Journal of Computational Science最新文献

{"title":"Towards universal MPI bindings for enhanced new language support","authors":"César Piñeiro ,&nbsp;Álvaro Vázquez ,&nbsp;Juan C. Pichel","doi":"10.1016/j.jocs.2025.102557","DOIUrl":"10.1016/j.jocs.2025.102557","url":null,"abstract":"<div><div>In the field of High Performance Computing (HPC), Message Passing Interface (MPI) is the most widely used and prevalent programming model. Only the low-level programming languages C, C++, and Fortran have bindings available in the standard. Although there are attempts to provide MPI bindings for other programming languages, these may be limited, which could lead to incompatibilities, performance overhead, and functional gaps. To address those problems, we present MPI4All, a brand-new tool designed to make the process of developing effective MPI bindings for any programming language more straightforward. Support for additional languages can be added with little difficulty, and MPI4All is independent of the MPI implementation. Programming language binding generators for Go and Java are included in the most recent version of MPI4All. We demonstrate their good performance results with respect to other state-of-the-art approaches. This work is an extended version of the ICCS-2024 conference paper (Piñeiro et al., 2024).</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102557"},"PeriodicalIF":3.1,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143636403","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":"AJAS: A high performance direct solver for advancing high precision astrometry","authors":"Konstantin Ryabinin,&nbsp;Gerasimos Sarras,&nbsp;Wolfgang Löffler,&nbsp;Olga Erokhina,&nbsp;Michael Biermann","doi":"10.1016/j.jocs.2025.102554","DOIUrl":"10.1016/j.jocs.2025.102554","url":null,"abstract":"<div><div>In astrometry, the determination of three-dimensional positions and velocities of stars based on observations from a space telescope suffers from the uncertainty of random and systematic errors. The systematic errors are introduced by imperfections of the telescope’s optics and detectors as well as in the pointing accuracy of the satellite. The fine art of astrometry consists of heuristically finding the best possible calibration model that will account for and remove these systematic errors. Since this is a process based on trial and error, appropriate software is needed that is efficient enough to solve the system of astrometric equations and reveal the astrometric parameters of stars for the given calibration model within a reasonable time. This paper is an extended version of the conference paper published and discussed at the International Conference on Computational Science 2024. In this work, we propose a novel software architecture and corresponding prototype of a direct solver optimized for running on supercomputers. The main advantages expected from this direct method over an iterative one are the numerical robustness, accuracy of the method, and the explicit calculation of the variance–covariance matrix for the estimation of the accuracy and correlation of the unknown parameters. This solver can handle astrometric systems with billions of equations within several hours. To reach the desired performance, we use state-of-the-art libraries and methods for hybrid parallel and vectorized computing. The calibration model based on Legendre polynomials is tested by generating synthetic observations on grid-shaped constellation with specified distortions. For these small-sized test data, the solver can recover perfectly the correct physical solution under the condition that the correct amount of eigenvalues is zeroed out. During the space mission, the calibration model should be carefully fine-tuned according to the real operating conditions. The developed solver is furthermore tested using mock science data related to the Japan Astrometry Satellite Mission for Infrared Exploration. Up to 9.2 billion observations of 115 thousand stars can be processed in 8.5 h utilizing 5000 CPUs. A linear scaling with the number of CPUs and a quadratic scaling with the number of observations is demonstrated.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102554"},"PeriodicalIF":3.1,"publicationDate":"2025-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143592646","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":"Influence of blood-related parameters for hyperthermia-based treatments for cancer","authors":"Gustavo Resende Fatigate ,&nbsp;Gustavo Coelho Martins ,&nbsp;Marcelo Lobosco ,&nbsp;Ruy Freitas Reis","doi":"10.1016/j.jocs.2025.102556","DOIUrl":"10.1016/j.jocs.2025.102556","url":null,"abstract":"<div><div>Hyperthermia is a cancer treatment method that uses controlled heat to induce tumor necrosis while preserving healthy tissue. This study uses computational simulations to investigate the effects of capillary network variability and blood flow dynamics on the thermal response during hyperthermia. A porous media bioheat model, coupled with uncertainty quantification (UQ) techniques using Monte Carlo simulations, was developed to analyze the influence of capillary angles, blood velocity, and capillary density on temperature distribution in biological tissues. The model demonstrates that under a range of physiological uncertainties, tumor tissues consistently reach the critical damage threshold temperature of <span><math><mrow><mn>4</mn><msup><mrow><mn>3</mn></mrow><mrow><mo>∘</mo></mrow></msup><mi>C</mi></mrow></math></span>, while healthy tissues remain below <span><math><mrow><mn>3</mn><msup><mrow><mn>8</mn></mrow><mrow><mo>∘</mo></mrow></msup><mi>C</mi></mrow></math></span>, minimizing collateral damage. To address the computational intensity of solving three-dimensional heat transfer equations with UQ analysis, high-performance computing methods were employed. A parallel implementation using CUDA achieved a speedup exceeding <span><math><mrow><mn>114</mn><mo>×</mo></mrow></math></span> compared to serial processing, while OpenMP achieved a <span><math><mrow><mn>16</mn><mo>×</mo></mrow></math></span> speedup.