Structural Variables Research Articles

Learning approaches to cost–benefit decision analysis of pile designs from site investigation plans

ABSTRACT The study introduces an approach to pile design optimisation in geotechnical engineering, integrating both Reinforcement Learning (RL) and Machine Learning (ML) algorithms with Cost–Benefit Decision Analysis (CBDA). The study explores three Site Investigation Plans (SIPs), described as Current (CSIP), Actual (ASIP) and Optimal (OSIP), that can be used to address uncertainties in subsurface conditions and pile design. RL is used to determine the optimal action in transitioning between these interrelated SIPs. Apart from the SIPs, the optimal action is also shown to depend on structural variables for the pile design. These variables are used in ML algorithms, such as Random Forest (RF), Gradient Boosting Machine (GBM) and Support Vector Machine (SVM), to predict the optimal action for decision-making of the SIP at a given site. The proposed CBDA framework is applied to a database consisting of 152 usable test piles across various U.S. states, which enables the assessment of SIPs under diverse geological conditions. The results show that RL effectively optimises SIP selection with RF achieving the highest overall accuracy in predicting optimal SIP actions. The study contributes a novel data-driven approach to CBDA and SIP evaluation in geotechnical decision-making.

Investigating phase changes in cyanide-containing organic compounds with high melting points based on Quantitative Structure-Property Relationship modelling

Cyanide-containing organic compounds exhibit heat-resistant properties due to their high melting points, which can exceed 250 °C. These compounds find widespread applications in various industries. To predict their thermal decomposition temperatures, a straightforward model is introduced in this study. Unlike complex descriptors and computer codes, this model relies on structural parameters specific to cyanide-containing organic compounds. The investigation revealed that certain structural variables significantly influence the onset decomposition temperature. Among these, the number and type of rings play a crucial role. The model was developed and tested using a dataset of 77 experimental thermal decomposition onset temperatures for cyanide-containing organic compounds. For training and test sets comprising 62 and 15 compounds, respectively, the predicted values demonstrated good agreement with experimental data. Specifically, the coefficient of determination (R2) was 0.922 for the training set and 0.897 for the test set. The root mean square deviation (RMSD) and average absolute deviation (AAD) were 14.37 and 17.11 K for the training set, and 11.10 and 13.80 K for the test set. Furthermore, the proposed model outperformed a more complex existing model when applied to 13 cyanide-containing organic compounds. Additionally, this method can be utilized for designing cyanide-containing organic compounds with desired decomposition temperatures.

Asymmetric Impact of Economic, And Structural Factors on the Development of the Non-Oil Export Sector in GCC Economies 1992-2022: Structural Reform Priorities to Sustainable Economic Growth

The current study empirically investigates to explore the links between economic and structural variables of non-oil export industry development in GCC economies. The research spans the years 1992 to 2022. It is understandable to wonder why some oil-producing countries have been able to diversify and build their economy while others have not. We postulate that disparities in the diversification and development patterns of oil producers are related to differences in their economic and structural characteristics. Divergence in diversification and development trends may result from such discrepancies. We measure the role of economic and structural elements to test this idea. The findings indicate that economic and structural factors play a critical role in the process of diversification and development. Countries have the higher-quality infrastructure, in particular. The findings imply that GCC authorities should improve robust macroeconomic conditions to support their non-oil export sector and boost institutional quality drivers to accelerate non-oil export sector development in their nations. The empirical findings can assist GCC policymakers in better designing the economic policies required to accomplish the GCC countries' comprehensive and sustainable economic development. Asymmetric mechanisms in non-oil export responses to economic and structural factor changes are likely to differ across economic sectors in GCC countries. A sectoral examination of this link could provide further information and supplement our analysis. New capital and infrastructure must be accommodated to support domestic and export manufacturing.

