Pub Date : 2026-05-01Epub Date: 2026-01-25DOI: 10.1016/j.cpc.2026.110052
Danh Nam Nguyen , Jae Hun Lee , Chun Sang Yoo
<div><div>The official OpenFOAM distributions are currently not well-suited for accurate simulations of laminar reacting flows, primarily due to the restrictive Sutherland transport model and the oversimplified unity Lewis number assumption. These limitations can be addressed by employing a detailed transport model (DTM) grounded in kinetic gas theory. However, this approach significantly increases computational cost. To resolve this trade-off, we present a newly developed framework, <em>DTLreactingFoam</em>, designed for simulating laminar flames with integrated detailed transport and chemical kinetics while maintaining computational efficiency. The first level of cost reduction is achieved by incorporating a polynomial-fit transport model (FTM). Further acceleration is provided by a time-correlated thermophysical property evaluation (coTHERM) method, which dynamically updates properties at each time step or iteration by exploiting their temporal correlations. The framework is validated through a series of canonical laminar flame simulations. The results show excellent agreement with experimental measurements and benchmark software, confirming the accurate implementation of both the DTM and FTM. Moreover, validation results demonstrate that coupling the coTHERM method with either the DTM or FTM enables high-fidelity laminar flame simulations with substantially reduced computational cost. Notably, using the coTHERM method in conjunction with the FTM achieves up to a 77% reduction in computational time compared to the direct use of the DTM, without compromising accuracy.</div><div><strong>PROGRAM SUMMARY</strong> <em>Program Title:</em> DTLreactingFoam <em>CPC Library link to program files:</em> (to be added by Technical Editor) <em>Developer’s repository link (OF-12):</em> <span><span>https://github.com/danhnam11/DTLreactingFoam-12</span><svg><path></path></svg></span> <em>Developer’s repository link (OF-10):</em> <span><span>https://github.com/danhnam11/DTLreactingFoam-10</span><svg><path></path></svg></span> <em>Developer’s repository link (OF-8):</em> <span><span>https://github.com/danhnam11/DTLreactingFoam-8</span><svg><path></path></svg></span> <em>Code Ocean capsule:</em> (to be added by Technical Editor) <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> C++ <em>Supplementary material: Nature of problem:</em> Using the detailed transport model (DTM) based on the principles of kinetic gas theory can accurately simulate laminar reacting flows in OpenFOAM (OF). However, the accuracy comes at the cost of significantly greater computational effort since all thermophyscal properties are recomputed in every single cell and at every time step throughout the simulation when using DTM in OF. <em>Solution method:</em> In reacting flow simulations, the evolution of thermodynamic state variables and species concentrations between successive steps are correlated. The change in these quantities from one step to the next are often minimal
{"title":"DTLreactingFoam: An efficient CFD tool for laminar reacting flow simulations using detailed chemistry and transport with time-correlated thermophysical properties","authors":"Danh Nam Nguyen , Jae Hun Lee , Chun Sang Yoo","doi":"10.1016/j.cpc.2026.110052","DOIUrl":"10.1016/j.cpc.2026.110052","url":null,"abstract":"<div><div>The official OpenFOAM distributions are currently not well-suited for accurate simulations of laminar reacting flows, primarily due to the restrictive Sutherland transport model and the oversimplified unity Lewis number assumption. These limitations can be addressed by employing a detailed transport model (DTM) grounded in kinetic gas theory. However, this approach significantly increases computational cost. To resolve this trade-off, we present a newly developed framework, <em>DTLreactingFoam</em>, designed for simulating laminar flames with integrated detailed transport and chemical kinetics while maintaining computational efficiency. The first level of cost reduction is achieved by incorporating a