Modern digital engineering design process commonly involves expensive repeated simulations on varying three-dimensional (3D) geometries. The efficient prediction capability of neural networks (NNs) makes them a suitable surrogate to provide design insights. Nevertheless, few available NNs can handle solution prediction on varying 3D shapes. We present a novel deep operator network (DeepONet) variant called Geom-DeepONet, which encodes parameterized 3D geometries and predicts full-field solutions on an arbitrary number of nodes. To the best of the authors' knowledge, this is the first attempt in the literature and is our primary novelty. In addition to expressing shapes using mesh coordinates, the signed distance function for each node is evaluated and used to augment the inputs to the trunk network of the Geom-DeepONet, thereby capturing both explicit and implicit representations of the 3D shapes. The powerful geometric encoding capability of a sinusoidal representation network (SIREN) is also exploited by replacing the classical feedforward neural networks in the trunk with SIREN. Additional data fusion between the branch and trunk networks is introduced by an element-wise product. A numerical benchmark was conducted to compare Geom-DeepONet to PointNet and vanilla DeepONet, where results show that our architecture trains fast with a small memory footprint and yields the most accurate results among the three with less than 2 MPa stress error. Results show a much lower generalization error of our architecture on unseen dissimilar designs than vanilla DeepONet. Once trained, the model can predict vector solutions, and speed can be over $10^5$ times faster than implicit finite element simulations for large meshes.

Unlike classical artificial neural networks, which require retraining for each new set of parametric inputs, the Deep Operator Network (DeepONet), a lately introduced deep learning framework, approximates linear and nonlinear solution operators by taking parametric functions (infinite-dimensional objects) as inputs and mapping them to complete solution fields. In this paper, two newly devised DeepONet formulations with sequential learning and Residual U-Net (ResUNet) architectures are trained for the first time to simultaneously predict complete thermal and mechanical solution fields under variable loading, loading histories, process parameters, and even variable geometries. Two real-world applications are demonstrated: 1- coupled thermo-mechanical analysis of steel continuous casting with multiple visco-plastic constitutive laws and 2- sequentially coupled direct energy deposition for additive manufacturing. Despite highly challenging spatially variable target stress distributions, DeepONets can infer reasonably accurate full-field temperature and stress solutions several orders of magnitude faster than traditional and highly optimized finite-element analysis (FEA), even when FEA simulations are run on the latest high-performance computing platforms. The proposed DeepONet model's ability to provide field predictions almost instantly for unseen input parameters opens the door for future preliminary evaluation and design optimization of these vital industrial processes.

This article aims to show the potential contribution of high-yielding rice varieties to achieve sustainable intensification in paddy farming, by focusing on a developing country. A comparative life cycle assessment of traditional vs. high-yielding varieties is carried out by comparing the area-based and yield-based results. Primary data are collected through a farm survey (49 farms in the Mazandaran province, Iran; spring 2018). The results highlight that high-yielding varieties can reduce the yield-scaled impacts. However, area-scaled impacts are subject to increase for most impact categories. Statistically significant trade-offs involve global warming potential (+13% per ha and −28% per t in high-yielding varieties) and fossil resource depletion (+15% per ha and −26% per t in high-yielding varieties). Pesticide management is the most alarming practice. High-yielding varieties increase pesticide consumption and related toxicity impacts both per t and per ha. This study is a new contribution to the literature by improving and broadening the mainstream productivity perspective of current life cycle assessment research about crop varieties. The lessons learnt from this study suggest that the trade-offs between yield-scaled and area-scaled impacts should be carefully considered by decision-makers and policymakers, especially in developing countries that, like Iran, are affected by the overexploitation of natural resources. Targeted policy and the development of farmer education and advisory services are needed to create the enabling conditions for farm management changes, including conscious use of production inputs while avoiding heuristics.

