical pipeline combining polynomial unfolding, bootstrap-resolved gap statistics, spectral dimension estimation, Brody index analysis, gap-ratio universality, number variance, inverse participation ratios, Rényi entropies, multifractal dimensions, and robustness-to-noise tests, we identify two sharply distinct spectral universality classes. Hierarchical fractals (Cantor and Random Cantor) converge to a Poisson-type spectral regime, with spectral dimension d s ≈ 2, vanishing Brody index β → 0, and gap ratio ⟨r⟩ approaching the Poisson limit. Subdivision fractals (Sierpiński and Koch) exhibit instead a rigid, non-RMT spectral class characterized by extreme level repulsion (⟨r⟩ → 1), Brody index β → 1, low number variance, and multifractal gap statistics, while remaining incompatible with both Poisson and Wigner-Dyson distributions. Scaling across refinement levels (3 ≤ L ≤ 7) reveals stable convergence of all indicators, demonstrating that the linear/integrable Poisson regime is contained as a limiting case within the broader class of hierarchical fractal spectra, whereas subdivision fractals generate a genuinely new universality class with no analogue in classical random matrix theory. These results provide the first systematic classification of spectral universality in one-dimensional fractal geometries and establish a quantitative bridge between geometry, spectral dimension, and level statistics in non-smooth metric spaces.
Multi-path grounding of high-speed maglev train struck by lightning
jianqiong zhang
xiangqiang Li

jianqiong zhang

and 2 more

March 20, 2026
Maglev train has no contact with rail in high-speed operation, which is significantly different from traditional wheel-rail train. A crucial issue of the lightning overvoltage on maglev train is grounding characteristics, however, which has not been fully and thoroughly investigated. In this letter, the multi-path grounding of maglev train is analyzed. A new and advantageous circuit model for researching the grounding characteristics and impacts is demonstrated. And, an experiment of lightning injection has been designed to validate the proposed model.
GALAR-TemporalNet: Anatomy-Guided Temporal Multi-Label Classification with Bidirectio...
Jiye Won

Jiye Won

and 2 more

March 20, 2026
A document by Jiye Won. Click on the document to view its contents.
Performance of the FMF First-Trimester Preeclampsia Screening Model and Aspirin Proph...
Bui Minh Cuong
Truong Huu Cuong

Bui Minh Cuong

and 5 more

March 20, 2026
Background: Early identification of women at high risk of preeclampsia allows timely preventive interventions, particularly aspirin prophylaxis. The first-trimester screening model developed by the Fetal Medicine Foundation (FMF) combines maternal characteristics with biophysical and biochemical markers to improve prediction accuracy. Objective: To evaluate the performance of the FMF first-trimester screening model for preeclampsia and to assess pregnancy outcomes following aspirin prophylaxis in high-risk women. Methods: This prospective cohort study included 1,187 singleton pregnancies at 11 weeks 3 days to 13 weeks 6 days of gestation who underwent first-trimester preeclampsia screening at Quang Ninh Obstetrics and Pediatrics Hospital, Vietnam, between November 2023 and November 2025. Screening was performed using the FMF algorithm incorporating maternal characteristics, mean arterial pressure (MAP), uterine artery pulsatility index (UtA-PI), and placental growth factor (PlGF). Women identified as high risk received aspirin prophylaxis. Pregnancy and neonatal outcomes were subsequently analyzed. Results: Among 1,187 screened pregnancies, 84 women (7.1%) were classified as high risk for preeclampsia. Each 1-mmHg increase in MAP was associated with an approximately 12.5% increase in the risk of preeclampsia, while each 0.1-unit increase in UtA-PI increased the risk by approximately 27.8%. Conversely, each 1 pg/mL decrease in PlGF was associated with a 7.0% increase in risk. The overall incidence of preeclampsia was 0.8% (9/1,187), with 66.7% of cases occurring in the high-risk group. No cases of early-onset preeclampsia (<34 weeks) were observed, and most cases occurred at ≥37 weeks and were non-severe. Women who developed preeclampsia had lower gestational age at delivery and lower neonatal birth weight. Intrauterine growth restriction and neonatal intensive care unit admission were more frequent among pregnancies complicated by preeclampsia and among those classified as high risk. Conclusion: First-trimester screening using the FMF model combined with early aspirin prophylaxis in high-risk women was associated with favorable maternal and neonatal outcomes. This strategy appears feasible and may be suitable for broader implementation in routine obstetric care in Vietnam.
“Cervical length surveillance for predicting spontaneous preterm birth in women with...
Nikit Kadam
Sanaria Raouf