polynomial-fit transport model (FTM). Further acceleration is provided by a time-correlated thermophysical property evaluation (coTHERM) method, which dynamically updates properties at each time step or iteration by exploiting their temporal correlations. The framework is validated through a series of canonical laminar flame simulations. The results show excellent agreement with experimental measurements and benchmark software, confirming the accurate implementation of both the DTM and FTM. Moreover, validation results demonstrate that coupling the coTHERM method with either the DTM or FTM enables high-fidelity laminar flame simulations with substantially reduced computational cost. Notably, using the coTHERM method in conjunction with the FTM achieves up to a 77% reduction in computational time compared to the direct use of the DTM, without compromising accuracy.</div><div><strong>PROGRAM SUMMARY</strong> <em>Program Title:</em> DTLreactingFoam <em>CPC Library link to program files:</em> (to be added by Technical Editor) <em>Developer’s repository link (OF-12):</em> <span><span>https://github.com/danhnam11/DTLreactingFoam-12</span><svg><path></path></svg></span> <em>Developer’s repository link (OF-10):</em> <span><span>https://github.com/danhnam11/DTLreactingFoam-10</span><svg><path></path></svg></span> <em>Developer’s repository link (OF-8):</em> <span><span>https://github.com/danhnam11/DTLreactingFoam-8</span><svg><path></path></svg></span> <em>Code Ocean capsule:</em> (to be added by Technical Editor) <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> C++ <em>Supplementary material: Nature of problem:</em> Using the detailed transport model (DTM) based on the principles of kinetic gas theory can accurately simulate laminar reacting flows in OpenFOAM (OF). However, the accuracy comes at the cost of significantly greater computational effort since all thermophyscal properties are recomputed in every single cell and at every time step throughout the simulation when using DTM in OF. <em>Solution method:</em> In reacting flow simulations, the evolution of thermodynamic state variables and species concentrations between successive steps are correlated. The change in these quantities from one step to the next are often minimal","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"322 ","pages":"Article 110052"},"PeriodicalIF":3.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146057321","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2026-01-21DOI: 10.1016/j.cpc.2026.110036
Sojeong Park , Wei Liu , Simon Julian Lauw , Wooseop Kwak , Chandra S. Verma , Hwee Kuan Lee
Molecular dynamics (MD) simulation is an essential tool for condensed matter physics, materials science, structural/mechanistic biology, and multi-agent systems. Despite their successes, traditional numerical integration methods for solving Hamilton’s equations of motion are computationally intensive, limiting simulations to short time scales. Recent advancements in machine learning have opened new avenues for accelerating MD simulations. This work introduces the Local Update Function network (LUFnet), a transformer-based neural network designed to increase time integration step sizes significantly while maintaining simulation stability and accuracy. LUFnet integrates local spatial and temporal information, enabling efficient rollout for long-time-scale simulations. By preserving key symmetries such as translational invariance, Galilean coordinate transformation invariance, and particle exchange symmetry, LUFnet achieves robust performance across different thermodynamic states. LUFnet is designed to accommodate general physical models (e.g., Lennard-Jones, Coulomb potential, on lattice systems). Its framework allows the model to be trained on small systems and directly applied to larger systems, maintaining computation efficiency and computation memory usage that scales linearly with the number of particles. Benchmarked on Lennard-Jones systems, LUFnet demonstrated minimal accuracy degradation even after rollout over many large time integration steps, offering an effective approach to molecular dynamics simulations.