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102556"},"PeriodicalIF":3.1,"publicationDate":"2025-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143576747","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":"Establishing a massively parallel computational model of the adaptive immune response","authors":"Aristotle Martin,&nbsp;Max Nezdyur,&nbsp;Cyrus Tanade,&nbsp;Amanda Randles","doi":"10.1016/j.jocs.2025.102555","DOIUrl":"10.1016/j.jocs.2025.102555","url":null,"abstract":"<div><div>Parallel agent-based models of the adaptive immune response can efficiently recapitulate emerging spatiotemporal properties of T-cell motility during clonal selection across multiple length and time scales. Here, we present a distributed, three-dimensional (3D) computational model of T-cell priming, and associated parallel data structures and algorithms that enable fully deterministic cell simulations at scale. We demonstrate performant usage of modern clusters with over 350x speedup, and explore trade-offs between simulation accuracy, code complexity, and communication overhead. This study highlights the potential for parallel 3D models to explore immunological research questions and guides implementation and performance considerations for this class of biology-inspired agent-based models.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"87 ","pages":"Article 102555"},"PeriodicalIF":3.1,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563386","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":"Unsupervised continual learning by cross-level, instance-group and pseudo-group discrimination with hard attention","authors":"Ankit Malviya,&nbsp;Sayak Dhole,&nbsp;Chandresh Kumar Maurya","doi":"10.1016/j.jocs.2025.102535","DOIUrl":"10.1016/j.jocs.2025.102535","url":null,"abstract":"<div><div>Extensive work has been done in supervised continual learning (SCL) , wherein models adapt to changing distributions with labeled data while mitigating catastrophic forgetting. However, this approach diverges from real-world scenarios where labeled data is scarce or non-existent. Unsupervised continual learning (UCL) emerges to bridge this disparity. Previous research has explored methods for unsupervised continuous feature learning by incorporating rehearsal to alleviate the problem of catastrophic forgetting. Although these techniques are effective, they may not be feasible for scenarios where storing training data is impractical. Moreover, rehearsal techniques may confront challenges pertaining to representation drifts and overfitting, particularly under limited buffer size conditions. To address these drawbacks, we employ parameter isolation as a strategy to mitigate forgetting. Specifically, we use task-specific hard attention to prevent updates to parameters important for previous tasks. In contrastive learning, loss is prone to be negatively affected by a reduction in the diversity of negative samples. Therefore, we incorporate instance-to-instance similarity into contrastive learning through both direct instance grouping and discrimination at the cross-level with local instance groups, as well as with local pseudo-instance groups. The masked model learns the features using cross-level discrimination, which naturally clusters similar data in the representation space. Extensive experimentation demonstrates that our proposed approach outperforms current state-of-the-art (SOTA) baselines by significant margins, all while exhibiting minimal or nearly zero forgetting, and without the need for any rehearsal buffer. Additionally, the model learns distinct task boundaries. It achieves an overall-average task and class incremental learning (TIL &amp; CIL) accuracy of 76.79% and 62.96% respectively with nearly zero forgetting, across standard datasets for varying task sequences ranging from 5 to 100. This surpasses SOTA baselines, which only reach 74.28% and 60.68% respectively in the UCL setting, where they experience substantial forgetting of almost over 4%. Moreover, our approach achieves performance nearly comparable to the SCL baseline and even surpasses it on some standard datasets, with a notable reduction in forgetting from almost 14.51% to nearly zero.