Abstract: Digital platforms produce a paradox: unprecedented connectivity alongside rising loneliness. Existing frameworks—built on assumptions of human senders and receivers—cannot explain this because the structure of communication has fundamentally changed.This paper introduces directionality as a formal variable tracing platform evolution from bidirectional social graphs, through unidirectional interest graphs, to Zero-Directionality: human–machine interaction in which the social other is entirely absent.We show that this zero-degree threshold enables two divergent trajectories. In the Inverted Loop (negative directionality), algorithms act preemptively, shifting the human from operator to operand. In the Triadic Mesh (triadic directionality), AI mediates between humans rather than replacing them, preserving human connection.Drawing on platform analysis, we examine how these trajectories reconfigure agency, citizenship, and social life, and identify degradation risks when mediation drifts into substitution. The framework extends platform studies to environments where the machine is communicative agent rather than intermediary.

Abstract: We present a parametrised charging infrastructure model developed to support the design of a hybrid-electric zero-emission vessel with corresponding charging infrastructure for operation along the Norwegian Coastal Express route. The charging model includes functionalities to analyse the required battery storage capacity and power ratings and locations of charging facilities for achieving battery-electric operation. We demonstrate the use of the charging model to analyse different zero-emission scenarios for the Norwegian Coastal Express route. In the presented example scenarios, the model takes as input the estimated energy demand for a new zero-emission vessel design for the Coastal Express in different weather conditions, and includes functionality to consider realistic port stays based on existing timetables and historical data of delays. The analyses show minimal required battery and illustrate a trade-off between charging power and battery capacity, as well as exemplifying the impact of different timetables as well as historic deviations on charging and energy delivered from the battery. The charging model presented is general and can be used for other routes than the Norwegian Coastal Express, as a tool for decision-makers to optimize for battery-electric operation whilst keeping the need for onboard storage capacity and charging infrastructure installations at a minimum.

Abstract: This article examines the impact of intermodal wagon technical specifica-tions and railway infrastructure parameters on electricity consumption in rail freight transport. To conduct this investigation, a three-stage analytical model was developed. The first stage establishes core assumptions, encompassing train lengths, rolling stock types, container configurations, infrastructure constraints, and the characteristics of the energy-consumption model. The second stage identifies technical constraints of specific wagons, determines representative train compositions, and executes loading simulations. The third stage focuses on evaluating energy efficiency across diverse loading scenarios. The case study demonstrates that specific energy consumption varies significantly with wagon type, train mass, and route characteristics, challenging the use of static energy-consumption values prevalent in current literature. Results indicate that 40-foot wagons incur high energy penalties due to their tare weight and axle count, despite maximizing loading capacity. While 60-foot wagons consume less energy, they result in a high frequency of empty slots under a 20 t/axle limit. Conversely, 80-foot wagons emerge as the most energy-efficient, particularly at a 22.5 t/axle limit. Mixed consists offer a balance of operational flexibility and competitive performance. Inter-estingly, extending train length from 600 m to 730 m increases volume but does not inherently reduce unit energy consumption. These findings underscore the necessity of aligning wagon fleet selection with infrastructure capabilities and cargo characteris-tics. Ultimately, this study provides actionable recommendations for planning ener-gy-efficient intermodal operations.

Abstract: This study evaluated the seasonal variation (July vs. September) in the chemical and flavonoid composition of Mediterranean natural pasture used for grazing Cilentana goats and assessed its impact on fresh goat cheese quality. A total of 36 pasture samples and 60 cheese samples were analyzed for proximate composition, fiber fractions, minerals, and flavonoids (HPLC). July pastures showed higher dry matter (95.4%), ash (8.74%), and lower fiber (28.7%), whereas September pastures had higher crude protein (17.7%), fiber (43.7%), and kaempferol (36.4 mg/kg DM). Cheese produced in July had higher fat (46.8% DM), ash (8.36%), NaCl (4.78%), and dry matter, while September cheeses showed slightly lower fat (42.2%), higher moisture, and a more acidic pH (5.90). Seasonal increases in kaempferol, lutein, quercetin, and CMG suggest enhanced antioxidant potential in late‐season forage. These findings highlight the importance of seasonal pasture monitoring to optimize feed quality and dairy product characteristics in extensive Mediterranean goat systems.