Regional patterns and drivers of modelled water flows along environmental, functional, and stand structure gradients in Spanish forests

Abstract. The study of the water cycle in the forest at large scales, such as countries, is challenging due to the difficulty of correctly estimating forest water flows. Hydrological models can be coupled with extensive forest data sources, such as national forest inventories, to estimate the water flow of forests over large extents, but so far the studies conducted have not analysed the role of stand structure variables or the functional traits of the forest on predicted blue and green water flows in detail. In this study, we modelled the water balance of Spanish forests using stand structure and species data from forest inventories to understand the effects of climate, stand structure, and functional groups on blue water flows. We calculated blue water and green water flows and expressed them relative to received precipitation. Relative blue water flow was mainly concentrated in the wetter regions (Atlantic and alpine biomes) of Spain (around 25 %) in comparison with the Mediterranean biomes (10 %–20 %) and during the autumn–winter season. The leaf area index (LAI) of the forest stand is the most important predictor of relative blue water, exhibiting a negative effect until it reaches a plateau at higher levels (around 2.5–3). Deciduous forests showed a greater relative blue water flow than evergreen functional groups (25 %–35 % and 10 %–25 %, respectively) primarily due to leaf fall during the autumn–winter season. This study highlights how green water is decoupled from blue water; namely, blue water depends on winter and autumn precipitation, while green water depends on the spring and summer water demand and how the species' functional traits (deciduous vs. evergreen) can influence blue water production.

Enhanced rock mass rating prediction from tunnel face imagery: A decision-supportive ensemble deep learning approach

Ensuring safety in tunnel construction necessitates a rapid and objective evaluation of exposed tunnel faces for proactive rock mass risk assessment. The diversity of rock types and discontinuities contributes to structural complexity and local variability, introducing subjectivity and uncertainty in traditional evaluations based on the experience of field engineers. In this study, a convolutional neural network (CNN) is employed for regression to enhance the objectivity and consistency of rock mass classification via rock mass rating (RMR). The model interprets the entire tunnel face image holistically by identifying and learning from complex patterns, rather than detecting specific geological features like fractured or discontinuity zones. The network model, comprising feature extraction and regression blocks, utilizes the EfficientNet family for computational efficiency and accuracy. Data augmentation and ensemble learning address prediction variability due to image quality, optimizing the model for accurate RMR predictions and providing error ranges. The RMR predictions using excavation site images closely follow the field-evaluated evolution. A server communication-based mobile application is developed for real-time RMR evaluation, enhancing its practical field applicability. In geology, tunnel, and geotechnical engineering, where decisions rely on extensive experience, our approach demonstrates that deep learning can enhance decision-making by analyzing large accumulated datasets to predict optimal outcomes.

The impact of mesh size, turbulence parameterization, and land‐surface‐exchange scheme on simulations of the mountain boundary layer in the hectometric range

AbstractThe horizontal grid spacing of numerical weather prediction models keeps decreasing towards the hectometric range. We perform limited‐area simulations with the Icosahedral Nonhydrostatic (ICON) model across horizontal grid spacings (1 km, 500 m, 250 m, 125 m) in the Inn Valley, Austria, and evaluate the model with observations from the Cross‐Valley Flow in the Inn Valley Investigated by Dual‐Doppler LIDAR Measurements (CROSSINN) measurement campaign. This allows us to investigate whether increasing the horizontal resolution automatically improves the representation of the flow structure, surface exchange, and common meteorological variables. Increasing the horizontal resolution results in an improved simulation of the thermally induced circulation. However, the model still faces challenges with scale interactions and the evening transition of the up‐valley flow. Differences between two turbulence schemes (1D turbulence kinetic energy (TKE) and 3D Smagorinsky) emerge due to their different surface transfer formulations, yielding a delayed evening transition in the 3D Smagorinsky scheme. Generally speaking, the correct simulation of the mountain boundary layer depends mostly on the representation of model topography and surface exchange, and the choice of turbulence parameterization is secondary.

