Journal of Computational Science最新文献

{"title":"A leading adaptive activation function for deep reinforcement learning","authors":"Chongjiexin Jia ,&nbsp;Tuanjie Li ,&nbsp;Hangjia Dong ,&nbsp;Chao Xie ,&nbsp;Wenxuan Peng ,&nbsp;Yuming Ning","doi":"10.1016/j.jocs.2025.102608","DOIUrl":"10.1016/j.jocs.2025.102608","url":null,"abstract":"<div><div>The activation function provides deep reinforcement learning with the capability to solve nonlinear problems. However, traditional activation functions have fixed parameter settings and cannot be adjusted adaptively based on constantly changing environmental conditions. This limitation frequently leads to slow convergence speed and inadequate performance of trained agents when confronted with highly complex nonlinear problems. This paper proposes a new method to enhance the ability of reinforcement learning to handle nonlinear problems. This method is mainly divided into two parts. Firstly, an activation function parameter initialization strategy based on environmental characteristics is adopted. Secondly, the Adam algorithm is used to dynamically update the activation function parameters. The activation function proposed in this paper is compared with both traditional activation functions and state-of-the-art activation functions through two experiments. Experimental data show that compared to ReLu, TanH, APA, and EReLu, its convergence speed in DQN tasks is improved by 3.89, 1.29, 0.981, and 2.173 times, respectively, and in SAC tasks, it is improved by 1.504, 1.013, 1.017, and 1.131 times, respectively. The results demonstrate that when the agent utilizes LaTanH as the activation function, it exhibits significant advantages in terms of convergence speed and performance and alleviates the problems of bilateral saturation and gradient vanishing.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"88 ","pages":"Article 102608"},"PeriodicalIF":3.1,"publicationDate":"2025-04-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143878732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Accelerating Particle-in-Cell Monte Carlo simulations with MPI, OpenMP/OpenACC and Asynchronous Multi-GPU Programming","authors":"Jeremy J. Williams ,&nbsp;Felix Liu ,&nbsp;Jordy Trilaksono ,&nbsp;David Tskhakaya ,&nbsp;Stefan Costea ,&nbsp;Leon Kos ,&nbsp;Ales Podolnik ,&nbsp;Jakub Hromadka ,&nbsp;Pratibha Hegde ,&nbsp;Marta Garcia-Gasulla ,&nbsp;Valentin Seitz ,&nbsp;Frank Jenko ,&nbsp;Erwin Laure ,&nbsp;Stefano Markidis","doi":"10.1016/j.jocs.2025.102590","DOIUrl":"10.1016/j.jocs.2025.102590","url":null,"abstract":"&lt;div&gt;&lt;div&gt;As fusion energy devices advance, plasma simulations play a critical role in fusion reactor design. Particle-in-Cell Monte Carlo simulations are essential for modeling plasma-material interactions and analyzing power load distributions on tokamak divertors. Previous work (Williams, 2024) introduced hybrid parallelization in BIT1 using MPI and OpenMP/OpenACC for shared-memory and multicore CPU processing. In this extended work, we integrate MPI with OpenMP and OpenACC, focusing on asynchronous multi-GPU programming with OpenMP Target Tasks using the “nowait” and “depend” clauses, and OpenACC Parallel with the “async(n)” clause. Our results show significant performance improvements: 16 MPI ranks plus OpenMP threads reduced simulation runtime by 53% on a petascale EuroHPC supercomputer, while the OpenACC multicore implementation achieved a 58% reduction compared to the MPI-only version. Scaling to 64 MPI ranks, OpenACC outperformed OpenMP, achieving a 24% improvement in the particle mover function. On the HPE Cray EX supercomputer, OpenMP and OpenACC consistently reduced simulation times, with a 37% reduction at 100 nodes. Results from MareNostrum 5, a pre-exascale EuroHPC supercomputer, highlight OpenACC’s effectiveness, with the “async(n)” configuration delivering notable performance gains. However, OpenMP asynchronous configurations outperform OpenACC at larger node counts, particularly for extreme scaling runs. As BIT1 scales asynchronously to 128 GPUs, OpenMP asynchronous multi-GPU configurations outperformed OpenACC in runtime, demonstrating superior scalability, which continues up to 400 GPUs, further improving runtime. Speedup and parallel efficiency (PE) studies reveal OpenMP asynchronous multi-GPU achieving an 8.77&lt;span&gt;&lt;math&gt;&lt;mo&gt;×&lt;/mo&gt;&lt;/math&gt;&lt;/span&gt; speedup (54.81% PE) and OpenACC achieving an 8.14&lt;span&gt;&lt;math&gt;&lt;mo&gt;×&lt;/mo&gt;&lt;/math&gt;&lt;/span&gt; speedup (50.87% PE) on MareNostrum 5, surpassing the CPU-only version. At higher node counts, PE declined across all implementations due to communication and synchronization costs. However, the asynchronous multi-GPU versions maintained better PE, demonstrating the benefits of asynchronous multi-GPU execution in reducing scalability bottlenecks. While the CPU-only implementation is faster in some cases, OpenMP’s asynchronous multi-GPU approach delivers better GPU performance through asynchronous data transfer and task dependencies, ensuring data consistency and avoiding race conditions. Using NVIDIA Nsight tools, we confirmed BIT1’s overall efficiency for large-scale plasma simulations, leveraging current and future exascale supercomputing