Abstract: Plant-based biomaterials are increasingly recognized as bio-instructive platforms capable of actively modulating immune responses rather than functioning solely as passive structural supports. In this context, the term plant-based is used operationally to denote photosynthetic biomass–derived platforms and includes both terrestrial plants and marine macroalgae, reflecting their shared richness in polysaccharides and secondary metabolites relevant to immune-engineering and regenerative medicine. Current evidence on plant-derived polysaccharides and phytochemicals is critically synthesized, including algal sulfated polysaccharides (fucoidan, alginate, carrageenan), terrestrial plant polysaccharides (e.g., Lycium barbarum and Aloe vera derivatives), and polyphenolic compounds, highlighting their roles as bioinstructive immunomodulators in biomedical contexts.Key immunoregulatory mechanisms are discussed, including macrophage polarization along an M1–M2 functional continuum, pattern-recognition receptor engagement, redox and metabolic regulation, and coordinated crosstalk between innate and adaptive immunity. Particular emphasis is placed on how material structure, molecular weight distribution, and chemical functionalization shape immune cell responses and downstream regenerative outcomes. Advanced delivery strategies, including polysaccharide-based hydrogels, nanocomposites, lipid-based phytosome formulations, and plant-derived extracellular vesicles (EVs), are reviewed as enabling technologies to enhance stability, bioavailability, and spatiotemporal control of plant-derived bioactives. Applications in wound, musculoskeletal, and bone regeneration are summarized with attention to tissue-specific immunological requirements. Key barriers to clinical translation are also addressed, including source variability, batch-to-batch reproducibility, establishment of structure–activity relationships, Good Manufacturing Practice (GMP) compliance, regulatory classification (medical device vs. drug vs. combination product), and ethical considerations related to sourcing and traditional knowledge. For clarity, extracellular vesicles (EVs) are used as an umbrella term encompassing heterogeneous vesicular subpopulations; the term “exosomes” is retained only when supported by subtype-specific characterization, as many studies report mixed EV preparations.

Abstract: Formal methods are software engineering approaches with a rigorous mathematical basis that can be used in helping to ensure the correctness of software systems, especially where safety or security is critical. Artificial Intelligence (AI) has developed very rapidly in the area of Generative AI (GenAI), where questions can be answered with increasingly impressive but with potentially unreliable and variable results. This paper surveys research in integrating the two approaches in a synergistic manner. Traditionally, such explorations have required significant manual efforts in searching for and evaluating existing research. However, most relevant publications are now accessible online, and AI tools are increasingly good at answering research questions with more and more reliability. This paper takes the approach of using GenAI to evaluate research questions on combining formal methods and AI-related techniques. The paper assesses the usefulness and validity of these results. With recent improvements in AI search tools, the approach is now a useful aid to researchers, significantly reducing the time needed to survey existing research, while always needing human checking by an expert. In addition, the combination of formal methods and AI approaches looks to be an interesting and beneficial research area with potential industrial-scale applications in the future.