Taxonomic and Functional Diversity of Reptiles in a Heterogeneous Landscape of Jalisco State, West-Central Mexico

Reptiles are threatened by anthropogenic landscape transformation, largely due to agriculture. However, ecosystems nowadays constitute a matrix of fragmented landscapes. We analyzed a heterogeneous landscape’s reptile taxonomic and functional diversity patterns across ten land cover/use types in Jalisco state, in west-central Mexico. At the alpha diversity level, we assessed the taxonomic diversity using q-order indices, and functional diversity with multidimensional indices (FRic, FDiv, and FEve) by land cover/use. We evaluated the differences in species composition among land cover/use types (beta diversity). We utilized multidimensional distance-based analyses (dbRDA) to evaluate the association between reptile diversity, habitat structure, and environmental variables. Species richness did not correlate with functional richness across land cover types overall, except for riparian habitat surrounded by crops (RH-C), which exhibited higher species richness and functional diversity. Secondary vegetation surrounded by temperate forest (SV-TF), riparian habitat surrounded by tropical dry forest (RH-TDF), and RH-C were the land cover/use types with the most functional groups. Herbaceous cover is crucial for preserving both reptile diversity facets in this landscape. These findings suggest that the availability of resources (e.g., riparian habitat and herbaceous cover) regardless of perturbation level could be more relevant for reptile diversity than the condition (tropical vs. temperate) due to the high plasticity and adaptation of the group. It is essential to recognize the ecological value of these habitats by adopting a holistic approach that values the intrinsic and ecological importance of reptile diversity.

Machine Learning for Evaluating Hospital Mobility: An Italian Case Study

This study delves into hospital mobility within the Italian regions of Apulia and Emilia-Romagna, interpreting it as an indicator of perceived service quality. Utilizing logistic regression alongside other machine learning techniques, we analyze the impact of structural, operational, and clinical variables on patient perceptions of quality, thus influencing their healthcare choices. The analysis of mobility trends has uncovered significant regional differences, emphasizing how the regional context shapes perceived service quality. To further enhance the analysis, SHAP (SHapley Additive exPlanations) values have been integrated into the logistic regression model. These values quantify the specific contributions of each variable to the perceived quality of service, significantly improving the interpretability and fairness of evaluations. A methodological innovation of this study is the use of these SHAP impact scores as weights in the data envelopment analysis (DEA), facilitating a comparative efficiency analysis of healthcare facilities that is both weighted and normative. The combination of logistic regression and SHAP-weighted DEA provides a deeper understanding of perceived quality dynamics and offers essential insights for optimizing the distribution of healthcare resources. This approach underscores the importance of data-driven strategies to develop more equitable, efficient, and patient-centered healthcare systems. This research significantly contributes to the understanding of perceived quality dynamics within the healthcare context and promotes further investigations to enhance service accessibility and quality, leveraging machine learning as a tool to increase the efficiency of healthcare services across diverse regional settings. These findings are pivotal for policymakers and healthcare system managers aiming to reduce regional disparities and promote a more responsive and personalized healthcare service.

A Lipid-Structured Model of Atherosclerosis with Macrophage Proliferation

Atherosclerotic plaques are fatty deposits that form in the walls of major arteries and are one of the major causes of heart attacks and strokes. Macrophages are the main immune cells in plaques and macrophage dynamics influence whether plaques grow or regress. Macrophage proliferation is a key process in atherosclerosis, particularly in the development of mid-stage plaques, but very few mathematical models include proliferation. In this paper we reframe the lipid-structured model of Ford et al. (J Theor Biol 479:48–63, 2019. https://doi.org/10.1016/j.jtbi.2019.07.003) to account for macrophage proliferation. Proliferation is modelled as a non-local decrease in the lipid structural variable. Steady state analysis indicates that proliferation assists in reducing eventual necrotic core lipid content and spreads the lipid load of the macrophage population amongst the cells. The contribution of plaque macrophages from proliferation relative to recruitment from the bloodstream is also examined. The model suggests that a more proliferative plaque differs from an equivalent (defined as having the same lipid content and cell numbers) recruitment-dominant plaque in the way lipid is distributed amongst the macrophages. The macrophage lipid distribution of an equivalent proliferation-dominant plaque is less skewed and exhibits a local maximum near the endogenous lipid content.