{"title":"Scalable neural network driven molecular dynamics simulation","authors":"Sojeong Park , Wei Liu , Simon Julian Lauw , Wooseop Kwak , Chandra S. Verma , Hwee Kuan Lee","doi":"10.1016/j.cpc.2026.110036","DOIUrl":"10.1016/j.cpc.2026.110036","url":null,"abstract":"<div><div>Molecular dynamics (MD) simulation is an essential tool for condensed matter physics, materials science, structural/mechanistic biology, and multi-agent systems. Despite their successes, traditional numerical integration methods for solving Hamilton’s equations of motion are computationally intensive, limiting simulations to short time scales. Recent advancements in machine learning have opened new avenues for accelerating MD simulations. This work introduces the Local Update Function network (LUFnet), a transformer-based neural network designed to increase time integration step sizes significantly while maintaining simulation stability and accuracy. LUFnet integrates local spatial and temporal information, enabling efficient rollout for long-time-scale simulations. By preserving key symmetries such as translational invariance, Galilean coordinate transformation invariance, and particle exchange symmetry, LUFnet achieves robust performance across different thermodynamic states. LUFnet is designed to accommodate general physical models (e.g., Lennard-Jones, Coulomb potential, on lattice systems). Its framework allows the model to be trained on small systems and directly applied to larger systems, maintaining computation efficiency and computation memory usage that scales linearly with the number of particles. Benchmarked on Lennard-Jones systems, LUFnet demonstrated minimal accuracy degradation even after rollout over many large time integration steps, offering an effective approach to molecular dynamics simulations.</div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"322 ","pages":"Article 110036"},"PeriodicalIF":3.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076756","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We present the first public release of plasmonX, a novel open-source code for simulating the plasmonic response of complex nanostructures. The code supports both fully atomistic and implicit descriptions of nanomaterials. In particular, it employs the frequency-dependent fluctuating charges (ωFQ) and dipoles (ωFQFμ) models to describe the response properties of atomistic structures, including simple and d-metals, graphene-based structures, and multi-metal nanostructures. For implicit representations, the Boundary Element Method is implemented in both the dielectric polarizable continuum model (DPCM) and integral equation formalism (IEF-PCM) variants. The distribution also includes a post-processing module that enables analysis of electric field-induced properties such as charge density and electric field patterns.
PROGRAM SUMMARY
Program Title: plasmonX CPC Library link to program files:https://doi.org/10.17632/zcd8fb4457.1Developer’s repository link:https://github.com/plasmonX/plasmonXLicensing provisions: GPLv3 Programming language: Fortran 2008, Python Nature of problem: Simulating the response properties of plasmonic metallic and graphene-based nanomaterials. Solution method: Fully atomistic frequency-dependent fluctuating charges (ωFQ) [1,2] and dipoles (ωFQFμ) [3] models and implicit, non-atomistic Boundary Element Methods (BEM) [4]. The approaches are implemented within the quasistatic approximation. Additional comments including restrictions and unusual features:The program has been mainly tested by using gfortran (versions 9–13) combined with the Math Kernel Library (MKL) provided by Intel.
References:
[1
]T. Giovannini, M. Rosa, S. Corni, C. Cappelli, A classical picture of subnanometer junctions: an atomistic Drude approach to nanoplasmonics, Nanoscale 11 (13) (2019) 6004-6015
[2
]T. Giovannini, L. Bonatti, M. Polini, C. Cappelli, Graphene plasmonics: Fully atomistic approach for realistic structures, J. Phys. Chem. Lett. 11 (18) (2020) 7595-7602.
[3
]T. Giovannini, L. Bonatti, P. Lafiosca, L. Nicoli, M. Castagnola, P. G. Illobre, S. Corni, C. Cappelli, Do we really need quantum mechanics to describe plasmonic properties of metal nanostructures?, ACS Photonics 9 (9) (2022) 3025-3034.
[4
]F. J. García de Abajo, A. Howie, Retarded field calculation of electron energy loss in inhomogeneous dielectrics, Phys. Rev. B 65 (11) (2002) 115418.