Nikit Kadam

and 3 more

March 20, 2026
Background – The value of transvaginal cervical length surveillance for predicting spontaneous preterm birth (sPTB) in women with congenital uterine anomalies (CUA) is unclear. Objective - A systematic review and diagnostic accuracy meta-analysis of cervical length surveillance for predicting sPTB in women with CUA Search strategy – A literature review of MEDLINE, EMBASE, and the Cochrane Library from inception to December 2025. Selection criteria - Observational studies evaluating 2 nd trimester cervical length measurement in pregnant women with CUAs. Data collection and analysis - Review Manager and Meta-DiSc, with pooled odds ratios (ORs), sensitivity, specificity, likelihood ratios, and SROC curves generated using a random-effects model. Results – 5 studies including 624 women met the inclusion criteria. sPTB occurred in 52% with a short cervix versus 13.2% with normal cervical length (OR 8.05, [4.52–14.35]). A short cervix increased the risk of sPTB before 37 weeks (OR 6.68, [3.68–12.11]) and before 35 weeks (OR 28.16, [5.64–140.48]). The association was strongest for bicornuate uterus (OR 40.98, [4.27–393.10]), though estimates were imprecise. At a 25-mm cervical length threshold, sensitivity was low (0.35, [0.23 - 0.49]), specificity was high (0.94, [0.91 - 0.96]), with a LR+ ratio of 5.51. A 30 mm threshold improved sensitivity (1.00, [0 – 1.00]) but reduced specificity (0.84, [0.67 - 0.94]), reflecting sparse data. The SROC curve indicated moderate-to-good discriminatory performance. Conclusion – Cervical length measurement yields clinically useful risk stratification, especially when a short cervix is present. At the conventional 25-mm threshold, high specificity but limited sensitivity makes it better for confirming than excluding risk. Higher thresholds may improve detection, but the optimal cut-offs are still uncertain.
השפעת עמדות כלפי אסלאם על ביצועי צוותים בארגונים מעורבים דתית
Tomer Menachem

Tomer Menachem

and 1 more

March 20, 2026
A document by Tomer Menachem. Click on the document to view its contents.
Extreme basal heat flow and presumptive subglacial thermal springs in Northeast Green...
William T Colgan

William T Colgan

and 11 more

March 20, 2026
A document by William T Colgan. Click on the document to view its contents.
Long-term trend of the White Stork (Ciconia ciconia) in Tver Region based on a nest-s...
Andrei Zinoviev

Andrei Zinoviev

March 20, 2026
Citizen-science datasets increasingly support long-term assessments of bird populations, but their value depends on how effectively historical nest inventories can be integrated with contemporary volunteer monitoring. I reconstruct the long-term trajectory of the White Stork Ciconia ciconia in Tver Region, western Russia, using three complementary data sources: historical publications, the registered Tver State University nest-site database and associated ArcGIS web application, and volunteer-based breeding observations collected between 2008 and 2016. By the mid-2010s, the regional database documented 432 known nest sites across 31 districts, compared with 194 nests reported for 1998. This increase indicates substantial growth in the known breeding network, although part of it reflects improved detection and reporting. Nest distribution was strongly concentrated in the south-west of the Region, while the north-east remained under-surveyed. Artificial substrates dominated the nesting population: 69.6% of known nests were on water towers and 8.2% on power-line poles, whereas only 2.9% were on trees. Breeding monitoring revealed marked interannual variation. In 2014, 68 of 70 monitored pairs bred successfully and produced 205 young (2.93 young per breeding pair), whereas in 2015 only 40 of 63 pairs were successful, producing 80 young in total (1.27 young per breeding pair). Productivity improved again in 2016 among nests with reliable brood counts. The combined evidence supports a long-term expansion and consolidation of the White Stork population in Tver Region, but also shows that uneven observer coverage limits formal year-to-year trend modelling. The regional database is already sufficient for robust spatial and historical inference and could become a fully quantitative monitoring system once annual coverage and validation are standardized. eBird broadened temporal coverage, and iNaturalist improved spatial inference and recent validation, but neither dataset was sufficient to replace directed nest monitoring for annual breeding-success analysis.
Parallel Node-Breaker Based Transient Stability Simulations Using Multi-Area Thevenin...
Vibhuti Sahu

