Pub Date : 2024-11-22DOI: 10.1016/j.jocs.2024.102467
Dennis Christensen , Per August Jarval Moen
In Bayesian statistics, the marginal likelihood (ML) is the key ingredient needed for model comparison and model averaging. Unfortunately, estimating MLs accurately is notoriously difficult, especially for models where posterior simulation is not possible. Recently, the idea of permutation counting was introduced, which provides an estimator which can accurately estimate MLs of models for exchangeable binary responses. Such data arise in a multitude of statistical problems, including binary classification, bioassay and sensitivity testing. Permutation counting is entirely likelihood-free and works for any model from which a random sample can be generated, including nonparametric models. Here we present perms, a package implementing permutation counting. Following optimisation efforts, perms is computationally efficient and can handle large data problems. It is available as both an R package and a Python library. A broad gallery of examples illustrating its usage is provided, which includes both standard parametric binary classification and novel applications of nonparametric models, such as changepoint analysis. We also cover the details of the implementation of perms and illustrate its computational speed via a simple simulation study.
在贝叶斯统计中，边际似然（ML）是模型比较和模型平均所需的关键要素。遗憾的是，准确估计 ML 是出了名的困难，尤其是对于无法进行后验模拟的模型。最近，有人提出了置换计数的概念，它提供了一种估计方法，可以准确估计可交换二元响应模型的 ML。这类数据出现在许多统计问题中，包括二元分类、生物测定和灵敏度测试。置换计数完全不需要似然，适用于任何可以生成随机样本的模型，包括非参数模型。这里我们介绍 perms，这是一个实现置换计数的软件包。经过优化，perms 的计算效率很高，可以处理大型数据问题。它既是一个 R 软件包，也是一个 Python 库。我们提供了大量示例来说明它的用法，其中既包括标准参数二元分类，也包括非参数模型的新型应用，如变化点分析。我们还介绍了 perms 的实现细节，并通过简单的模拟研究说明了其计算速度。
{"title":"perms: Likelihood-free estimation of marginal likelihoods for binary response data in Python and R","authors":"Dennis Christensen , Per August Jarval Moen","doi":"10.1016/j.jocs.2024.102467","DOIUrl":"10.1016/j.jocs.2024.102467","url":null,"abstract":"<div><div>In Bayesian statistics, the marginal likelihood (ML) is the key ingredient needed for model comparison and model averaging. Unfortunately, estimating MLs accurately is notoriously difficult, especially for models where posterior simulation is not possible. Recently, the idea of permutation counting was introduced, which provides an estimator which can accurately estimate MLs of models for exchangeable binary responses. Such data arise in a multitude of statistical problems, including binary classification, bioassay and sensitivity testing. Permutation counting is entirely likelihood-free and works for any model from which a random sample can be generated, including nonparametric models. Here we present <span>perms</span>, a package implementing permutation counting. Following optimisation efforts, <span>perms</span> is computationally efficient and can handle large data problems. It is available as both an R package and a Python library. A broad gallery of examples illustrating its usage is provided, which includes both standard parametric binary classification and novel applications of nonparametric models, such as changepoint analysis. We also cover the details of the implementation of <span>perms</span> and illustrate its computational speed via a simple simulation study.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"84 ","pages":"Article 102467"},"PeriodicalIF":3.1,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699237","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}
This paper considers a new approach for people flow estimation in buildings from elevator trip records and corresponding load data, and the resulting model is used on the virtual sensing platform we have developed. People flow data can be used to improve elevator performance through optimal car assignments to hall calls by a group controller and are useful for estimating occupant distributions as heat loads allowing for optimized air-conditioning control to realize energy savings. Available data from an elevator controller is insufficient for exact people flow estimation and therefore this problem becomes under-defined. Our virtual sensing platform adopts equation-based modeling and optimization-based parameter estimation, which estimates application-related parameters from available sensor data, allowing for over- or under-defined situations among sensory information, but better mathematical formulation is essential for accurate parameter estimation on this virtual sensing platform. Accordingly, we propose a new method to define an elevator trip-wise mathematical formulation by modifying pre-defined base equations or defining additional equations. The key idea is that each elevator trip has different features, including sparsity, that are useful for improving accuracy and can be successfully formulated as simultaneous equations that our virtual sensing platform accepts. The procedure for defining a mathematical formulation is invoked after trip data are obtained and we refer this procedure as “on-the-fly mathematical formulation.” The formulated trip-wise equations are combined as simultaneous equations for estimating people flow over a given period on the virtual sensing platform by mathematical optimization.