我们提出了第一个公开发布的plasmonX，一个新颖的开源代码，用于模拟复杂纳米结构的等离子体响应。代码支持纳米材料的完全原子描述和隐式描述。特别是，它采用频率相关的波动电荷（ωFQ）和偶极子（ωFQFμ）模型来描述原子结构的响应特性，包括简单金属和d金属，石墨烯基结构和多金属纳米结构。对于隐式表示，边界元方法在介电极化连续介质模型（DPCM）和积分方程形式（IEF-PCM）变体中实现。该分布还包括一个后处理模块，可以分析电场诱导的特性，如电荷密度和电场模式。项目摘要项目标题：plasmonX CPC库链接到程序文件：https://doi.org/10.17632/zcd8fb4457.1开发人员的存储库链接：https://github.com/plasmonX/plasmonX许可条款：GPLv3编程语言：Fortran 2008， Python问题的性质：模拟等离子体金属和石墨烯基纳米材料的响应特性。求解方法：全原子频率相关波动电荷（ωFQ）[1,2]和偶极子（ωFQFμ）[3]模型和隐式非原子边界元方法（BEM）[4]。这些方法是在准静态近似内实现的。该程序主要通过使用gfortran（版本9-13）和intel提供的数学内核库（MKL）进行测试。参考：[1]T。张晓明，张晓明，张晓明，等。亚纳米等离子体动力学的研究进展，光子学报，36 (5)(2019):649 - 649 [j] . [j]。李建军，李建军，李建军，等。石墨烯等离子体：真实结构的全原子方法，物理学报。化学。科学通报，11(18)(2020)7595-7602。Giovannini， L. Bonatti， P. Lafiosca， L. Nicoli， M. Castagnola， P. G. Illobre， S. Corni， C. Cappelli，我们真的需要量子力学来描述金属纳米结构的等离子体特性吗？王晓明，光电子学报9(9)(2022):3025-3034。J. García de Abajo， A. Howie，非均匀介质中电子能量损失的延迟场计算，物理学报。Rev. b65(11)(2002) 115418。
{"title":"plasmonX: An open-source code for nanoplasmonics","authors":"Tommaso Giovannini , Pablo Grobas Illobre , Piero Lafiosca , Luca Nicoli , Luca Bonatti , Stefano Corni , Chiara Cappelli","doi":"10.1016/j.cpc.2026.110035","DOIUrl":"10.1016/j.cpc.2026.110035","url":null,"abstract":"<div><div>We present the first public release of <span>plasmonX</span>, a novel open-source code for simulating the plasmonic response of complex nanostructures. The code supports both fully atomistic and implicit descriptions of nanomaterials. In particular, it employs the frequency-dependent fluctuating charges (<em>ω</em>FQ) and dipoles (<em>ω</em>FQF<em>μ</em>) models to describe the response properties of atomistic structures, including simple and <em>d</em>-metals, graphene-based structures, and multi-metal nanostructures. For implicit representations, the Boundary Element Method is implemented in both the dielectric polarizable continuum model (DPCM) and integral equation formalism (IEF-PCM) variants. The distribution also includes a post-processing module that enables analysis of electric field-induced properties such as charge density and electric field patterns.</div></div><div><h3>PROGRAM SUMMARY</h3><div><em>Program Title:</em> plasmonX <em>CPC Library link to program files:</em> <span><span>https://doi.org/10.17632/zcd8fb4457.1</span><svg><path></path></svg></span> <em>Developer’s repository link:</em> <span><span>https://github.com/plasmonX/plasmonX</span><svg><path></path></svg></span> <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> Fortran 2008, Python <em>Nature of problem:</em> Simulating the response properties of plasmonic metallic and graphene-based nanomaterials. <em>Solution method:</em> Fully atomistic frequency-dependent fluctuating charges (<em>ω</em>FQ) [1,2] and dipoles (<em>ω</em>FQF<em>μ</em>) [3] models and implicit, non-atomistic Boundary Element Methods (BEM) [4]. The approaches are implemented within the quasistatic approximation. <em>Additional comments including restrictions and unusual features:</em>The program has been mainly tested by using gfortran (versions 9–13) combined with the Math Kernel Library (MKL) provided by Intel.</div><div><strong>References:</strong><ul><li><span>[1</span><span><div>]<em>T. Giovannini, M. Rosa, S. Corni, C. Cappelli, A classical picture of subnanometer junctions: an atomistic Drude approach to nanoplasmonics, Nanoscale 11 (13) (2019) 6004-6015</em></div></span></li><li><span>[2</span><span><div>]<em>T. Giovannini, L. Bonatti, M. Polini, C. Cappelli, Graphene plasmonics: Fully atomistic approach for realistic structures, J. Phys. Chem. Lett. 11 (18) (2020) 7595-7602.</em></div></span></li><li><span>[3</span><span><div>]<em>T. Giovannini, L. Bonatti, P. Lafiosca, L. Nicoli, M. Castagnola, P. G. Illobre, S. Corni, C. Cappelli, Do we really need quantum mechanics to describe plasmonic properties of metal nanostructures?, ACS Photonics 9 (9) (2022) 3025-3034.</em></div></span></li><li><span>[4</span><span><div>]<em>F. J. García de Abajo, A. Howie, Retarded field calculation of electron energy loss in inhomogeneous dielectrics, Phys. Rev. B 65 (11) (2002) 115418.</em></div></span></li></ul></div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"322 ","pages":"Article 110035"},"PeriodicalIF":3.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076721","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-05-01Epub Date: 2026-01-21DOI: 10.1016/j.cpc.2026.110039
Ping-Hsuan Tsai , Seung Whan Chung , Debojyoti Ghosh , John Loffeld , Youngsoo Choi , Jonathan L. Belof