Abstract: Background: Regulatory frameworks such as the Belmont Report, the Common Rule, and the Declaration of Helsinki require informed consent to ensure participants understand a study’s purpose and can make voluntary decisions about their involvement. Regulations including the General Data Protection Regulation (Regulation (EU) 2016/679) further emphasise that consent must be freely given and revocable without disadvantage. Although informed consent forms (ICFs) are intended to be clear and accessible, they have become increasingly lengthy and complex. Large language models (LLMs) offer potential to navigate and interpret this complexity and have shown promise in biomedical information extraction tasks. However, their susceptibility to hallucinations limits reliability in high stakes settings. Retrieval augmented generation (RAG) can mitigate such errors. This study evaluates the integration of LLMs with RAG for reviewing data reuse language in ICFs and their ability to interpret complex textual structures. Methods: Firstly, we processed 438 ICFs from different trials, including multi-countries, languages and versions of ICFs. Using expertly curated prompts, we extracted information about data reuse using GPT-4.1. Comparing the LLM-generated data reuse outputs with human expert ground truth, we evaluated accuracy and the time required to extract information for each ICF. To further validate the workflow, we evaluated an independent set of 488 ICFs spanning additional trials, languages, and regions. For this cohort, we assessed the correctness of LLM outputs along with the quality of supporting evidence provided by the model. Results: Across 438 ICFs, the system achieved 81.6% accuracy, which increased to 90% in a subsequent evaluation of additional 488 ICFs after prompt optimisation. Using a RAG-based approach, the system accurately extracted data reuse information across multiple languages and identified nuanced international regulatory requirements. Conclusion: This approach has the potential to significantly alleviate administrative burdens by automating labour-intensive processes, while also generating insights that could inform the standardisation of ICF creation. Ultimately, these advancements may contribute to reduce the complexity of ICFs, thereby improving their readability and comprehensibility for participants.

Abstract: Groundwater quantification is essential for sustainable water resources management, yet it is often hampered by limited data availability and difficulties in measuring spring discharges. This study investigates three carbonate aquifers in Central Italy’s Abruzzo region: the Genzana–Greco, Morrone, and Marsicano mountains. The aim is to resolve uncertainties in spring attribution, and groundwater flow patterns using isotopic analyses combined with field surveys. The Genzana–Greco aquifer was examined to clarify the sources of the Acquachiara spring and the previously unreported Germina spring, assessing whether recharge occurs locally or from the carbonate massif. In the Morrone mountain aquifer, discharge gains along the Pescara River through the Gole di Popoli were quantified, and spring isotopic compositions were compared to the main basal spring Giardino to better define groundwater contributions. For the Marsicano mountain aquifer, the role of Lake Scanno in feeding the Villalago springs was investigated through isotopic analysis of inflows, downstream springs, and basal aquifer discharge points to constrain the hydrogeological water budget. Overall, the integration of isotopic tracers with hydrological measurements allowed a more precise characterization of aquifer recharge areas, mean residence times, and groundwater flow paths, improving the understanding of regional water resources in a complex carbonate setting.

Abstract: For every prime p and every integer a, the backward finite difference δp(a) := a^p − (a − 1)^p equals the cyclotomic binary form Φp(a, a − 1), where Φp(X, Y) is the homogenisation of the p-th cyclotomic polynomial, and hence equals the norm NQ(ζp)/Q(a − ζp(a − 1)). For p = 3 this specialises to the identity δ3(a) = NZ[ω](a − ω(a − 1)), where ω = e^(2πi/3), connecting the individual cubic finite difference obtained by differencing the classical sum formula of Nicomachus of Gerasa (~100 CE) with the Eisenstein norm that appears in Euler's factorisation of a^3 + b^3. We develop this identity in three directions: (a) General cyclotomic framework. For each prime p, every prime divisor q of δp(a) satisfies q ≡ 1 (mod p), imposing an arithmetic sieve whose density ~1/(p−1) grows increasingly severe with p. (b) Arithmetic density. The values {δ3(a)}a≥1 form a thin subfamily of the Löschian numbers (norms in Z[ω]), with counting function ~√(N/3) versus the Landau-Ramanujan asymptotic CN/√log N for all Löschian numbers up to N. (c) Three-language equivalence. For the cubic case we prove a precise equivalence among: (i) divisibility of δ3(a), (ii) multiplicative order modulo q, and (iii) splitting of q in Z[ω]. We also give an elementary proof of the base case 1 + b^3 = c^3 (no positive-integer solutions), and derive 3-adic constraints on any hypothetical solution to a^3 + b^3 = c^3 via the Lifting-the-Exponent Lemma, without invoking unique factorisation in Z[ω].