{"title":"On-the-fly mathematical formulation for estimating people flow from elevator load data in smart building virtual sensing platforms","authors":"Koichi Kondo , Ryosuke Ohori , Kiyotaka Matsue , Hiroyuki Aizu","doi":"10.1016/j.jocs.2024.102488","DOIUrl":"10.1016/j.jocs.2024.102488","url":null,"abstract":"<div><div>This paper considers a new approach for people flow estimation in buildings from elevator trip records and corresponding load data, and the resulting model is used on the virtual sensing platform we have developed. People flow data can be used to improve elevator performance through optimal car assignments to hall calls by a group controller and are useful for estimating occupant distributions as heat loads allowing for optimized air-conditioning control to realize energy savings. Available data from an elevator controller is insufficient for exact people flow estimation and therefore this problem becomes under-defined. Our virtual sensing platform adopts equation-based modeling and optimization-based parameter estimation, which estimates application-related parameters from available sensor data, allowing for over- or under-defined situations among sensory information, but better mathematical formulation is essential for accurate parameter estimation on this virtual sensing platform. Accordingly, we propose a new method to define an elevator trip-wise mathematical formulation by modifying pre-defined base equations or defining additional equations. The key idea is that each elevator trip has different features, including sparsity, that are useful for improving accuracy and can be successfully formulated as simultaneous equations that our virtual sensing platform accepts. The procedure for defining a mathematical formulation is invoked after trip data are obtained and we refer this procedure as “on-the-fly mathematical formulation.” The formulated trip-wise equations are combined as simultaneous equations for estimating people flow over a given period on the virtual sensing platform by mathematical optimization.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"84 ","pages":"Article 102488"},"PeriodicalIF":3.1,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699238","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}
Pub Date : 2024-11-18DOI: 10.1016/j.jocs.2024.102460
Marcin Czajkowski, Krzysztof Jurczuk, Marek Kretowski
This study introduces the Relative Multi-test Classification Tree (RMCT), a novel classification method tailored for multi-omics data analysis. The RMCT method combines the interpretative power of decision trees with the analytical precision of Relative eXpression Analysis (RXA) to address the complex task of examining biomedical data derived from diverse high-throughput technologies. The proposed RMCT approach discerns patterns within and across omics layers, yielding an accurate and interpretable classifier. In each internal node of RMCT, we create a multitest - group of Top-Scoring-Pair tests, that capture the ordering relationships among features from various omics. Multi-tests are optimized for maximal reduction of Gini impurity, and ensuring consistency in decision-making. We address computational challenges by advanced GPU parallelization, remarkably improving RMCT’s time performance. Through experimental validation on diverse multi-omics datasets, RMCT has demonstrated superior performance compared to traditional tree-based solutions, particularly in terms of accuracy and clarity of predictions. This method effectively reveals intricate interactions and relationships within multi-omics data, marking it as a useful addition to bioinformatics and biomedicine. This work represents a thorough extension of our preliminary research, which was initially presented at the twenty-third edition of the International Conference on Computational Science (ICCS). It expands the initial concept of integrating decision trees with RXA for multi-omics data classification, deepening the analytical methodologies, further optimizing the GPU computing, and broadening the experimental validation.