Despite advancements in high-performance computing and modern numerical algorithms, computational cost remains prohibitive for multi-query kinetic plasma simulations. In this work, we develop data-driven reduced-order models (ROMs) for collisionless electrostatic plasma dynamics, based on the kinetic Vlasov-Poisson equation. Our ROM approach projects the equation onto a linear subspace defined by the proper orthogonal decomposition (POD) modes. We introduce an efficient tensorial method to update the nonlinear term using a precomputed third-order tensor. We capture multiscale behavior with a minimal number of POD modes by decomposing the solution manifold into multiple time windows and creating temporally local ROMs. We consider two strategies for decomposition: one based on the physical time and the other based on the electric field energy. Applied to the 1D1V Vlasov–Poisson simulations, that is, prescribed E-field, Landau damping, and two-stream instability, we demonstrate that our ROMs accurately capture the total energy of the system both for parametric and time extrapolation cases. The temporally local ROMs are more efficient and accurate than the single ROM. In addition, in the two-stream instability case, we show that the energy-windowing reduced-order model (EW-ROM) is more efficient and accurate than the time-windowing reduced-order model (TW-ROM). With the tensorial approach, EW-ROM solves the equation approximately 90 times faster than Eulerian simulations while maintaining a maximum relative error of 7.5% for the training data and 11% for the testing data.
{"title":"Local reduced-order modeling for electrostatic plasmas by physics-informed solution manifold decomposition","authors":"Ping-Hsuan Tsai , Seung Whan Chung , Debojyoti Ghosh , John Loffeld , Youngsoo Choi , Jonathan L. Belof","doi":"10.1016/j.cpc.2026.110039","DOIUrl":"10.1016/j.cpc.2026.110039","url":null,"abstract":"<div><div>Despite advancements in high-performance computing and modern numerical algorithms, computational cost remains prohibitive for multi-query kinetic plasma simulations. In this work, we develop data-driven reduced-order models (ROMs) for collisionless electrostatic plasma dynamics, based on the kinetic Vlasov-Poisson equation. Our ROM approach projects the equation onto a linear subspace defined by the proper orthogonal decomposition (POD) modes. We introduce an efficient tensorial method to update the nonlinear term using a precomputed third-order tensor. We capture multiscale behavior with a minimal number of POD modes by decomposing the solution manifold into multiple time windows and creating temporally local ROMs. We consider two strategies for decomposition: one based on the physical time and the other based on the electric field energy. Applied to the 1D1V Vlasov–Poisson simulations, that is, prescribed E-field, Landau damping, and two-stream instability, we demonstrate that our ROMs accurately capture the total energy of the system both for parametric and time extrapolation cases. The temporally local ROMs are more efficient and accurate than the single ROM. In addition, in the two-stream instability case, we show that the energy-windowing reduced-order model (EW-ROM) is more efficient and accurate than the time-windowing reduced-order model (TW-ROM). With the tensorial approach, EW-ROM solves the equation approximately 90 times faster than Eulerian simulations while maintaining a maximum relative error of 7.5% for the training data and 11% for the testing data.</div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"322 ","pages":"Article 110039"},"PeriodicalIF":3.4,"publicationDate":"2026-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076755","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2025-12-31DOI: 10.1016/j.cpc.2025.110013
Han Li , Ke Xiao , Yangchen Xu , Haoze Zhang , Zhenyi Chen , Runze Mao , Zhi X. Chen
<div><div>Recent advances in deep learning for solving flame chemical kinetics offer promising solutions to the long-standing trade-off between accuracy and computational efficiency in combustion simulations. This work introduces DFODE-kit, an open-source Python package designed to replace the conventional, computationally intensive integration of chemical source terms governed by high-dimensional, stiff ordinary differential equations (ODEs), thereby substantially accelerating chemistry evaluation in combustion simulations. The package provides: i) an efficient sampling module that extracts high-quality thermochemical states from low-dimensional manifolds in canonical flames; ii) an effective data augmentation module that enriches the dataset to approximate the high-dimensional composition space encountered in turbulent flames; and (iii) an optimized neural network training module with multiscale preprocessing and physics-informed constraints to enhance model fidelity and stability. The trained models are seamlessly integrated into our previously released CFD solver DeepFlame<span><span><sup>1</sup></span></span>, and can also be adapted for use with other widely used platforms such as OpenFOAM via custom interface modifications. Illustrative examples for <em>a posteriori</em> validations demonstrate that DFODE-kit models achieve excellent predictive accuracy. Furthermore, in isolated chemistry evaluations, the