Abstract: This study proposes SWEET-RL, a reinforcement learning framework for training LLM agents in multi-turn collaborative reasoning tasks involving human or agent partners. A step-wise critic is trained using intermediate evaluation signals derived from task progression rather than final answers. The method is evaluated on ColBench, consisting of 3,800 multi-turn collaboration sessions across software development and design tasks. SWEET-RL improves long-horizon task success rates by 24.3% and reduces dialogue-level error accumulation by 35.1%, demonstrating stronger robustness in extended collaborative interactions.

Abstract: A novel two–stage procedure for approximating solutions of nonlinear systems is introduced. The scheme employs two evaluations of the vector function F together with a single Jacobian computation, followed by the resolution of two linear subproblems that share an identical coefficient matrix. This structure reduces the computational burden and enhances the adaptability of the method with respect to existing alternatives. The design of the algorithm is motivated by criteria relating efficiency to the total number of functional evaluations, ensuring that the resulting strategy achieves the optimal convergence order permitted within this framework. A proof of the local convergence order is provided, and its accuracy is supported by a series of experiments on distinct nonlinear models, including problems arising from differential equations. The numerical evidence confirms that the developed technique reaches the theoretical convergence rate and performs favorably when compared with other methods of equal order. Moreover, we examine the dynamical features of the related parametric variant, offering additional understanding of its stability properties and iterative behavior.

Abstract: Enterprise architecture (EA) principles provide normative guidance for architectural evolution, yet validating whether EA models comply with such principles is typically performed manually and does not scale to continuous governance. This paper presents an ontology-based validation approach that enables automated compliance checking of ArchiMate models against EA principles. The approach (i) semantically lifts ArchiMate models into RDF/OWL as ontology instances grounded in ArchiMEO, (ii) structures natural-language principles using SBVR Structured English to reduce ambiguity and support traceability, (iii) enriches the resulting knowledge graph with inferred architectural relations through derivation rules, and (iv) operationalizes validation using SHACL constraints and SPARQL queries that produce explainable violation reports linked to concrete model elements. The approach is developed following Design Science Research and evaluated in three case studies (two real-world organizational settings and one controlled educational setting).The evaluation demonstrates that the approach supports repeatable execution of principle checks on evolving models, improves traceability of violations for architecture review and decision-making, and reduces manual effort by shifting substantial parts of compliance checking from human interpretation to automated constraint validation.

Abstract: Despite numerous studies on carbohydrate metabolism in gestational diabetes mellitus (GDM), the role of amino acid metabolic disturbances in the mother-fetus system remains insufficiently characterized, even though amino acids play a critical role in the develop-ment of fetal macrosomia (FM) and the programming of offspring metabolic health. This study included 62 mother–newborn dyads, stratified into clinical groups based on the presence of GDM and FM. Quantitative amino acid analysis was performed in maternal serum, umbilical cord serum, and amniotic fluid samples. Statistical analysis included Kruskal–Wallis, Mann–Whitney, chi-square tests, Spearman's correlation, and machine learning methods (Random Forest) with SHAP value calculation. Metabolic pathway analysis was conducted using MetaboAnalyst. Specific amino acid markers were identi-fied for each biological compartment. In maternal serum, GDM markers included glycine, 1-methylhistidine, γ-aminobutyric acid, lysine, and tryptophan, all showing significantly decreased levels. In cord blood, 11 amino acids exhibited reduced concentrations in GDM, including glutamine, glycine, asparagine, methionine, and proline. In amniotic fluid, GDM was associated with increased levels of lysine and 1-methylhistidine. In GDM complicated by FM, cord blood showed elevated lysine, proline, leucine, and al-lo-isoleucine, whereas amniotic fluid in this group was characterized by low homocitrul-line, asparagine, and lysine alongside high histidine levels. Correlation analysis revealed multiple associations between amino acids and clinical parameters, including an inverse correlation of fetal weight with homocitrulline and positive correlations with lysine and isoleucine. Metabolic pathway analysis indicated that GDM markers in maternal serum are associated with disturbances in biotin, glutamate, and carnitine metabolism, whereas cord blood markers implicated a broader spectrum of processes, including amino acid and purine metabolism. In amniotic fluid from GDM with FM, the methylhistidine me-tabolism pathway was additionally enriched, potentially reflecting specific alterations in neonatal muscle metabolism. GDM is accompanied by differential alterations in the amino acid profile across all investigated biological compartments, with the combination of GDM and FM characterized by unique metabolic signatures. The identified amino ac-ids may serve as potential biomarkers for early prediction of GDM and its complications, and offer prospects for targeted correction of metabolic disturbances.