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"86 ","pages":"Article 102535"},"PeriodicalIF":3.1,"publicationDate":"2025-02-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143465131","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 cluster-based opposition differential evolution algorithm boosted by a local search for ECG signal classification","authors":"Mehran Pourvahab ,&nbsp;Seyed Jalaleddin Mousavirad ,&nbsp;Virginie Felizardo ,&nbsp;Nuno Pombo ,&nbsp;Henriques Zacarias ,&nbsp;Hamzeh Mohammadigheymasi ,&nbsp;Sebastião Pais ,&nbsp;Seyed Nooreddin Jafari ,&nbsp;Nuno M. Garcia","doi":"10.1016/j.jocs.2025.102541","DOIUrl":"10.1016/j.jocs.2025.102541","url":null,"abstract":"<div><div>Electrocardiogram (ECG) signals, which capturethe heart's electrical activity, are used to diagnose and monitor cardiac problems. The accurate classification of ECG signals, particularly for distinguishing among various types of arrhythmias and myocardial infarctions, is crucial for the early detection and treatment of heart-related diseases. This paper proposes a novel approach based on an improved differential evolution (DE) algorithm for ECG signal classification for enhancing the performance. In the initial stages of our approach, the preprocessing step is followed by the extraction of several significant features from the ECG signals. These extracted features are then provided as inputs to an enhanced multi-layer perceptron (MLP). While MLPs are still widely used for ECG signal classification, using gradient-based training methods, the most widely used algorithm for the training process, has significant disadvantages, such as the possibility of being stuck in local optimums. This paper employs an enhanced differential evolution (DE) algorithm for the training process as one of the most effective population-based algorithms. To this end, we improved DE based on a clustering-based strategy, opposition-based learning, and a local search. Clustering-based strategies can act as crossover operators, while the goal of the opposition operator is to improve the exploration of the DE algorithm. The weights and biases found by the improved DE algorithm are then fed into six gradient-based local search algorithms. In other words, the weights found by the DE are employed as an initialization point. Therefore, we introduced six different algorithms for the training process (in terms of different local search algorithms). In an extensive set of experiments, we showed that our proposed training algorithm could provide better results than the conventional training algorithms.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"86 ","pages":"Article 102541"},"PeriodicalIF":3.1,"publicationDate":"2025-02-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143471302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Community-based voting approach to enhance the spreading dynamics by identifying a group of influential spreaders in complex networks","authors":"Suman Nandi ,&nbsp;Mariana Curado Malta ,&nbsp;Giridhar Maji ,&nbsp;Animesh Dutta","doi":"10.1016/j.jocs.2025.102540","DOIUrl":"10.1016/j.jocs.2025.102540","url":null,"abstract":"<div><div>Exploring a group of influential spreaders to acquire maximum influence has become an emerging area of research in complex network analysis. The main challenge of this research is to identify the group of important nodes that are scattered broadly, such that the propagation ability of information is maximum to a network. Researchers proposed many centrality-based approaches with certain limitations to identify the influential nodes (spreaders) considering different properties of the networks. To find a group of spreaders, the VoteRank (a voting mechanism) based method produces effective results with low time complexity, where in each iteration, the node votes for its neighbors by its voting capability, and the node obtaining the maximum vote score is identified as an influential spreader. The major loophole of existing VoteRank methods is measuring the voting capability based on the degree, k-shell index, or contribution of neighbors methods, which does not efficiently identify the spreaders from the diverse regions based on their spreading ability. In this paper, we propose a novel Community-based VoteRank method (CVoteRank) to identify a group of influential spreaders from diverse network regions by which the diffusion process is enhanced. Firstly, we measure every node’s spreading ability based on intra- and inter-connectivity structure in a community, which signifies the local and global importance of the node. To identify the seed nodes, we assign the spreading ability to that node’s voting capability and iteratively calculate the voting score of a node based on its neighboring voting capability and its spreading ability. Then, the node acquiring the maximum voting score is identified as the influential spreader in each iteration. Finally, to solve the problem of influence overlapping, CVoteRank reduces the voting capability of the neighboring nodes of the identified spreader. The efficiency of CVoteRank is evaluated and compared with the different state-of-the-art methods on twelve real networks. Utilizing the stochastic susceptible–infected–recovered epidemic method, we calculate the infected scale, final infected scale, and the average shortest path length among the identified spreaders. The experimental results show that CVoteRank identifies the most efficient spreaders with the highest spreading ability within a short period and the maximum reachability, and the identified spreaders are situated at diverse portions of the networks.