Robust decentralized adaptive compensation for the multi-axial real-time hybrid simulation benchmark

Real-time hybrid simulation (RTHS) is a powerful and highly reliable technique integrating experimental testing with numerical modeling for studying rate-dependent components under realistic conditions. One of its key advantages is its cost-effectiveness compared to large-scale shake table testing, which is attained by selectively conducting experimental testing on critical parts of the analyzed structure, thus avoiding the assembly of the entire system. One of the fundamental advancements in RTHS methods is the development of multi-dimensional dynamic testing. In particular, multi-axial RTHS (maRTHS) aims to prescribe multi-degree-of-freedom (MDOF) loading from the numerical substructure over the test specimen. Under these conditions, synchronization is a significant challenge in multiple actuator loading assemblies. This study proposes a robust and decentralized adaptive compensation (RoDeAC) method for the next-generation maRTHS benchmark problem. An initial calibration of the dynamic compensator is carried out through offline numerical simulations. Subsequently, the compensator parameters are updated in real-time during the test using a recursive least squares adaptive algorithm. The results demonstrate outstanding performance in experiment synchronization, even in uncertain conditions, due to the variability of reference structures, seismic loading, and multi-actuator properties. Notably, this achievement is accomplished without needing detailed information about the test specimen, streamlining the procedure and reducing the risk of specimen deterioration. Additionally, the tracking performance of the tests closely aligns with the reference structure, further affirming the excellence of the outcomes.

Navigating the Impact of Foreign Aid: A Holistic View of Educational Aid Disbursements in Developing Nations

Background: Foreign aid is a fundamental instrument of international cooperation, intended theoretically to ameliorate the disparities in saving-investment dynamics, knowledge acquisition, and foreign exchange within developing nations. Despite over half a century of endeavours to evaluate its efficacy, the ongoing discourse on whether aid engenders predominantly positive outcomes in recipient countries remains inconclusive. Consequently, the nuanced impact of foreign aid on addressing challenges in developing nations remains a topic necessitating comprehensive and rigorous analysis. Objective: The primary aim of this study is to augment the existing body of knowledge within the domain of foreign aid by scrutinising the impact of education aid on improving educational achievements, particularly at the primary and secondary education tiers, by focusing on the overall, gender-wise, income level-wise, and donor-wise effect. This study focuses on delineating the influence of aid on the Primary Completion Rate (PCR), recognising its role as a precursor to shaping the Secondary School Level Net Enrollment Rate (SERNet). Method: This study utilised recent data on education aid disbursements across 50 low—and lower-income countries over 19 years (2002-2020). It specifically examined the cross-sector effects of this aid within the education sector, focusing on its impact on both the PCR and the SERNet. The analysis was conducted using the two-stage least squares methodology using the instrumental variables (IV-2SLS). Result: The findings revealed a statistically significant and positive correlation between education aid and the PCR, while no substantial relationship was observed with the SERNet. This discrepancy indicates that aid to improve primary school completion may not necessarily lead to enhancements in secondary school enrollments. Additionally, the analysis demonstrated that economic factors, governance, and school-related structural variables also play crucial roles in improving educational outcomes in developing countries. Conclusion: These outcomes suggest that the efficacy of foreign aid exhibits variance across different scenarios. The challenges concerning PCR and SERNet likely diverge, implying that assessing aid impacts singularly might be inadequate. Tailoring an aid to address specific issues could yield more favourable outcomes. Paper Type: Research Paper

Advancements in hand-drawn chemical structure recognition through an enhanced DECIMER architecture