Waste water in the galvanic process contains high concentrations of heavy metals that pose a direct danger to humans and the environment. Conventional methods for their removal are quite expensive and generate a large amount of waste. The development of new and improvement of existing methods for the removal of heavy metals from galvanic wastewater are the subject of many studies. Compared to other purification methods, the adsorption is becoming an increasingly popular method of wastewater purification, especially if the adsorbent is cheap, easily available and does not require any other treatment before use. Therefore, the aim of the work was to investigate the possibility of using natural bentonite for the removal of heavy metal ions from multi-component water systems of the galvanic industry. For this purpose, the physico-chemical characterization of natural bentonite was performed, and then the influence of pH value, time and temperature on the adsorption efficiency was examined. The results of adsorption showed that natural bentonite can be used as an adsorbent for the removal of heavy metal ions from waste galvanic waters, and that at pH 5 it achieves the maximum removal efficiency for Cu(II):Cr(III):Ni(II) ions in the percentage ratio 100 : 99.990 : 99.998. The results showed that the highest removal efficiency for Cu (II) ions was achieved in the first 10 minutes, and 20 minutes for Cr (III) and Ni (II) ions. The maximum efficiency of Cu (II) removal was achieved at all temperatures, while for Cr (III) 99.99% and Ni (II) 100% maximum efficiency was achieved at 35°C, which indicates that the adsorption process is endothermic. The experimental results of the adsorption of Cu (II) metal ions are in good agreement with the Langmuir and Freundlich theoretical models, while for Cr (III) and Ni (II) ions they are in better agreement with the Langmuir adsorption model.

Abstract Histologic transformation to small cell lung cancer (tSCLC) is a rare but increasingly recognised mechanism of acquired resistance to tyrosine kinase inhibitors (TKI) in patients with epidermal growth factor receptor (EGFR)-positive non-small cell lung cancer (NSCLC). Beyond its acknowledged role in TKI resistance, histologic transformation to SCLC might be an important, yet under-recognised, mechanism of resistance in NSCLC treated with immunotherapy. Our review identified 32 studies that investigated tSCLC development in patients with EGFR-mutated NSCLC treated with TKI therapy and 16 case reports of patients treated with immunotherapy. It revealed the rarity of tSCLC, with a predominance of EGFR exon 19 mutations and limited therapeutic options and outcomes. Across all analysed studies in EGFR-mutated NSCLC treated with TKI therapy, the median time to tSCLC development was ∼17 months, with a median overall survival of 10 months. Histologic transformation of EGFR-mutated NSCLC to SCLC is a rare, but challenging clinical problem with a poor prognosis. A small number of documented cases of tSCLC after immunotherapy highlight the need for rebiopsies at progression to diagnose this potential resistance mechanism. Further research is needed to better understand the mechanisms underlying this phenomenon and to develop more effective treatment strategies for patients with tSCLC.

Next-generation mobile networks, such as 5G/6G, have envisioned the possibility of direct communication between user devices, known as Device-to-Device (D2D) communication. Given that in D2D networks, traffic is transmitted ad-hoc from device to device, the range and quality of service are directly dependent on the number of nodes forming the D2D network. Therefore, we need to incentivize users to participate in the network operation through appropriate compensation for the provided resources and work done. A D2D network formed this way is inherently decentralized, making blockchain the primary choice as a technology. In this paper, we propose a new blockchain-based protocol for active tracing of IP traffic via in-band network telemetry. The experiment demonstrates that the proposed protocol can record all nodes participating in traffic forwarding in the D2D network through active traffic monitoring. Blockchain-based microtransactions can use participation records provided by our protocol to incentivize users to expand and strengthen the D2D network.

The applications presented in this conference paper focus on the development of a mobile and web application serving as a planner with a focus on tracking persons with Down syndrome. These innovative technological solutions contribute to the development of independence and functionality for persons with Down syndrome while emphasizing the importance of inclusivity in society. In addition to focusing on organizing activities, the mobile and web applications provide support and facilitate daily tasks. The web application allows parents/guardians/teachers to add new activities to the planner and track the progress of these activities. On the other hand, the mobile application enables persons with Down syndrome to record their activities within the application, considering their specific challenges, and customizing the user interface to their needs.