Abstract: S. Tomé and Principe (STP) islands have been studied in recent years for their wide range of medicinal plants which exhibit several biological activities of great medicinal interest for some diseases. Experimental planning for optimization of several parameters was carried out by a full factorial of two levels of three factors for secondary metabolite extraction from Rauvolfia caffra leaves by using water and hexane at 25 and 40 ºC and 200 rpm for 0 and 5 days of incubation/extraction. The best conditions for highest extraction of phenolic compounds (i.e 89.90 moles gallic acid equivalent/g leaves)) was obtained at 25ºC, in H20 and 5 days of incubation. Several phytochemical assays were performed for characterization of these plant extracts and the highest levels of TFC, DPPH and Reducing power were obtained with aqueous plant extraction at 25ºC and for 5 days of incubation whereas leaves extraction with water at 40º C for 5 days of incubation revealed highest levels of ABTS scavenging activity. The levels of SOD and superoxide radical scavenging activities were highest with plant extraction with hexane at 25 and 40ºC for 5 days of incubation, respectively. The present report consists of a novel and intrinsic synchronous fluorescence and phosphorescence characterization of secondary metabolites from this plant extract. Intrinsic and non-destructive synchronous fluorescence was carried out in the range of 250 to 750 nm with a Δλ range of 5–30 nm which exhibited peaks at 320, 530, 550, 590, 650, 675, 690, 700, 710 nm in hexane plant extracts whereas aqueous extracts revealed only peaks at 382, 430, 460 and 530 nm. On the other hand, intrinsic and non-destructive synchronous phosphorescence was also performed which exhibited peaks at 430, 500 and 540 nm in aqueous extracts whereas hexane extracts revealed peaks at 320, 530, 560, 655, 675, 690 and 710 nm, respectively. 3-D spectra of secondary metabolites confirmed the peaks at 290, 320, 345, 400, 490 and 675 nm in plant extracts. FTIR spectroscopy was selected to investigate the structural properties of secondary metabolites in these plant extracts. Therefore, the present work describes a novel characterization of secondary metabolites by a non-destructive and intrinsic synchronous fluorescence techniques for plant extracts.

Abstract: Background: The standard cosmological model provides an excellent phenomenological description of the Universe. Motivated by persistent cosmological tensions, particularly in the Hubble constant (H0), this study proposes and tests the Gibbs Energy Redistribution Theory (GERT) as a thermodynamically grounded alternative to the expansion history model. Our central hypothesis is that a dynamical expansion history derived from fundamental thermodynamic principles empirically alleviates cosmological tensions, including the Hubble tension. At this stage, it provides a more physically coherent description of cosmic evolution, interpreting effects traditionally attributed to ad hoc dark components as emergent thermodynamic manifestations. Methods: We introduce a phenomenological, thermodynamically motivated model, the Gibbs Energy Redistribution Theory (GERT), in which the effective contributions of the matter- and lambda-like sectors are promoted to smooth, density-controlled functions, yielding a dynamical expansion history H(z) within the Friedmann Equation framework. We compared the resulting H(z) predictions with cosmic microwave background (CMB) shift-parameter constraints, baryon acoustic oscillation (BAO) distance measurements, and Type Ia supernova data. The analysis pipeline used standard open source scientific Python tools. Results: For the baseline implementation (the minimal-parameter reference fit described in the Methods), we obtained an excellent global fit (degrees of freedom (dof): χ2/dof ≈ 0.99) against CMB shift-parameter constraints, BAO distance measurements, and Type Ia supernova data, and inferred H0 ≈ 72.5 km s−1 Mpc−1, consistent with local determinations (e.g., SH0ES project). We quantified the deviations from the standard model in the diagnostic plots of H(z) and distance moduli. Conclusions: The framework yields concrete, testable predictions for late time expansion behavior, offering a physically coherent and causal narrative for cosmic evolution, and can be further constrained by future low-redshift probes.