Accurate recognition of hand-drawn chemical structures is crucial for digitising hand-written chemical information in traditional laboratory notebooks or facilitating stylus-based structure entry on tablets or smartphones. However, the inherent variability in hand-drawn structures poses challenges for existing Optical Chemical Structure Recognition (OCSR) software. To address this, we present an enhanced Deep lEarning for Chemical ImagE Recognition (DECIMER) architecture that leverages a combination of Convolutional Neural Networks (CNNs) and Transformers to improve the recognition of hand-drawn chemical structures. The model incorporates an EfficientNetV2 CNN encoder that extracts features from hand-drawn images, followed by a Transformer decoder that converts the extracted features into Simplified Molecular Input Line Entry System (SMILES) strings. Our models were trained using synthetic hand-drawn images generated by RanDepict, a tool for depicting chemical structures with different style elements. A benchmark was performed using a real-world dataset of hand-drawn chemical structures to evaluate the model's performance. The results indicate that our improved DECIMER architecture exhibits a significantly enhanced recognition accuracy compared to other approaches.Scientific contributionThe new DECIMER model presented here refines our previous research efforts and is currently the only open-source model tailored specifically for the recognition of hand-drawn chemical structures. The enhanced model performs better in handling variations in handwriting styles, line thicknesses, and background noise, making it suitable for real-world applications. The DECIMER hand-drawn structure recognition model and its source code have been made available as an open-source package under a permissive license.Graphical

DNA structural features and variability of complete MHC locus sequences.

The major histocompatibility (MHC) locus, also known as the Human Leukocyte Antigen (HLA) genes, is located on the short arm of chromosome 6, and contains three regions (Class I, Class II and Class III). This 5Mbp locus is one of the most variable regions of the human genome, yet it also encodes a set of highly conserved and important proteins related to immunological response. Genetic variations in this region are responsible for more diseases than in the entire rest of the human genome. However, information on local structural features of the DNA is largely ignored. With recent advances in long-read sequencing technology, it is now becoming possible to sequence the entire 5Mbp MHC locus, producing complete diploid haplotypes of the whole region. Here, we describe structural maps based on the complete sequences from six different homozygous HLA cell lines. We find long-range structural variability in the different sequences for DNA stacking energy, position preference and curvature, variation in repeats, as well as more local changes in regions forming open chromatin structures, likely to influence gene expression levels. These structural maps can be useful in visualizing large scale structural variation across HLA types, in particular when this can be complemented with epigenetic signals.

Head motion in the UK Biobank imaging subsample: longitudinal stability, associations with psychological and physical health, and risk of incomplete data.

Participant motion in brain magnetic resonance imaging is associated with processing problems including potentially non-useable/incomplete data. This has implications for representativeness in research. Few large studies have investigated predictors of increased motion in the first instance. We exploratively tested for association between multiple psychological and physical health traits with concurrent motion during T1 structural, diffusion, average resting-state and task functional magnetic resonance imaging in N = 52 951 UK Biobank imaging subsample participants. These traits included history of cardiometabolic, inflammatory, neurological and psychiatric conditions, as well as concurrent cognitive test scores and anthropometric traits. We tested for stability in motion in participants with longitudinal imaging data (n = 5305, average 2.64 years later). All functional and T1 structural motion variables were significantly intercorrelated (Pearson r range 0.3-0.8, all P < 0.001). Diffusion motion variables showed weaker correlations around r = 0.1. Most physical and psychological phenotypes showed significant association with at least one measure of increased motion including specifically in participants with complete useable data (highest β = 0.66 for diabetes versus resting-state functional magnetic resonance imaging motion). Poorer values in most health traits predicted lower odds of complete imaging data, with the largest association for history of traumatic brain injury (odds ratio = 0.720, 95% confidence interval = 0.562 to 0.923, P = 0.009). Worse psychological and physical health are consistent predictors of increased average functional and structural motion during brain imaging and associated with lower odds of complete data. Average motion levels were largely consistent across modalities and longitudinally in participants with repeat data. Together, these findings have implications for representativeness and bias in imaging studies of generally healthy population samples.