DNN models attain up to <em>O</em>(10<sup>2</sup>) acceleration compared with CVODE, while end-to-end CFD runs typically see multi-fold speed-ups. The package, dataset, and example scripts are released to support reproducible benchmarking and community adoption. <strong>PROGRAM SUMMARY</strong><em>Program Title:</em> DFODE-kit <em>CPC Library link to program files:</em> (to be added by Technical Editor) <em>Developer’s repository link:</em> <span><span>https://github.com/deepflame-ai/DFODE-kit</span><svg><path></path></svg></span> <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> Python <em>Nature of problem:</em>In combustion systems, chemical source terms are governed by stiff ODEs, where stiffness arises from the inherent multiscale nature of chemical kinetics. Specifically, the vastly disparate timescales between fast and slow reactions, combined with strong nonlinear coupling among species, give rise to numerically stiff systems that require extremely small time steps for stable and accurate integration. As a result, ODE integration often dominates the computational cost of high-fidelity reacting flow simulations, limiting their scalability and physical resolution. <em>Solution method:</em> To address the computational challenges posed by stiff chemical ODE integration, deep learning provides a promising alternative, owing to its powerful nonlinear regression capabilities. When trained on high-fidelity thermochemical datasets, deep learning models can accurately approximate the complex relationships between thermoch
{"title":"DFODE-Kit: Deep learning package for solving flame chemical kinetics with high-dimensional stiff ordinary differential equations","authors":"Han Li , Ke Xiao , Yangchen Xu , Haoze Zhang , Zhenyi Chen , Runze Mao , Zhi X. Chen","doi":"10.1016/j.cpc.2025.110013","DOIUrl":"10.1016/j.cpc.2025.110013","url":null,"abstract":"<div><div>Recent advances in deep learning for solving flame chemical kinetics offer promising solutions to the long-standing trade-off between accuracy and computational efficiency in combustion simulations. This work introduces DFODE-kit, an open-source Python package designed to replace the conventional, computationally intensive integration of chemical source terms governed by high-dimensional, stiff ordinary differential equations (ODEs), thereby substantially accelerating chemistry evaluation in combustion simulations. The package provides: i) an efficient sampling module that extracts high-quality thermochemical states from low-dimensional manifolds in canonical flames; ii) an effective data augmentation module that enriches the dataset to approximate the high-dimensional composition space encountered in turbulent flames; and (iii) an optimized neural network training module with multiscale preprocessing and physics-informed constraints to enhance model fidelity and stability. The trained models are seamlessly integrated into our previously released CFD solver DeepFlame<span><span><sup>1</sup></span></span>, and can also be adapted for use with other widely used platforms such as OpenFOAM via custom interface modifications. Illustrative examples for <em>a posteriori</em> validations demonstrate that DFODE-kit models achieve excellent predictive accuracy. Furthermore, in isolated chemistry evaluations, the DNN models attain up to <em>O</em>(10<sup>2</sup>) acceleration compared with CVODE, while end-to-end CFD runs typically see multi-fold speed-ups. The package, dataset, and example scripts are released to support reproducible benchmarking and community adoption. <strong>PROGRAM SUMMARY</strong><em>Program Title:</em> DFODE-kit <em>CPC Library link to program files:</em> (to be added by Technical Editor) <em>Developer’s repository link:</em> <span><span>https://github.com/deepflame-ai/DFODE-kit</span><svg><path></path></svg></span> <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> Python <em>Nature of problem:</em>In combustion systems, chemical source terms are governed by stiff ODEs, where stiffness arises from the inherent multiscale nature of chemical kinetics. Specifically, the vastly disparate timescales between fast and slow reactions, combined with strong nonlinear coupling among species, give rise to numerically stiff systems that require extremely small time steps for stable and accurate integration. As a result, ODE integration often dominates the computational cost of high-fidelity reacting flow simulations, limiting their scalability and physical resolution. <em>Solution method:</em> To address the computational challenges posed by stiff chemical ODE integration, deep learning provides a promising alternative, owing to its powerful nonlinear regression capabilities. When trained on high-fidelity thermochemical datasets, deep learning models can accurately approximate the complex relationships between thermoch","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"321 ","pages":"Article 110013"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145923460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-11DOI: 10.1016/j.cpc.2026.110025