{"title":"Enhancing multi-omics data classification with relative expression analysis and decision trees","authors":"Marcin Czajkowski, Krzysztof Jurczuk, Marek Kretowski","doi":"10.1016/j.jocs.2024.102460","DOIUrl":"10.1016/j.jocs.2024.102460","url":null,"abstract":"<div><div>This study introduces the Relative Multi-test Classification Tree (RMCT), a novel classification method tailored for multi-omics data analysis. The RMCT method combines the interpretative power of decision trees with the analytical precision of Relative eXpression Analysis (RXA) to address the complex task of examining biomedical data derived from diverse high-throughput technologies. The proposed RMCT approach discerns patterns within and across omics layers, yielding an accurate and interpretable classifier. In each internal node of RMCT, we create a multitest - group of Top-Scoring-Pair tests, that capture the ordering relationships among features from various omics. Multi-tests are optimized for maximal reduction of Gini impurity, and ensuring consistency in decision-making. We address computational challenges by advanced GPU parallelization, remarkably improving RMCT’s time performance. Through experimental validation on diverse multi-omics datasets, RMCT has demonstrated superior performance compared to traditional tree-based solutions, particularly in terms of accuracy and clarity of predictions. This method effectively reveals intricate interactions and relationships within multi-omics data, marking it as a useful addition to bioinformatics and biomedicine. This work represents a thorough extension of our preliminary research, which was initially presented at the twenty-third edition of the International Conference on Computational Science (ICCS). It expands the initial concept of integrating decision trees with RXA for multi-omics data classification, deepening the analytical methodologies, further optimizing the GPU computing, and broadening the experimental validation.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"84 ","pages":"Article 102460"},"PeriodicalIF":3.1,"publicationDate":"2024-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699236","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}
Pub Date : 2024-11-15DOI: 10.1016/j.jocs.2024.102473
Shima Esfandiari, Mohammad Reza Moosavi
Identifying influential nodes is crucial in network science for controlling diseases, sharing information, and viral marketing. Current methods for finding vital spreaders have problems with accuracy, resolution, or time complexity. To address these limitations, this paper presents a hybrid approach called the Bubble Method (BM). First, the BM assumes a bubble with a radius of two surrounding each node. Then, it extracts various attributes from inside and near the surface of the bubble. These attributes are the k-shell index, k-shell diversity, and the distances of nodes within the bubble from the central node. We compared our method to 12 recent ones, including the Hybrid Global Structure model (HGSM) and Generalized Degree Decomposition (GDD), using the Susceptible–Infectious–Recovered (SIR) model to test its effectiveness. The results show the BM outperforms other methods in terms of accuracy, correctness, and resolution. Its low computational complexity renders it highly suitable for analyzing large-scale networks.
在网络科学中，识别有影响力的节点对于控制疾病、共享信息和病毒营销至关重要。目前寻找重要传播者的方法在准确性、分辨率或时间复杂性方面存在问题。为了解决这些局限性，本文提出了一种名为 "气泡法"（BM）的混合方法。首先，气泡法假定每个节点周围都有一个半径为 2 的气泡。然后，从气泡内部和表面附近提取各种属性。这些属性包括 k 壳指数、k 壳多样性以及气泡内节点与中心节点的距离。我们使用易感-感染-恢复（SIR）模型，将我们的方法与包括混合全局结构模型（HGSM）和广义度分解（GDD）在内的 12 种最新方法进行了比较，以检验其有效性。结果表明，BM 在准确性、正确性和分辨率方面都优于其他方法。它的计算复杂度低，非常适合分析大规模网络。
{"title":"Identifying influential nodes in complex networks through the k-shell index and neighborhood information","authors":"Shima Esfandiari, Mohammad Reza Moosavi","doi":"10.1016/j.jocs.2024.102473","DOIUrl":"10.1016/j.jocs.2024.102473","url":null,"abstract":"<div><div>Identifying influential nodes is crucial in network science for controlling diseases, sharing information, and viral marketing. Current methods for finding vital spreaders have problems with accuracy, resolution, or time complexity. To address these limitations, this paper presents a hybrid approach called the Bubble Method (BM). First, the BM assumes a bubble with a radius of two surrounding each node. Then, it extracts various attributes from inside and near the surface of the bubble. These attributes are the k-shell index, k-shell diversity, and the distances of nodes within the bubble from the central node. We compared our method to 12 recent ones, including the Hybrid Global Structure model (HGSM) and Generalized Degree Decomposition (GDD), using the Susceptible–Infectious–Recovered (SIR) model to test its effectiveness. The results show the BM outperforms other methods in terms of accuracy, correctness, and resolution. Its low computational complexity renders it highly suitable for analyzing large-scale networks.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"84 ","pages":"Article 102473"},"PeriodicalIF":3.1,"publicationDate":"2024-11-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699234","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}