Vibhuti Sahu

and 1 more

March 20, 2026
This paper indroduces space-parallel transient stability simulation (TSS) solver tailored for large-scale Node-Breaker (NB) power system models using the Multi-Area Thevenin Equivalents (MATE) framework. The proposed approach features a distributed Sparse Tableau Approach (STA) enabling efficient breaker-level substation modeling without explicit topology processing, development of MATE equations for NB models, mapping matrices to formulate MATE based network equations for NB models and a weighted graph partitioning technique to ensure balanced parallel workloads. By exploiting both space and task parallelism, the solver achieves time efficient performance even with NB model matrices up to 50 times larger than their Bus-Branch (BB) counterparts. Extensive tests on the Polish 2383-bus, Pegase 9241-bus, and Pegase 13651-bus systems, each configured with Breaker-and-Half schemes, demonstrate execution times of only 2.8s ∼ 3.3s, 23.5s ∼ 28s, 30s ∼ 34.6s respectively for a 10s run on a 64-core, 2.2 GHz, 256 GB RAM compute node. These results establish the proposed MATE-NB solver as a scalable, high-fidelity, and computationally efficient solution for TSS of large-scale NB networks.
AROMA: Adaptive Orchestration for Robust and Cost-Efficient Multi-Agent LLM Systems
Tianyu Yin

Tianyu Yin

and 1 more

March 20, 2026
Multi-Agent Large Language Model Systems (MAS) confront significant challenges stemming from systemic design flaws, inter-agent misalignments, and prohibitive operational costs. These systems frequently offer only modest performance gains, or even exhibit setbacks, while incurring substantial increases in token consumption due to prevalent failure modes like improper task decomposition and information overload. To address these critical issues, we propose Adaptive Orchestration for Robust Multi-Agent LLM Systems (AROMA), a novel framework designed to dynamically perceive, diagnose, and adaptively orchestrate multi-agent collaboration. AROMA incorporates capabilities for real-time failure identification, intelligent adjustment of system parameters, roles, and communication strategies, and optimization for efficient task completion with minimal overhead. Through extensive experiments on complex benchmarks, AROMA demonstrates enhanced task success rates and a substantial reduction in average token cost compared to existing baselines, alongside a significant mitigation of collaboration failure modes. Our findings confirm AROMA's efficacy in improving robustness, efficiency, and generalizability, paving the way for more reliable and economically sustainable multi-agent LLM deployments.
“In Defense of the Language Mavens.”
Kevin Reilly