infrastructures. Asynchronous data transfers and dedicated GPU assignments to MPI ranks enhance performance, with OpenMP’s asynchronous multi-GPU implementation utilizing OpenMP Target Tasks with “nowait” and “depend” clauses outperforming other configurations. This makes OpenMP the preferred application programming interface when performance portability, high thr","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"88 ","pages":"Article 102590"},"PeriodicalIF":3.1,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143874008","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An embedding-based method for inferring novel interlayers in multilayer networks","authors":"Pietro Cinaglia","doi":"10.1016/j.jocs.2025.102592","DOIUrl":"10.1016/j.jocs.2025.102592","url":null,"abstract":"<div><div>In biology, networks are applied for modelling heterogeneous entities (e.g., gene, disease, drugs) and their own interactions. In this context, the multilayer networks allow modelling multiple types of interactions on independent layers, which are in turn interconnected by interlayer edges. Link prediction is a crucial task, e.g., which allows discovering of novel relationships between biological entities (e.g., proteins and genes). The state-of-the-art reports several methods focused on link prediction. However, no one is specifically designed for inferring entire interlayers between the unconnected layers of a multilayer network. In this paper, we presented an in-house method for the inference of entire interlayers from pairs of unconnected layers of interest. The proposed method exploits two main approaches: the first constructs a set of primitive links between unconnected layers of interest, based on properties intrinsic to graph network models; the second refines these based on more complex features denoted from node embeddings to infer the candidate interlayer edges, which ultimately constitute the resulting interlayer. In our experimentation, the proposed method has exhibited an effective capability in inferring novel interlayers, even when the number of nodes within the layers of interest increase. Performance was evaluated by using several well-known Key Performance Indicators. Briefly, results showed an improvement by +15.73% and +116.38% in terms of F1-Score and MCC, respectively. Furthermore, the accuracy improved on average by +46.30%, as can also be seen from ROC-AUC and PR-AUC, which showed +44.48% and +38.45%, respectively.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"88 ","pages":"Article 102592"},"PeriodicalIF":3.1,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143859557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A three-stage framework combining neural networks and Monte Carlo tree search for approximating analytical solutions to the Thomas–Fermi equation","authors":"Hassan Dana Mazraeh ,&nbsp;Kourosh Parand","doi":"10.1016/j.jocs.2025.102582","DOIUrl":"10.1016/j.jocs.2025.102582","url":null,"abstract":"<div><div>This study presents an innovative framework that integrates physics-informed neural networks with Monte Carlo tree search to develop an approximate analytical solution for the Thomas–Fermi equation. The framework operates in three stages. Initially, a physics-informed neural network is used to generate a numerical approximation of the Thomas–Fermi equation. Subsequently, the Monte Carlo tree search algorithm identifies an analytical expression that closely approximates the numerical solution from the first stage, resulting in an initial analytical solution. In the final stage, the physics-informed neural network is employed once more to optimize the coefficients of the expression found by Monte Carlo tree search, further refining the accuracy of the solution. Experimental results validate the effectiveness of this approach, demonstrating its capability to solve the challenging and nonlinear Thomas–Fermi equation, for which an exact analytical solution is not available.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102582"},"PeriodicalIF":3.1,"publicationDate":"2025-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143816994","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Time series forecasting for multidimensional telemetry data based on Generative Adversarial Network in a Digital Twin","authors":"João Carmo de Almeida Neto ,&nbsp;Leandro Santiago de Araújo ,&nbsp;Leopoldo André Dutra Lusquino Filho ,&nbsp;Claudio Miceli de Farias","doi":"10.1016/j.jocs.2025.102589","DOIUrl":"10.1016/j.jocs.2025.102589","url":null,"abstract":"<div><div>The research related to Digital Twins has been increasing in recent years. Besides the mirroring of the physical word into the digital, there is the need of providing services related to the data collected and transferred to the virtual world. One of these services is the forecasting of physical part future behavior, that could lead to applications, like preventing harmful events or designing improvements to get better performance. One strategy used to predict any system operation is the use of time series models like Autoregressive Integrated Moving Average (ARIMA) or Long-Short Term Memory (LSTM) and improvements implemented using these algorithms. Recently, deep learning techniques based on generative models such as Generative Adversarial Networks (GANs) have been proposed to create time series, and the use of LSTM has gained more relevance in time series forecasting, but both have limitations that restrict the forecasting results. Another issue found in the literature is the challenge of handling multivariate environments/applications in time series generation. Therefore, new methods need to be studied in order to fill these gaps and, consequently, provide better resources for creating useful Digital Twins. In the proposed method we introduce the integration of a Bidirectional LSTM (BiLSTM) layer with a time series obtained by GAN that leads to improved forecasting of all feature of the available dataset in terms of accuracy. The obtained results demonstrate improved prediction performance.