Pub Date : 2024-11-14DOI: 10.1016/j.jocs.2024.102468
Sophie Robert-Hayek , Soraya Zertal , Philippe Couvée
Black-box auto-tuning methods have been proven to be efficient for tuning configurable computer hardware, including those encountered within the High Performance Computing (HPC) ecosystem. However, because of the shared nature of HPC clusters and the complexity of the software and hardware stacks, the measurement of the performance function can be tainted by noise during the tuning process, which can reduce and sometimes prevent the benefit of the tuning approach. A usual choice for performing the tuning in spite of these interference is to add a resampling step at each iteration to reduce uncertainty, but this approach can be time-consuming and must be done carefully. In this paper, we propose a new resampling and filtering algorithm called EVADyR (Efficient Value Aware Dynamic Resampling). Compared to the state of the art, it finds a better exploration versus exploitation trade-off by resampling only promising configuration and increases the level of confidence around the suggested solution as the tuning process advances. This algorithm was able to tune efficiently two I/O accelerators highly sensitive to interference, in two different scenarios. Compared to Standard Error Dynamic Resampling (SEDR), a state of the art noise reduction strategy, we show that EVADyR is able to reduce the distance to the optimum by 93.5% and 24.7% for the two I/O accelerators respectively, as well as speed-up the experiment duration by 45.8% and 58.1% because less iterations are needed to reach the found optimum. Our results prove the importance of using noise reduction strategies whenever tuning systems running in production.
{"title":"EVADyR: A new dynamic resampling algorithm for auto-tuning noisy High Performance Computing systems","authors":"Sophie Robert-Hayek , Soraya Zertal , Philippe Couvée","doi":"10.1016/j.jocs.2024.102468","DOIUrl":"10.1016/j.jocs.2024.102468","url":null,"abstract":"<div><div>Black-box auto-tuning methods have been proven to be efficient for tuning configurable computer hardware, including those encountered within the High Performance Computing (HPC) ecosystem. However, because of the shared nature of HPC clusters and the complexity of the software and hardware stacks, the measurement of the performance function can be tainted by noise during the tuning process, which can reduce and sometimes prevent the benefit of the tuning approach. A usual choice for performing the tuning in spite of these interference is to add a resampling step at each iteration to reduce uncertainty, but this approach can be time-consuming and must be done carefully. In this paper, we propose a new resampling and filtering algorithm called EVADyR (Efficient Value Aware Dynamic Resampling). Compared to the state of the art, it finds a better exploration versus exploitation trade-off by resampling only promising configuration and increases the level of confidence around the suggested solution as the tuning process advances. This algorithm was able to tune efficiently two I/O accelerators highly sensitive to interference, in two different scenarios. Compared to Standard Error Dynamic Resampling (SEDR), a state of the art noise reduction strategy, we show that EVADyR is able to reduce the distance to the optimum by 93.5% and 24.7% for the two I/O accelerators respectively, as well as speed-up the experiment duration by 45.8% and 58.1% because less iterations are needed to reach the found optimum. Our results prove the importance of using noise reduction strategies whenever tuning systems running in production.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"84 ","pages":"Article 102468"},"PeriodicalIF":3.1,"publicationDate":"2024-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142699235","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}
Pub Date : 2024-11-14DOI: 10.1016/j.jocs.2024.102469
Ruby Varshney , Anjuman Ara Khatun , Haider Hasan Jafri
We study the transition to phase synchronization in an ensemble of Stuart–Landau oscillators interacting on a star network. We observe that by introducing frequency-weighted coupling and timescale variations in the dynamics of nodes, the system exhibits a first-order explosive transition to phase synchrony. Further, we extend this study to understand the nature of synchronization in the case of two coupled star networks. If the coupled star networks are identical, we observe that with increasing inter-star coupling strength, the hysteresis width initially increases, reaches a maximum value, then decreases before saturating. If the interacting star networks are non-identical, we observe that the transition to the coherent state is preceded by the occurrence of intermittent in-phase and anti-phase synchrony for small inter-star coupling. However, for large values of coupling strengths, we observe that the intermittent state disappears and the hysteresis width changes as in coupled identical star networks. We characterize these transitions by plotting the Lyapunov exponents for the system and the master stability function.