Kevin Reilly

March 20, 2026
Language, Reason, and the Costs of Abandoning StandardsI have been doing some reading in recent weeks of one of my favorite authors, Stephen Pinker, specifically a work, which has become something of a classic, published some time ago, The Language Instinct .1  In this fascinating work, Pinker includes a discussion of certain strands within contemporary language studies that advocate treating nonstandard dialects as pedagogically equivalent to standard academic language in analytic and institutional contexts. Increasingly, the trend to “deprioritize” the rules of the so-called “hegemonic” idiom has spread not only in the academic literature (especially in discourse within the “critical studies” community), but also into elementary and secondary educational contexts, where traditional instruction gives way to curricula which eschew formal semantics, structural rigor, analytical grammar and attention to conventions of standard speech. These are, in the view of many, tedious and “prescriptivist,” if not altogether oppressive, in effect. The implications of this trend bear attention.The desire for inclusivity, and a respect for the identity, and the legitimacy, of non-standard dialects has served an important corrective function. It has countered naïve assumptions about intelligence, worth, or expressive capacity based on the accent of speakers (particularly, within an educational context, members of traditionally “disadvantaged” or marginalized communities), or their native idiom, and they have rightly underscored that all natural languages, and dialects, as tools for the conveyance of meaning, are systematic, rule-governed, and capable of sustaining coherence within their speech communities. The aforementioned scholarship, among critical theorists, repetitively makes this case, and often advocates for the “disruption” of normative practice in prioritizing standard language in education. While many proponents of critical language awareness advocate additive models (combining standard instruction with critical reflection), there remains a growing pedagogical tendency—particularly in some applied contexts—to de-emphasize explicit instruction in standard forms. Alongside the salutary recognition of the potential of nonstandard idioms to convey meaning, the advocates of disruption and those who decentralize standard language – both in the study of the conventions of semantics, grammar, syntax and literature of the so-called hegemonic culture, and even (increasingly) in second language studies – are encouraged to develop in students “critical language awareness.” This means they must analyze how standard language ideologies function to maintain power, reproduce social inequality, and marginalize non-standard speakers. Standard language in this view should be depreciated, and “traditional” tenets of rigor and formality loosened, if not abandoned.2This, in my view, is a troubling tendency. A growing demand within the academic community that teachers should disparage the role of standard language as a distinct and indispensable instrument of analytic reasoning, education, and public discourse, and should ignore or relax altogether the requirements of standard language within those domains, comes at a genuine intellectual cost. This trend derives from a conflation of two claims that ought to be kept rigorously distinct: 1) that non-standard dialects are linguistically legitimate, and 2) that they are equally suited to the purposes of rigorous analysis, formal exposition, and institutional persuasion. The first claim is well supported; the second is not. Linguistic legitimacy concerns internal rule-governed structure. Analytic suitability concerns the external demands of institutional reasoning. These are different evaluative criteria.The function of standard language is not merely social or aesthetic. It is epistemic. Standard language, and its concomitant conventions, are a deliberately constrained linguistic instrument, refined over time to facilitate explicit reasoning among large, heterogeneous populations lacking shared background assumptions. Their norms - precision of reference, explicit premising, linear argumentation, controlled affect, and stable semantics—are not simply useless, arbitrary conventions imposed by cultural elites.They are, ultimately, technologies of clarity. Like any technology, they are historically contingent and socially distributed — but their function is instrumental rather than symbolic. Their value lies not in who first wielded them, but in what they enable: durable inferenceacross differences.This point becomes clearer when we consider the nature of analytic reasoning itself. Formal and informal logic alike require that premises be identifiable, that terms be defined and used consistently, that inferential steps be traceable, and that conclusions follow recognizably from what precedes them. These requirements impose significant cognitive demands, particularly on novice thinkers – such as students in elementary school and in secondary education. Language varieties or discourse registers that tolerate imprecision, ambiguity, shifting reference, lack of logical parallelism, ellipsis, emotional compression or implicit premises, and which rely heavily on coterie-defined metaphorical, metonymical and other figurative devices may function admirably in high-context, oral, or in-group settings. But when such registers are imported, unaltered, into analytic contexts, they tend to obscure inferential structure and permit, rather than resist, intellectual shortcuts.This observation should not be misunderstood as a claim about the intellectual capacities of speakers. The capacity to reason analytically is not distributed along linguistic lines. Rather, the issue is whether a linguistic system - or, more precisely, a set of linguistic norms - enforces the disciplines that analytic reasoning requires. Standard academic language, as institutionally codified, has evolved to externalize these disciplines; informal and high-context registers are