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"88 ","pages":"Article 102589"},"PeriodicalIF":3.1,"publicationDate":"2025-04-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854326","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A tripartite evolutionary game for strategic decision-making in live-streaming e-commerce","authors":"Georgia Fargetta,&nbsp;Laura R.M. Scrimali","doi":"10.1016/j.jocs.2025.102585","DOIUrl":"10.1016/j.jocs.2025.102585","url":null,"abstract":"<div><div>The rapid growth of live-streaming has transformed traditional e-commerce into an interactive and immersive experience, giving birth to live-streaming e-commerce. This paper investigates the strategic interactions between brands, social media influencers, and consumers under this mechanism. Using evolutionary game theory, we model decision-making dynamics across these three parties and analyze how their strategies develop over time. Our framework incorporates contractual penalties between brands and influencers, rewards for influencers, product returns, and subscription fees to capture realistic market behaviors. We derive replicator dynamics equations for each participant group and identify stable equilibrium strategies for the entire system. The application of replicator dynamics offers valuable perspectives on temporary states and strategies that achieve long-term equilibrium. We also present numerical simulations to validate the effectiveness of our model. In addition, we show how parameters, such as penalties and rewards, influence strategy selection and allow the system to achieve stability successfully. This research provides actionable recommendations for optimizing partnerships in live-streaming e-commerce supply chains.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102585"},"PeriodicalIF":3.1,"publicationDate":"2025-04-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143816993","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Parameter tuning of the firefly algorithm by three tuning methods: Standard Monte Carlo, quasi-Monte Carlo and latin hypercube sampling methods","authors":"Geethu Joy ,&nbsp;Christian Huyck ,&nbsp;Xin-She Yang","doi":"10.1016/j.jocs.2025.102588","DOIUrl":"10.1016/j.jocs.2025.102588","url":null,"abstract":"<div><div>There are many different nature-inspired algorithms in the literature, and almost all such algorithms have algorithm-dependent parameters that need to be tuned. The proper setting and parameter tuning should be carried out to maximize the performance of the algorithm under consideration. This work is the extension of the recent work on parameter tuning by Joy et al. (2024) presented at the International Conference on Computational Science (ICCS 2024), and the Firefly Algorithm (FA) is tuned using three different methods: the Monte Carlo method, the Quasi-Monte Carlo method and the Latin Hypercube Sampling. The FA with the tuned parameters is then used to solve a set of six different optimization problems, and the possible effect of parameter setting on the quality of the optimal solutions is analyzed. Rigorous statistical hypothesis tests have been carried out, including Student’s t-tests, F-tests, non-parametric Friedman tests and ANOVA. Results show that the performance of the FA is not influenced by the tuning methods used. In addition, the tuned parameter values are largely independent of the tuning methods used. This indicates that the FA can be flexible and equally effective in solving optimization problems, and any of the three tuning methods can be used to tune its parameters effectively.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102588"},"PeriodicalIF":3.1,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820613","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Discrete fractional neural networks within the framework of octonions: A preliminary exploration","authors":"Jie Ran ,&nbsp;Yonghui Zhou ,&nbsp;Thabet Abdeljawad ,&nbsp;Hao Pu","doi":"10.1016/j.jocs.2025.102586","DOIUrl":"10.1016/j.jocs.2025.102586","url":null,"abstract":"<div><div>Conventional neural networks constructed on real or complex domains have limitations in capturing multi-dimensional data with memory effects. This work is a preliminary exploration of discrete fractional neural network modeling within the framework of octonions. Initially, by introducing the discrete fractional Caputo difference operator into the octonion domain, we establish a novel system of discrete fractional delayed octonion-valued neural networks (DFDOVNNs). The new system provides a theoretical support for developing neural network algorithms that are useful for solving complex, multi-dimensional problems with memory effects in the real world. We then use the Cayley–Dickson technique to divide the system into four discrete fractional complex-valued neural networks to deal with the non-commutative and non-associative properties of the hyper-complex domain. Next, we establish the existence and uniqueness of the equilibrium point to the system based on the homeomorphism theory. Furthermore, by employing the Lyapunov theory, we establish some straightforward and verifiable linear matrix inequality (LMI) criteria to ensure global Mittag-Leffler stability of the system. In addition, an effective feedback controller is developed to achieve the system’s drive-response synchronization in the Mittag-Leffler sense. Finally, two numerical examples support the theoretical analysis. This research introduces a novel direction in neural network studies that promises to significantly advance the fields of signal processing, control systems, and artificial intelligence.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102586"},"PeriodicalIF":3.1,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143785566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Dispersal- and harvesting-induced dynamics of single-species inhabited in minimal ring-shaped patches","authors":"Arjun Hasibuan ,&nbsp;Bapan Ghosh ,&nbsp;Asep K. Supriatna","doi":"10.1016/j.jocs.2025.102581","DOIUrl":"10.1016/j.jocs.2025.102581","url":null,"abstract":"<div><div>We investigate two discrete-time models of a single-species dispersed between three patches located on a ring. The dynamic models are formulated by identical logistic maps with linear coupling. The co-existing equilibrium completely depends on the intrinsic growth rate and carrying capacity. However, stability depends only on intrinsic growth rate and dispersal rate. We shall analytically present the stability analysis of both the trivial and coexisting equilibrium in a two parameter plane. Our main focus is to explore the bifurcations at the coexisting equilibrium and their consequences in ecology. Increasing dispersal rate leads to a period-doubling bifurcation in the bi-directional dispersal model followed by a Neimark-Sacker bifurcation arising from each periodic branch. Our analysis reports the existence of either three stable 2-cycles or three distinct quasi-periodic modes resulting in either periodic–periodic–periodic multistability or quasiperiodic–quasiperiodic–quasiperiodic multistability. In contrast to the bi-directional model, only a Neimark-Sacker occurs in the uni-directional dispersal model for increasing dispersal rate. This uni-directional dispersal strategy does not exhibit any multistability. The co-existing equilibrium may experience an instability switching in both models while introducing harvesting in one of the patches. Under harvesting, the bi-directional model could induce a Neimark-Sacker bifurcation which is impossible to occur for increasing dispersal. We shall estimate the effort levels to achieve the same amount of harvested yield in both uni- and bi-directional dispersal models. These results might be interesting from biological conservation and fishery management viewpoints.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102581"},"PeriodicalIF":3.1,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143808259","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Modeling the effect of fear-inducing awareness programs on smoking cessation","authors":"Jyoti Maurya,&nbsp;Mamta Kumari,&nbsp;A.K. Misra","doi":"10.1016/j.jocs.2025.102584","DOIUrl":"10.1016/j.jocs.2025.102584","url":null,"abstract":"<div><div>Smoking remains a significant public health challenge, contributing to numerous preventable diseases and mortality worldwide. Addressing this issue requires innovative strategies to enhance smoking cessation rate. In this research work, we develop a mathematical model to evaluate the impact of fear-inducing awareness programs on promoting smoking cessation. The model incorporates the parameters depicting the behavioral changes of individuals to capture the dynamic interplay between fear-driven awareness programs and smoking behavior. We analyze the local and global stability of the equilibria obtained from the model, providing a comprehensive understanding of the system’s dynamics. Furthermore, we identify critical bifurcation phenomena, including saddle–node and transcritical bifurcations, occurring in both forward and backward directions, which elucidate the system’s qualitative changes under parameter variations. Numerical simulations are conducted using smoking prevalence data from the United States of America (USA) to validate the analytical results and explore the influence of key parameters on smoking behavior. Our findings highlight that intensifying the fear component within awareness programs is more effective in promoting smoking cessation compared to merely increasing the number of such programs.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102584"},"PeriodicalIF":3.1,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143808258","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}