{"title":"Explosive synchronization in interacting star networks","authors":"Ruby Varshney , Anjuman Ara Khatun , Haider Hasan Jafri","doi":"10.1016/j.jocs.2024.102469","DOIUrl":"10.1016/j.jocs.2024.102469","url":null,"abstract":"<div><div>We study the transition to phase synchronization in an ensemble of Stuart–Landau oscillators interacting on a star network. We observe that by introducing frequency-weighted coupling and timescale variations in the dynamics of nodes, the system exhibits a first-order explosive transition to phase synchrony. Further, we extend this study to understand the nature of synchronization in the case of two coupled star networks. If the coupled star networks are identical, we observe that with increasing inter-star coupling strength, the hysteresis width initially increases, reaches a maximum value, then decreases before saturating. If the interacting star networks are non-identical, we observe that the transition to the coherent state is preceded by the occurrence of intermittent in-phase and anti-phase synchrony for small inter-star coupling. However, for large values of coupling strengths, we observe that the intermittent state disappears and the hysteresis width changes as in coupled identical star networks. We characterize these transitions by plotting the Lyapunov exponents for the system and the master stability function.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"83 ","pages":"Article 102469"},"PeriodicalIF":3.1,"publicationDate":"2024-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142663389","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}
We develop a novel fourth-order compact finite difference scheme to solve nonlinear singular ordinary differential equations. Such problems occur in many fields of science and engineering, such as studying the equilibrium of an isothermal gas sphere, reaction–diffusion in a spherical permeable catalyst, etc. These problems are challenging to solve because of their singularity or nonlinearity. By our proposed method, we can easily solve these complex problems without removing or modifying the singularity. To construct the new fourth-order compact difference method, Initially, we created a uniform mesh within the solution domain and developed a compact finite difference scheme. This scheme approximates the derivatives at the boundary nodal points to handle the problem’s singularity effectively. Employing a matrix analysis approach, we discussed the convergence analysis of the methods. To demonstrate its efficacy, we apply our approach to solve various real-life problems from the literature. The new method offers high-order accuracy with minimal grid points and provides better numerical results than the nonstandard finite difference method and exponential compact finite difference method.
{"title":"A new fourth-order compact finite difference method for solving Lane-Emden-Fowler type singular boundary value problems","authors":"Nirupam Sahoo , Randhir Singh , Ankur Kanaujiya , Carlo Cattani","doi":"10.1016/j.jocs.2024.102474","DOIUrl":"10.1016/j.jocs.2024.102474","url":null,"abstract":"<div><div>We develop a novel fourth-order compact finite difference scheme to solve nonlinear singular ordinary differential equations. Such problems occur in many fields of science and engineering, such as studying the equilibrium of an isothermal gas sphere, reaction–diffusion in a spherical permeable catalyst, etc. These problems are challenging to solve because of their singularity or nonlinearity. By our proposed method, we can easily solve these complex problems without removing or modifying the singularity. To construct the new fourth-order compact difference method, Initially, we created a uniform mesh within the solution domain and developed a compact finite difference scheme. This scheme approximates the derivatives at the boundary nodal points to handle the problem’s singularity effectively. Employing a matrix analysis approach, we discussed the convergence analysis of the methods. To demonstrate its efficacy, we apply our approach to solve various real-life problems from the literature. The new method offers high-order accuracy with minimal grid points and provides better numerical results than the nonstandard finite difference method and exponential compact finite difference method.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"83 ","pages":"Article 102474"},"PeriodicalIF":3.1,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142663387","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}
Pub Date : 2024-11-12DOI: 10.1016/j.jocs.2024.102466
Arun Govind Neelan , G. Sai Krishna , Vinoth Paramanantham
Addressing discontinuities in fluid flow problems is inherently difficult, especially when shocks arise due to the nonlinear nature of the flow. While handling discontinuities is a well-established practice in computational fluid dynamics (CFD), it remains a major challenge when applying physics-informed neural networks (PINNs). In this study, we compare the shock-resolving capabilities of traditional CFD methods with those of PINNs, highlighting the advantages of the latter. Our findings show that PINNs exhibit less dissipative behavior compared to conventional techniques. We evaluated the performance of both PINNs and traditional methods on linear and nonlinear test cases, demonstrating that PINNs offer superior shock-resolving properties. Notably, PINNs can accurately resolve inviscid shocks with just three grid points, whereas traditional methods require at least seven points. This suggests that PINNs are more effective at resolving shocks and discontinuities when using the same grid for both PINN and CFD simulations. However, it is important to note that PINNs, in this context, are computationally more expensive than traditional methods on a given grid.