Assessment of Mycological Possibility Using Machine Learning Models for Effective Inclusion in Sustainable Forest Management

The integral role of wild fungi in ecosystems, including provisioning, regulating, cultural, and supporting services, is well recognized. However, quantifying and predicting wild mushroom yields is challenging due to spatial and temporal variability. In Mediterranean forests, climate-change-induced droughts further impact mushroom production. Fungal fruiting is influenced by factors such as climate, soil, topography, and forest structure. This study aims to quantify and predict the mycological potential of Lactarius deliciosus in sustainably managed Mediterranean pine forests using machine learning models. We utilize a long-term dataset of Lactarius deliciosus yields from 17 Pinus pinaster plots in Soria, Spain, integrating forest-derived structural data, NASA Landsat mission vegetation indices, and climatic data. The resulting multisource database facilitates the creation of a two-stage ‘mycological exploitability’ index, crucial for incorporating anticipated mycological production into sustainable forest management, in line with what is usually done for other uses such as timber or game. Various Machine Learning (ML) techniques, such as classification trees, random forest, linear and radial support vector machine, and neural networks, were employed to construct models for classification and prediction. The sample was always divided into training and validation sets (70-30%), while the differences were found in terms of Overall Accuracy (OA). Neural networks, incorporating critical variables like climatic data (precipitation in January and humidity in November), remote sensing indices (Enhanced Vegetation Index, Green Normalization Difference Vegetation Index), and structural forest variables (mean height, site index and basal area), produced the most accurate and unbiased models (OAtraining = 0.8398; OAvalidation = 0.7190). This research emphasizes the importance of considering a diverse array of ecosystem variables for quantifying wild mushroom yields and underscores the pivotal role of Artificial Intelligence (AI) tools and remotely sensed observations in modeling non-wood forest products. Integrating such models into sustainable forest management plans is crucial for recognizing the ecosystem services provided by them.

Structural characteristics and non-linear fiscal multipliers

We identify structural characteristics of economies which determine the fiscal multipliers in normal times. We find that structural variables determine fiscal multipliers in a non-linear fashion where fiscal multipliers change significantly across the threshold of structural variables. Using the recently developed Panel Latent Threshold Model, we identify latent groups of countries intrinsic to the structural characteristics of the variables. Results suggest that improvement in human development and financial infrastructure measures play a prominent role in enhancing the fiscal multiplier effects across the economies. We also find that economies can achieve higher fiscal multipliers by keeping a check on the import propensity. We identify female labour force participation as one of the significant structural determinants of the fiscal multiplier, where many latent groups find parallels with the declining portion of the feminisation U-shaped hypothesis. Overall, the results suggest that countries should pursue targeted structural change-oriented fiscal policy to achieve higher economic growth.

Implementation of Specific Nutrition Interventions to Prevent Stunting in Pregnant Women in the Oesapa Community Health Center Area East Nusa Tenggara

Policies related to stunting prevention have been in place since 2013. However, the prevalence of stunting in East Nusa Tenggara Province is still the highest in Indonesia from 2013 to 2022. One of the areas in the province where the prevalence decreased from 2021 to 2022 is Kupang City. Oesapa Community Health Centre was appointed by the Indonesian Ministry of Health as a pilot health centre representing Kupang City. The purpose of this study is to describe the implementation of specific nutritional interventions for handling stunting based on the variables of communication, bureaucratic structure, resources, and disposition in the target group of pregnant women which will be associated with the target achievement of specific nutritional interventions based on Presidential Regulation No 72 of 2021. The research used qualitative methods with triangulation of informants totalling 5 people, data triangulation, and triangulation of methods, namely in-depth interviews and observations. The results showed that interventions in the form of providing additional food to pregnant women with chronic energy deficiency conditions and pregnant women who consume blood supplement tablets at least 90 tablets during pregnancy have successfully reached the target. Other Community Health Centres can make the Oesapa Community Health Centre a barometer in providing health services, especially in specific nutrition interventions aimed at target groups of pregnant women so that they work together in accelerating the reduction in stunting prevalence.