Abstract:

Background: Mesh implantation is the standard of care in hernia repair. However, penetrating suture fixation may contribute to chronic pain and tissue irritation. This pilot study evaluates the feasibility of a hybrid fixation technique using a biodegradable UV-curable adhesive biomaterial in inguinal hernia repair.Methods: Ten male patients (20-40 years) with unilateral inguinal hernia underwent open repair and were allocated into two groups (n = 5 each): hybrid fixation approach (part of the mesh was secured conventionally and the remaining portion was stabilized with an experimental adhesive UV-curable biomaterial within 3 minutes) and conventional mesh fixation. Pain (VAS) and patient-reported outcomes (CCS, EuraHS QoL, SF-36) were assessed at day 1, day 8, 6 weeks, 12 months, and 24 months. Ultrasonography and thermography were analysed when available as exploratory assessments.Results: The adhesive-assisted partial self-stabilization reduced operative time compared with conventional fixation (52.0 ± 3.1 vs 60.2 ± 3.7 min). Postoperative pain (VAS) in the hybrid group decreased from 2.6 ± 0.55 on day 1 to 0.8 ± 0.84 on day 8, with complete resolution by 6 weeks. Foreign-body sensation (CCS) decreased from day 1 to 6 weeks in both groups (hybrid: 54.08% to 30.38%, control: 65.32% to 36.57%). No intraoperative complications and no hernia recurrences were observed during the 24-month follow-up. Overall SF-36 scores increased from 77.8 preoperatively to 92.4 at 24 months. Conclusions: In this pilot cohort, hybrid fixation using the UV-curable adhesive was feasible and was associated with shorter operative time, with no intraoperative complications and no recurrences observed during follow-up. Further studies of hybrid mesh fixation on larger cohorts are warranted.

Abstract: Reliable aboveground biomass (AGB) estimates for woody crops are required for carbon accounting and MRV; however, it remains unclear how LiDAR modality and sampling geometry influence plot-scale and tree-scale AGB predictions in intensively managed orchards. We benchmarked four LiDAR modalities across three Mediterranean woody-crop sites in Córdoba (Spain), IFAPA, Doña María, and Villaseca using open national airborne laser scanning (PNOA/ALS), Riegl ALS, unmanned laser scanning (ULS), and mobile laser scanning (MLS). The field inventory used 58 fixed-area plots (20×50 m; 0.1 ha) collected in December 2024-January 2025 (1,867 trees) and species-specific allometries based on D2r to derive tree and plot AGB; carbon was computed using wood carbon fractions (0.445 olive; 0.457 almond) and CO2e via IPCC conversion. Plot-level LiDAR metrics (e.g., mean height, p95, maximum height, and cover proxies) were extracted from normalized point clouds and modeled with Random Forest, XGBoost, and an ensemble under an 80/20 train-test split. Mean field AGB differed among sites (33.89, 30.94 and 12.76 Mg ha−1 for Villaseca, Doña María, and IFAPA). In the provided summaries, XGBoost achieved the lowest errors at IFAPA (RMSE = 0.400 Mg ha−1; R2 = 0.994) and Villaseca (RMSE = 0.872 Mg ha−1; R2 = 0.995), whereas PNOA was competitive at Doña María (RMSE = 0.725 Mg ha−1; R2 = 0.994). The results support cross-platform LiDAR for orchard AGB mapping and identify conditions under which open national LiDAR can enable scalable MRV. In addition, we evaluated TreeQSM-based quantitative structure models (QSMs) as an independent tree-level 3D reconstruction approach and examined their site-dependent agreement with field inventory estimates.