typically not structured with those analytic demands in mind. What is permitted is frequently exploited, particularly by students still learning how to discipline their thought, to the detriment of cogency. The advantage of standard language is not structural superiority, but the institutionalization of constraints that externalize logical discipline.Consider, for example, features often cited in discussions of non-standard speech: double negatives, emotionally charged vocabulary, topic drift, or reliance on shared contextual assumptions. Within the speech communities in which these features are normative, they are not logically defective; they are, indeed, rule-governed and intelligible. But they are poorly aligned with analytic goals. Double negation, while unambiguous within a dialect, can complicate scope in formal reasoning. It takes the rigidity of formal logic, the agreement of the linguistic community as to application of meaning within the norms of “standard” language to resolve this. This is the case for example in Iberian languages, Langue d’oc , Langue d’oil and Italo Romance. In these contexts, the double negative construction and, for example, dative redundancy are standardized as to their interpretation within a closely framed set of linguistic applications. Emotional loading, meanwhile, collapses descriptive rigor into subjectivism. Topic drift, imprecise reference and highly figurative language often obfuscate reasoning and undermine argumentative coherence. Implicit premises and non sequitur render reasoning opaque for readers not already inclined to agree. Along with a failure to rigorously define conventions of semantics, grammar, syntax and punctuation, as well as spelling and pronunciation, these features may not make analytic reasoning entirely impossible. They do, however, make it more difficult, and far more subject to misunderstanding and misinterpretation. They detract from rigorous discourse, they do not assist it. And, in educational settings and communication requiring rigor, assistance matters.Standard language functions as a kind of intellectual scaffolding. Its stylistic constraints externalize cognitive discipline. They force writers to slow down, to specify what they mean, to anticipate objections, and to make inferential commitments explicit. This is extremely important in academic contexts, in the public domain, and in business. In this sense, standard language “carries the water” for analytic thought not because it is inherently more logical, but because it institutionalizes habits of clarity. To abandon those habits in the name of linguistic equivalence is a major category error. It confuses respect of language evolutionary realities and cultural diversity with utility and rigor. Castilian was once the vulgar language of disenfranchised peasants within the complex of the pax romana. Over millennia, it evolved into a highly logical discursive language spectacularly suited for analytical rigor. Standard language, agreed upon by all, fulfills this highly utilitarian function as well as aesthetic objectives. It does so by default; non-standard and informal registers generally do not - much as Basil Bernstein’s distinction between restricted (high-context, assumption-laden) and elaborated (explicit, low-context) codes – which highlights the institutional mismatch when schools deprecate traditional standards or allow expression for heterogeneous consumption exclusively in high context idioms.3The consequences of this confusion in the educational domain and the acceptance, indeed the encouragement, of nonstandard language are increasingly visible. Students are often encouraged to view demands for standard language as matters of taste. Worse, the teaching of standard language is framed – especially within the so-called “critical” language studies community – through an academic perspective that interprets language primarily through the lens of power, domination, and resistance. In this view, standard language is disparaged as the language of the hegemon, the colonialist. It must cede its place to the patois, the languages and dialectical peculiarities of the so-called disenfranchised. Language study and language teaching must no longer prioritize the “standard” language of power elites. Rather, enlightened language theory must be, to use the term of consensus, “disruptive”.In reading such statements, I rarely understand, specifically, what harms are being remediated and what, other than the value of clarity, coherence and logical discourse, is being disrupted. This iconoclasm is seen somehow as a meritorious or virtuous redress of historical injustices which must be corrected. It seems that among such theoreticians, little thought is given to the fact that standard language – because it is the language of the hegemon – has evolved precisely to provide tools for rational thought – as such rational discourse, and its logic, provides the framework for advancement of the social, scientific and artistic achievements of the dominant culture. Institutional centers of power have historically codified linguistic norms that support the forms of reasoning valued within those institutions (reasoning construed as “effective” for achieving the utilitarian objectives of their societies and cultural norms upon which the reigning societal edifices are erected). Instead of disparaging the teaching of the language which fosters access to these norms – including not only the logic, the grammar, the syntax and agreed upon semantics of the hegemonic dialect, but also the indispensable linguistic niceties which constitute the phatic web within which all productive discourse takes place (is experienced as inviting, nonthreatening and conducive to dialogue) –, some pedagogical approaches, in seeking to resist linguistic hierarchy, actually end limiting students’ access to the linguistic tools most rewarded by institutions. In so doing, I suggest they are being cognitively dissonant – they are subverting their own implied desire to provide access to power to those whom