{"title":"Physics-informed neural networks and higher-order high-resolution methods for resolving discontinuities and shocks: A comprehensive study","authors":"Arun Govind Neelan , G. Sai Krishna , Vinoth Paramanantham","doi":"10.1016/j.jocs.2024.102466","DOIUrl":"10.1016/j.jocs.2024.102466","url":null,"abstract":"<div><div>Addressing discontinuities in fluid flow problems is inherently difficult, especially when shocks arise due to the nonlinear nature of the flow. While handling discontinuities is a well-established practice in computational fluid dynamics (CFD), it remains a major challenge when applying physics-informed neural networks (PINNs). In this study, we compare the shock-resolving capabilities of traditional CFD methods with those of PINNs, highlighting the advantages of the latter. Our findings show that PINNs exhibit less dissipative behavior compared to conventional techniques. We evaluated the performance of both PINNs and traditional methods on linear and nonlinear test cases, demonstrating that PINNs offer superior shock-resolving properties. Notably, PINNs can accurately resolve inviscid shocks with just three grid points, whereas traditional methods require at least seven points. This suggests that PINNs are more effective at resolving shocks and discontinuities when using the same grid for both PINN and CFD simulations. However, it is important to note that PINNs, in this context, are computationally more expensive than traditional methods on a given grid.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"83 ","pages":"Article 102466"},"PeriodicalIF":3.1,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142663591","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}
Pub Date : 2024-11-12DOI: 10.1016/j.jocs.2024.102465
Deepika Shukla, C. Ravindranath Chowdary
In academia, research collaboration plays a vital role in enhancing the research quality and enriching the academic profile of the authors. Recommending appropriate collaborators from a vast scholarly database, particularly for newcomers, poses a challenging cold-start problem. This study addresses a cold-start problem in peer recommendation, considering a dynamic coauthorship graph as a network structure of academic collaborators. As the coauthorship graph is quite large and complex, an efficient indexing method is essential for speeding up the initial search of similar coauthors. The study introduces an efficient Global Inverted List for indexing research areas and active authors in the coauthorship network. An attribute-based search and filtering mechanism is proposed to identify relevant collaborators, followed by the application of k-means clustering and doc2vec metrics to rank and select top recommendations. A cold user is associated with attributes that identify coauthors with similar research interests. For each attribute of the cold user, the model searches the associated authors from the GIL. Further, two filtering approaches are applied to refine the retrieved author list. The first ensures that the authors have a significant presence in the specified research areas, whereas the second one helps avoid recommending authors with only superficial connections to the cold user. The model creates a feature matrix of filtered authors using the publication features of authors. The k-means clustering applied to the feature matrix generates clusters, among which the model chooses only those with seed nodes i.e. the clusters which are having seed nodes are selected for further process. Selected clusters are ranked using doc2vec metrics, with the top-ranked cluster providing the final recommendation. The model recommends the top members of the selected cluster, where is the length of the recommendations provided to the new user. Our extensive experiments show the efficacy of the proposed model.