Abstract: The triangle finding problem is a cornerstone of complex network analysis, serving as the primitive for computing clustering coefficients and transitivity. This paper presents \texttt{Aegypti}, a practical algorithm for triangle detection and enumeration in undirected graphs. By combining a descending degree-ordered vertex-iterator with a hybrid strategy that adapts to graph density, \texttt{Aegypti} achieves a worst-case runtime of $\mathcal{O}(m^{3/2})$ for full enumeration, which matches the bound established by Chiba and Nishizeki for arboricity-based listing algorithms. For the detection variant ($\texttt{first\_triangle}=\text{True}$), we prove that sorting by non-increasing degree enables early termination in $\mathcal{O}(n\log n + d_{\max}^2)$ worst-case time when the maximum-degree vertex participates in a triangle, where the quadratic factor in $d_{\max}$ reduces to $\mathcal{O}(d_{\max}/C(v_{\max}))$ in expectation when the local clustering coefficient $C(v_{\max}) > 0$. Experiments on complement graphs of DIMACS maximum-clique benchmark instances confirm that detection terminates sub-millisecond on the majority of instances, while the matrix-multiplication baseline requires substantially more time on the same inputs.

Abstract: Automatic modulation classification (AMC) is a core capability for spectrum monitoring, adaptive receivers, and electronic support. Most radio-frequency machine learning (RFML) studies train multi-class classifiers on benchmark datasets that contain a single modulation per recording at baseband. In operational settings, however, the objective is often to detect only a small set of signals of interest, making large multi-class models unnecessarily expensive to train and deploy. This paper investigates an alternative workflow based on targeted binary transformer detectors and evaluates their robustness under practical RF complications. Using the RadioML 2018.01A dataset, we construct binary detection tasks with BPSK as the signal of interest and introduce three increasingly realistic conditions: (i) center-frequency shifts away from baseband, (ii) sampling-rate mismatches via decimation and interpolation, and (iii) multi-signal mixtures where modulations co-occur either in frequency (simultaneous transmissions) or in time (temporal concatenation). The results show that baseband-trained detectors do not generalize to center-frequency-shifted signals, and multi-signal interference can cause complete detection failure unless explicitly modeled during training. We investigate early-exit transformer inference to reduce computation on high-confidence examples, showing it maintains (and occasionally improves) detection performance. We also evaluate inter-modulation transfer learning and intra-modulation adaptation from baseband to mixed- and multi-signal scenarios.

Abstract:

Rust, caused by Puccinia arachidis, is one among the most destructive fungal diseases constraining global groundnut (Arachis hypogaea L.) production. While the development of disease-resistant varieties stands as the most effective approach to preventing substantial yield losses, the genetic mechanisms underlying resistance to rust is not yet well understood, emphasizing the necessity for further detailed research. In this study, 184 accessions from the ICRISAT groundnut mini-core collection were evaluated for rust resistance at Dharwad, India, across multiple seasons, as well as in Vietnam for one season. Whole-genome resequencing-based genome-wide association study (GWAS) identified five highly significant marker trait associations (MTAs) for rust resistance (p = 5.22 × 10-13 to 7.21 × 10-08). Among these, two robust rust-associated kompetitive allele specific PCR (KASP) markers, snpAH00607 at chromosome Ah01 and snpAH00609 at chromosome Ah17, were validated across diverse set of breeding and pre-breeding lines. These markers were linked to candidate genes encoding sterol C4-methyl oxidase 1-2, implicated in brassinosteroid-mediated salicylic acid signalling, and MYB transcription factor known to be associated with defense responses. The identified SNPs, validated markers, and candidate genes will serve as important resources for marker-assisted breeding of rust disease resistant groundnut varieties.