Yongsen He , Rui Wang , Yanming Wang , Siyu Liu
Numerical modelling of solution-based laser-induced synthesis (LIS) remains challenging due to the presence of strongly coupled multiphysics processes operating across broad temporal and spatial scales, from ultrafast nanosecond laser heating to macroscopic material deposition on the order of seconds. To address the computational cost of single-scale simulations, this study proposes a multiscale Lattice Boltzmann Method (LBM) framework based on implementation of information transfer via moment-space projection, which reconstructs distribution functions using a pseudo-inverse operator to effectively filter high-order non-equilibrium noise at grid interfaces. Furthermore, a subcycling strategy is employed to enforce consistent dimensionless transport parameters across subdomains, eliminating the need for explicit temporal interpolation. Validation through Von Neumann analysis and lid-driven cavity benchmarks confirms the method’s unconditional linear stability and second-order spatial accuracy. When applied to the LIS process, the framework successfully couples thermal, hydrodynamic, and chemical fields, achieving a 91.9% reduction in lattice count and an 88.9% reduction in CPU time compared to uniform single-scale LBM, without compromising physical fidelity. This work provides a scalable and efficient approach for simulating additive manufacturing processes characterized by inherent spatiotemporal disparities spanning multiple orders of magnitude.
{"title":"A multiscale lattice Boltzmann framework based on moment-space information transfer: Application to laser-induced synthesis","authors":"Yongsen He , Rui Wang , Yanming Wang , Siyu Liu","doi":"10.1016/j.cpc.2026.110025","DOIUrl":"10.1016/j.cpc.2026.110025","url":null,"abstract":"<div><div>Numerical modelling of solution-based laser-induced synthesis (LIS) remains challenging due to the presence of strongly coupled multiphysics processes operating across broad temporal and spatial scales, from ultrafast nanosecond laser heating to macroscopic material deposition on the order of seconds. To address the computational cost of single-scale simulations, this study proposes a multiscale Lattice Boltzmann Method (LBM) framework based on implementation of information transfer via moment-space projection, which reconstructs distribution functions using a pseudo-inverse operator to effectively filter high-order non-equilibrium noise at grid interfaces. Furthermore, a <span><math><mrow><mn>1</mn><mo>:</mo><msup><mrow><mi>r</mi></mrow><mn>2</mn></msup></mrow></math></span> subcycling strategy is employed to enforce consistent dimensionless transport parameters across subdomains, eliminating the need for explicit temporal interpolation. Validation through Von Neumann analysis and lid-driven cavity benchmarks confirms the method’s unconditional linear stability and second-order spatial accuracy. When applied to the LIS process, the framework successfully couples thermal, hydrodynamic, and chemical fields, achieving a 91.9% reduction in lattice count and an 88.9% reduction in CPU time compared to uniform single-scale LBM, without compromising physical fidelity. This work provides a scalable and efficient approach for simulating additive manufacturing processes characterized by inherent spatiotemporal disparities spanning multiple orders of magnitude.</div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"321 ","pages":"Article 110025"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974175","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-11DOI: 10.1016/j.cpc.2026.110029
Fei Fei , Donghui Liu , Lefeng Xie , Zhiyuan Ren , Yuan Hu
Based on the Ellipsoidal–Statistical BGK (ESBGK) model developed by Hild and Pfeiffer [J. Comput. Phys. 514, 113226 (2024)], the unified stochastic particle (USP) method is extended to the simulation of polyatomic gas mixtures. By decomposing the collision term into macroscopic and microscopic components and solving the macroscopic part in conjunction with the particle motion, the USP method achieves the asymptotic-preserving property for the Navier-Stokes equations and second-order accuracy in the fluid limit. The proposed scheme is verified through some 1D and 2D benchmark cases, including Couette flow, Poiseuille flow, Shock wave, and hypersonic flow past a cylinder. The USP method results are in good agreement with the Direct Simulation Monte Carlo (DSMC) data across a wide range of Knudsen numbers. Additionally, the proposed USP method demonstrates superior accuracy and efficiency compared to the traditional stochastic particle (SP) method, making it a more suitable choice for complex multi-scale gas dynamics problems.