在学术界，研究合作对提高研究质量和丰富作者的学术形象起着至关重要的作用。从庞大的学术数据库中推荐合适的合作者，尤其是对于新人来说，是一个具有挑战性的冷启动问题。本研究将动态共同作者图视为学术合作者的网络结构，从而解决了同行推荐中的冷启动问题。由于共同作者图谱相当庞大和复杂，高效的索引方法对于加快相似共同作者的初始搜索至关重要。本研究介绍了一种高效的全局反向列表（GIL），用于索引共同作者网络中的研究领域和活跃作者。研究提出了一种基于属性的搜索和过滤机制来识别相关的合作者，然后应用 k-means 聚类和 doc2vec 指标来排列和选择顶级推荐。冷用户与识别具有相似研究兴趣的合作者的属性相关联。对于冷用户的每个属性，模型都会从 GIL 中搜索相关作者。此外，还采用了两种过滤方法来完善检索到的作者列表。第一种方法确保作者在指定的研究领域有重要影响力，而第二种方法则有助于避免推荐与冷用户只有表面联系的作者。该模型利用作者的发表特征创建了筛选作者的特征矩阵。对特征矩阵进行 k-means 聚类生成 k 个聚类，模型只选择其中有种子节点的聚类，即选择有种子节点的聚类进行下一步处理。选定的聚类使用 doc2vec 指标进行排名，排名靠前的聚类提供最终推荐。该模型推荐所选簇中排名前 L 的成员，其中 L 是向新用户提供的推荐的长度。我们的大量实验证明了所建议模型的有效性。
{"title":"A model to address the cold-start in peer recommendation by using k-means clustering and sentence embedding","authors":"Deepika Shukla, C. Ravindranath Chowdary","doi":"10.1016/j.jocs.2024.102465","DOIUrl":"10.1016/j.jocs.2024.102465","url":null,"abstract":"<div><div>In academia, research collaboration plays a vital role in enhancing the research quality and enriching the academic profile of the authors. Recommending appropriate collaborators from a vast scholarly database, particularly for newcomers, poses a challenging cold-start problem. This study addresses a cold-start problem in peer recommendation, considering a dynamic coauthorship graph as a network structure of academic collaborators. As the coauthorship graph is quite large and complex, an efficient indexing method is essential for speeding up the initial search of similar coauthors. The study introduces an efficient Global Inverted List <span><math><mrow><mo>(</mo><mi>G</mi><mi>I</mi><mi>L</mi><mo>)</mo></mrow></math></span> for indexing research areas and active authors in the coauthorship network. An attribute-based search and filtering mechanism is proposed to identify relevant collaborators, followed by the application of k-means clustering and doc2vec metrics to rank and select top recommendations. A cold user is associated with attributes that identify coauthors with similar research interests. For each attribute of the cold user, the model searches the associated authors from the GIL. Further, two filtering approaches are applied to refine the retrieved author list. The first ensures that the authors have a significant presence in the specified research areas, whereas the second one helps avoid recommending authors with only superficial connections to the cold user. The model creates a feature matrix of filtered authors using the publication features of authors. The k-means clustering applied to the feature matrix generates <span><math><mi>k</mi></math></span> clusters, among which the model chooses only those with seed nodes i.e. the clusters which are having seed nodes are selected for further process. Selected clusters are ranked using doc2vec metrics, with the top-ranked cluster providing the final recommendation. The model recommends the top <span><math><mi>L</mi></math></span> members of the selected cluster, where <span><math><mi>L</mi></math></span> is the length of the recommendations provided to the new user. Our extensive experiments show the efficacy of the proposed model.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"83 ","pages":"Article 102465"},"PeriodicalIF":3.1,"publicationDate":"2024-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142663388","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}
Pub Date : 2024-11-10DOI: 10.1016/j.jocs.2024.102461
Suyue Han , Bin Liu , Jun Shu , Zuli He , Xinyu Xia , Ke Pan , Hourui Ren
Strong geological disasters have caused persistent losses in society, economy, and ecological environments. Given the unique geographical settings of the stricken areas, their resilience is prone to damage or even loss. Comprehensive risk assessment of natural disasters is the core content and important foundation for building regional resilience. Therefore, conducting dynamic characteristics analysis of resilience in mountainous disaster areas impacted by strong earthquake geological disasters is vital for ensuring the region's high-quality and sustainable development. This article takes the 51 stricken areas of Wenchuan earthquake as the research object. To this end, social, economic and ecological environmental data from 2008 to 2020 was hereby collected. Initially, a regional resilience assessment system based on "socio-economic-ecological environment" was established, considering the long-term and spatial heterogeneity of geological disasters. Secondly, the regional resilience assessment model was constructed using Spectral clustering-genetic algorithm-improved entropy weight method. Following that, the dynamic characteristics of regional resilience were quantitatively analyzed from two aspects, including change velocity state and change rate trend. Finally, based on the regional resilience characteristics, differentiated resilience enhancement strategies were proposed. Collectively, the results revealed that: (1) From a geological disaster standpoint, the risk in post-earthquake disaster areas exhibited a strikingly rapid decline, with the spatial distribution of geological disaster risk being notably higher in the central areas and diminishing towards the peripheries. (2) Overall, the regional resilience of the 51 stricken areas showed a "V-shaped" trend, with a significant upturn since 2012. (3) From the perspective of dynamic characteristics, more counties (cities) presented an upward trend. (4) The 51 stricken areas were hereby divided into the "benchmarking type", the "declination type", the "backward type", and the "potential type". In conclusion, the current study enhances the technical framework for evaluating regional resilience and provides technical support for the construction of resilient cities.
{"title":"Quantitative assessment and dynamic characteristic measurement of regional resilience: From the perspective of post-earthquakes effects","authors":"Suyue Han , Bin Liu , Jun Shu , Zuli He , Xinyu Xia , Ke Pan , Hourui Ren","doi":"10.1016/j.jocs.2024.102461","DOIUrl":"10.1016/j.jocs.2024.102461","url":null,"abstract":"<div><div>Strong geological disasters have caused persistent losses in society, economy, and ecological environments. Given the unique geographical settings of the stricken areas, their resilience is prone to damage or even loss. Comprehensive risk assessment of natural disasters is the core content and important foundation for building regional resilience. Therefore, conducting dynamic characteristics analysis of resilience in mountainous disaster areas impacted by strong earthquake geological disasters is vital for ensuring the region's high-quality and sustainable development. This article takes the 51 stricken areas of Wenchuan earthquake as the research object. To this end, social, economic and ecological environmental data from 2008 to 2020 was hereby collected. Initially, a regional resilience assessment system based on \"socio-economic-ecological environment\" was established, considering the long-term and spatial heterogeneity of geological disasters. Secondly, the regional resilience assessment model was constructed using Spectral clustering-genetic algorithm-improved entropy weight method. Following that, the dynamic characteristics of regional resilience were quantitatively analyzed from two aspects, including change velocity state and change rate trend. Finally, based on the regional resilience characteristics, differentiated resilience enhancement strategies were proposed. Collectively, the results revealed that: (1) From a geological disaster standpoint, the risk in post-earthquake disaster areas exhibited a strikingly rapid decline, with the spatial distribution of geological disaster risk being notably higher in the central areas and diminishing towards the peripheries. (2) Overall, the regional resilience of the 51 stricken areas showed a \"V-shaped\" trend, with a significant upturn since 2012. (3) From the perspective of dynamic characteristics, more counties (cities) presented an upward trend. (4) The 51 stricken areas were hereby divided into the \"benchmarking type\", the \"declination type\", the \"backward type\", and the \"potential type\". In conclusion, the current study enhances the technical framework for evaluating regional resilience and provides technical support for the construction of resilient cities.</div></div>","PeriodicalId":48907,"journal":{"name":"Journal of Computational Science","volume":"83 ","pages":"Article 102461"},"PeriodicalIF":3.1,"publicationDate":"2024-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142663613","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号