they encourage to disparage standard forms. Research in educational sociology and writing studies suggests that explicit instruction in elaborated codes improves students’ ability to produce decontextualized, analytically structured arguments.4,5In today‘s public discourse, including political discourse, regrettably there is a trend increasingly to favor immediacy and affect over argument. This can be seen in both social and legacy media. Commentary frequently substitutes assertion for analysis, confident that expressive force will compensate for logical thinness. In such an environment, the erosion of linguistic standards is not merely a cultural shift; it is an epistemic one.To insist on standard language in analytic domains and in the public discourse, and to require students to learn and to wield it effectively, and with rigor, is not to denigrate non-standard dialects or to deny their expressive richness. It is simply to recognize that different communicative goals require different linguistic disciplines. It is a recognition that, in fact, not all registers are interchangeable. We do not accuse mathematics of elitism because it demands symbolic precision, nor chemistry because it requires a specialized vocabulary. We understand that certain forms of inquiry demand particular tools. Language is no different.This recognition carries clear implications for education, academia, and public life. Students should be taught—explicitly and unapologetically—that mastery of standard language is not a betrayal of identity but an acquisition of intellectual power and access to what Lisa Delpit calls “the language of power”.5   Academic institutions should resist the temptation to relax linguistic standards under the mistaken belief that rigor is exclusionary. Public figures, especially those who aspire to persuade across differences, should model the disciplined use of language appropriate to analytic, business and civic reasoning.   It is important to emphasize that nothing in this argument entails that nonstandard dialects are cognitively deficient or aesthetically inferior. Nor does it deny that standard language carries some historical associations with exclusion. The question at issue is not dignity, but function. Educational institutions must decide which linguistic norms best serve the development of transferable analytic competence.   While analytic reasoning is in principle independent of linguistic form, the absence of externalized constraints increases cognitive load, particularly for novice thinkers. None of this requires suppressing dialects, policing informal speech, or denying the legitimacy of linguistic variation in its proper contexts. It requires only the courage to say what was once taken for granted: that clarity is not oppressive , and precision is not prejudice. Recognizing the functional value of standard language does not require denying expressive richness or cultural legitimacy in other forms. I think we are better served with a stance not advocating replacement, but rather breadth of repertoire. This allows for expressive richness while recalling that a shared standard language remains one of the most powerful instruments we possess for thinking together in common.   NOTES1. Stephen Pinker, The Language Instinct: How the Mind Creates Language, 9th ed, Harper Perennial, 2007.2. Examples in the literature are myriad. A few: The new Journal from the U of Pennsylvania, Racial Justice in Multilingual Education(RJME), the first edition of which was published in August of 2025; The “Standard Language Ideology Statement”, published by the the Department of Linguistics at the University of Michigan, July 2021:   https://lsa.umich.edu/linguistics/about-us/values-statement/standard-language-ideology-statement.In this latter, one reads the following: “Linguists do not support the widely held assumption that there is a standard language that should be adopted by all, and our department condemns penalties that come with not using such language. Standard language ideology is a construct that establishes a hierarchy between varieties. It misleads language users into believing that some varieties are better than others and can perpetuate harmful patterns of linguistic discrimination - discrimination that is often a proxy for ethnic, gender, class, and regional discrimination.”3. Basil Bernstein, Class, Codes and Control, Vol. 1 (Routledge, 1971); see also his 1962 article in American Anthropologist. Bernstein argues convincingly that elaborated codes are essential for formal reasoning and institutional success, and failing to teach them disadvantages students structurally, independent of any deficit in the restricted code itself.4. Pierre Bourdieu, Language and Symbolic Power, ed. John B. Thompson (Harvard University Press, 1991), esp. ”The Economics of Linguistic Exchanges” and ”Authorized Language.” Bourdieu argues that refusing to teach the dominant language in the name of equality deprives repressed groups of instruments for social mobility and institutional participation.5. Lisa Delpit, ”Skills and Other Dilemmas of a Progressive Black Educator,” Harvard Educational Review (1986); see also, Lisa Delpit, “The Silenced Dialogue: Power and Pedagogy in Educating Other People’s Children,” Harvard Educational Review 58:280–298 (1988). Delpit argues that explicit instruction in the ”language of power” (standard/edited English) is an ethical obligation, especially for marginalized students, and that withholding it is paternalistic rather than liberatory. This complements pedagogical work on writing as cognitive scaffolding, such as Flower and Hayes, “A Cognitive Process Theory of Writing.” College Composition and Communication, 1981.
Emergent Cooperative Dynamics and Causal Treatment Effects in Large Language Model Mu...
Vikas Ramachandra

Vikas Ramachandra

March 20, 2026
Emergent Cooperative Dynamics and Causal Treatment Effects in Large Language Model Multi-Agent Ecosystems
← Previous 1 2 … 10 11 12 13 14 15 16 17 18 … 2754 2755 Next →

| Powered by Authorea.com

  • Home