Implementation of Skills-Based Vocational Education Programmes for Youths Empowerment and Community Development in Ogun State, Nigeria

Due to some institutional, structural, personnel, and environmental variables of performance limitations, the expectations of vocational education programmes seem to have been compromised. This paper therefore presents an investigation on the implementation of vocational education programmes for community empowerment in Ogun State, Nigeria. Descriptive survey research was adopted for the study. From a total population of 384,740 technical education students, a sample of 380 students and 60 instructors were selected using a random sampling technique. A 5-point Likert structured questionnaire titled: “Vocational Education and Community Empowerment Scale (VECES)”, r=0.86 was used for data collection. Mean and standard deviation scores were used to answer the research questions, while a t-test was used to test the hypotheses at a 0.05 level of significance. Results showed that inadequate training facilities and an acute shortage of qualified instructors challenged vocational education programmes for community empowerment. Respondents did not significantly differ in their opinions on the contents, challenges and strategies to enhance the effective implementation of vocational education programmes. It was concluded that youths in the communities were yet to be empowered because of the performance limitations associated with implementing vocational education programmes as expected in the state’s policy document. Based on these findings, it was recommended among others that government and private sectors should provide qualified instructors, equipment and facilities in technical and vocational institutions for the acquisition of skills.

Floral traits and functional role of whorls in pollinator attraction of Magnolia grandiflora L.

Abstract Species within the primitive genus Magnolia may often produce specialized floral traits and behaviour to grab the attention of potential pollinators from their surroundings. These reproductive traits in plants undergo various selection pressures and frequently bring forth variations to adapt to the new habitat that may result in speciation. We have aimed to understand the floral traits, the functional role of floral parts, and the variable response of visitors in Magnolia grandiflora belonging to an ancient order Magnoliales. Our observations reveal that the floral traits of M. grandiflora resemble those of basal angiosperms, conforming to the primitive existence of the genus Magnolia. The inconsistency in floral form reflected that M. grandiflora has been experiencing various selection pressures from biotic and abiotic factors. Inconsistency in floral equations and diagrams reflect the structural variability in flowers, which can directly impact pollination and reproductive output. SEM images of pollen grains reveal that the pollen grains were monocolpate and boat-shaped in structure. GC-MS analysis showed that VOCs in stigmatic exudation of the flower were largely composed of terpene hydrocarbons such as Perthenine, β-Elemene, β-Caryophyllene, α-Humulene, Bicyclogermacrene, Germacrene A and D, etc. Beyond pollinators attraction, VOCs play a major role in repelling unwanted visitors and in improving the defensive mechanisms in Magnolia. Beetles, bees, and flies were found to be the active pollinators and the behaviour of bees indicates that bees were trying to replace the role of beetles in pollinating M. grandiflora. In summary, floral whorls were diversely functionalized to ensure maximum reproduction in M. grandiflora.

Efficient graph representation in graph neural networks for stress predictions in stiffened panels

Machine learning (ML) and deep learning (DL) techniques have gained significant attention as reduced order models (ROMs) to computationally expensive structural analysis methods, such as finite element analysis (FEA). Graph neural network (GNN) is a particular type of neural network which processes data that can be represented as graphs. This allows for efficient representation of complex geometries that can change during the conceptual design of a structure or a product. In this study, we propose a novel graph embedding for the efficient representation of 3D stiffened panels by considering separate plate domains as vertices. This approach is considered using Graph Sampling and Aggregation (GraphSAGE) to predict stress distributions in stiffened panels with varying geometries. A comparison between a finite element-vertex graph representation is conducted to demonstrate the effectiveness of the proposed approach. A comprehensive parametric study is performed to examine the effect of structural variables on stress predictions. A wide range of geometries is considered, material nonlinearity, a few boundary conditions, together with uniform and patch loading at various positions. The study involves straight and curved panels with uni- and bi-directional stiffeners. The proposed unit-vertex representation of the panel requires only about 2% of GPU memory and about 4% of training time in comparison to a finite element-vertex embedding. The GraphSAGE model with the proposed unit-vertex representation accurately captures stress distribution across all panels, achieving an average prediction accuracy of 92.3% for the maximum von Mises stress. Our results demonstrate the immense potential of graph neural networks with the proposed graph embedding as a robust reduced-order model for 3D structures.