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"86 ","pages":"Article 102540"},"PeriodicalIF":3.1,"publicationDate":"2025-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143429405","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":"Deep dive into generative models through feature interpoint distances","authors":"Dariusz Jajeśniak ,&nbsp;Piotr Kościelniak,&nbsp;Arkadiusz Zajdel,&nbsp;Marcin Mazur","doi":"10.1016/j.jocs.2025.102539","DOIUrl":"10.1016/j.jocs.2025.102539","url":null,"abstract":"<div><div>This paper introduces the Interpoint Inception Distance (IID) as a new approach for evaluating deep generative models. It is based on reducing the measurement of discrepancy between multidimensional feature distributions to one-dimensional interpoint comparisons. Our method provides a general tool for deriving a wide range of evaluation measures. The Cramér Interpoint Inception Distance (CIID) is notable for its theoretical properties, including a Gaussian-free structure of feature distribution and a strongly consistent estimator. Our experiments, conducted on both synthetic and large-scale real or generated data, suggest that CIID is a promising competitor to the Fréchet Inception Distance (FID), which is currently the primary metric for evaluating deep generative models. This article is an extended version of the ICCS 2024 conference paper (Jajeśniak et al., 2024) <span><span>[1]</span></span>.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"86 ","pages":"Article 102539"},"PeriodicalIF":3.1,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143396220","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 novel approach for overlapping community detection in social networks based on the attraction","authors":"Kuo Chi ,&nbsp;Hui Qu ,&nbsp;Ziheng Fu","doi":"10.1016/j.jocs.2024.102508","DOIUrl":"10.1016/j.jocs.2024.102508","url":null,"abstract":"<div><div>The growing scale of networks makes the study of social networks increasingly difficult. Overlapping community detection can both make the network easier to analyze and manage by detecting communities and better represent the intersection between communities. In this paper, a novel approach for overlapping community detection in social networks is proposed. First, the nodes with local maximum degree are selected from the global network to form initial communities. Next, if the attraction between a community and its surrounding node exceeds a set threshold, these nodes can be directly attracted to that community. Then repeat the above process iteratively until communities no longer change, and nodes that have not yet been divided into communities are regarded as overlapping nodes if they are attracted to two or more communities all greater than the set threshold. In addition, the membership of an overlapping node<del>s</del> in a related community can be calculated by computing the ratio of the attraction of that community to the overlapping node to the sum of the attractions that the node has. Finally, experimental results on 4 synthetic networks and 6 real-world networks show that the proposed algorithm is effective in detecting overlapping communities and performs better compared to some existing algorithms.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"85 ","pages":"Article 102508"},"PeriodicalIF":3.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143177671","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":"Poisson-based framework for predicting count data: Application to traffic counts in Prague areas","authors":"Evženie Uglickich ,&nbsp;Ivan Nagy","doi":"10.1016/j.jocs.2025.102534","DOIUrl":"10.1016/j.jocs.2025.102534","url":null,"abstract":"<div><div>In this paper, we address the task of modeling and predicting count data, with an application to traffic counts on selected urban roads in Prague. We investigated the relationship between multiple counts, designating one of them as the target variable (e.g., data from a key road section) and the others as explanatory counts. Defining traffic count data as the number of vehicles passing through a selected road section per unit of time, we use a framework based on Poisson models to develop a progressive methodology, which we compared with existing models. Working with multimodal count data, we propose the following main steps for the methodology: (i) cluster analysis of explanatory counts using recursive Bayesian estimation of Poisson mixtures; (ii) target count model estimation via local Poisson regressions at identified locations, capturing local relationships between target and explanatory counts; and (iii) prediction of target counts through real-time location detection. The algorithm’s properties were first investigated using simulated data and then validated with real traffic counts. Experimental results indicate that the proposed algorithm outperforms classical Poisson and negative binomial regressions, decision tree and random forest classifiers, as well as a multi-layer perceptron, in predicting traffic count data across various quality metrics, even for weakly correlated data. Applied to traffic count data, the promising performance demonstrated by the proposed algorithm offers an optimistic vision for traffic prediction and urban planning, suggesting its potential as a valuable tool for enhancing transportation efficiency by optimizing the timing of city traffic lights to improve traffic flow.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"85 ","pages":"Article 102534"},"PeriodicalIF":3.1,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143335841","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}