Software development is implemented in several key phases, one of which is software testing. Software testing consists of selecting techniques for the purpose of finding software defects and bugs in the process of writing code. There are several ways and approaches that lead us to that purpose, with the goal of selecting the most adequate method in terms of cost, complexity, and efficiency. In this paper, we will take a deeper dive into mutation testing techniques. Mutation testing techniques are fault-based and focus more on test structures than the input data, which is considered the testing start point. The basic concept of mutation testing consists of a few steps, which will be covered in this paper, and metrics that measure how effective the tests really are. With a few code examples, we will show why code coverage, which is mostly taken as a measure while testing, is sometimes not the most reliable source and does not give a full picture when talking about the quality of written tests.

Embedded real-time clock systems have a large number of applications in practice. The main issue is the accuracy of time they show, which is why performing time synchronization is very important for their usability and reliability. This paper proposes an embedded real-time analogue clock that uses an AdaFruit NeoPixel LED ring for visualizing current time. Three different colors are used for showing hour, minute and second values, whereas different levels of brightness are used for describing accurate values of time to the level of milliseconds. An Ethernet LAN module is used for performing time synchronization via a remote NTP server. Dynamic synchronization interval change is used for removing the effect of the microcontroller clock error on the accuracy of the shown time. After being put to use, the system was able to perform multiple functions successfully, including the conveying of information to the user when the clock is out of sync.

This paper aimed to explore ways to organize Spotify playlists, relying on clustering algorithms. Clustering algorithms were performed on playlists with extracted and standardized audio features obtained from the Spotify API, and the algorithms used were KMeans, DBSCAN, Affinity Propagation, and Spectral Clustering. Their performances were measured with the silhouette score, execution time, and inspection of clustered tracks, where it was determined that KMeans was the best algorithm in this case. Even though the execution time of KMeans is the third best, its silhouette score is the highest with 0.263. With this model, it is possible to effectively perform a mood-based organization of one's Spotify playlist, by dividing it into multiple smaller ones that share similar audio features.

With the growing requirements to keep the security of supply higher than ever the room for failures is getting smaller in today's power systems, while the increased integration of distributed renewable energy sources is additionally complicating fault detection. By using big data that is collected in modern power systems, artificial intelligence algorithms can significantly improve the capabilities of traditional protection schemes. However, the choice of the artificial intelligence algorithm can significantly impact the scheme accuracy. This paper analyses a novel approach for power system fault detection and classification by using automated machine learning procedure that iterates over different data transformations, machine learning algorithms, and hyperparameters to select the best model. By simulating and testing tens of thousands of fault scenarios on a realistic test system, the suggested approach resulted with robustness and high accuracy.

Recognizing the increasing importance of renewable energy sources, specifically wind farms, in today's power environments, this paper aims to clarify the complex interactions between these renewable energy facilities and distribution grids functioning under low-demand conditions. This particular case comes with inherent limitations that must be considered by taking into account all the factors that can influence the performance of the wind farm under these conditions. The modelling procedure and the simulation of the connection of the wind farm to the power system in rural area was performed using EMTP-RV software. The mean annual production of the wind power plant and the behaviour of the wind power plant in the event of failure in a real power system were calculated. Also, the power quality was examined in agreement with the Network Code of the transmission system of Bosnia and Herzegovina.

In the era of Industry 4.0, service robot path planning has emerged as a pivotal element in the optimization of logistic tasks within manufacturing, warehousing and service applications. In this context, the adoption of advanced path planning algorithms, such as the Grey Wolf Optimizer (GWO) swarm algorithm, play a key role in enabling these robots to navigate through complex environments with precision and agility. Harnessing the power of bio-inspired algorithms, our framework establishes a methodical and effective approach to the intricate task of service robot path planning.

Denial of Service (DoS) attacks, particularly the distributed variant known as DDoS, are easily initiated but pose significant challenge in terms of mitigation, especially in the case of DDoS. These attacks involve the use of a vast number of packets, often generated by specialized programs and scripts, crafted for specific attack types like SYN flood, ICMP Smurf, and similar. Malicious DoS packets share similar attributes, such as packet length, interval time, destination port, TCP flags, and the number of connections to the same host or service. To rapidly identify anomalous packets amidst legitimate traffic, we propose a system that incorporates the Newcombe-Benford power law and Kolmogorov-Smirnov test. This approach enables the detection of matching first occurrences of leading digits, such as packet size indicating the use of automated scripts for malicious purposes, and the count of connections to the same host or service.