{"title":"A unified stochastic particle method for polyatomic gas mixtures","authors":"Fei Fei , Donghui Liu , Lefeng Xie , Zhiyuan Ren , Yuan Hu","doi":"10.1016/j.cpc.2026.110029","DOIUrl":"10.1016/j.cpc.2026.110029","url":null,"abstract":"<div><div>Based on the Ellipsoidal–Statistical BGK (ESBGK) model developed by Hild and Pfeiffer [J. Comput. Phys. 514, 113226 (2024)], the unified stochastic particle (USP) method is extended to the simulation of polyatomic gas mixtures. By decomposing the collision term into macroscopic and microscopic components and solving the macroscopic part in conjunction with the particle motion, the USP method achieves the asymptotic-preserving property for the Navier-Stokes equations and second-order accuracy in the fluid limit. The proposed scheme is verified through some 1D and 2D benchmark cases, including Couette flow, Poiseuille flow, Shock wave, and hypersonic flow past a cylinder. The USP method results are in good agreement with the Direct Simulation Monte Carlo (DSMC) data across a wide range of Knudsen numbers. Additionally, the proposed USP method demonstrates superior accuracy and efficiency compared to the traditional stochastic particle (SP) method, making it a more suitable choice for complex multi-scale gas dynamics problems.</div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"321 ","pages":"Article 110029"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974176","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2025-12-18DOI: 10.1016/j.cpc.2025.110008
Julian Soltau , Arne Walter , Frank Duschek , Thomas Dekorsy
We present a new simulation framework for the detection of aerosol fluorescence with integration spheres. Utilizing a Monte Carlo based ray-tracing approach, aerosol fluorescence within integrating sphere setups is simulated from photon generation through laser excitation over interactions with the setup components to losses and finally detection. Through modular design, the position and number of openings, sensors, etc. can be freely configured. Therefore, potential experimental setups can be evaluated with regard to overall performance, bottlenecks can be identified and the impact of different component parameters determined.
PROGRAM SUMMARY
Program Title: AFIS - Aerosol Fluorescence in Integrating Spheres
CPC Library link to program files:https://doi.org/10.17632/nj9dg3tr6d.1
Licensing provisions: BSD 3-clause
Programming language: Python
Nature of problem: Measuring (bio-)aerosol fluorescence is a complex task, especially for thin aerosols. In order to evaluate new experimental setups utilizing an integrating sphere, simulation data is essential to asses which system configurations yield promising results. Therefore, a simulation environment capable of calculating the different interactions within the setup is necessary, ideally providing a high level of customizability for the simulated setups.
Solution method: The AFIS simulation framework utilizes a ray-tracing approach based on a classical Monte Carlo description of the involved processes. Through batch-wise processing and penalization the computational efficiency is increased.
{"title":"AFIS - A simulation framework for detection of aerosol fluorescence with integrating spheres","authors":"Julian Soltau , Arne Walter , Frank Duschek , Thomas Dekorsy","doi":"10.1016/j.cpc.2025.110008","DOIUrl":"10.1016/j.cpc.2025.110008","url":null,"abstract":"<div><div>We present a new simulation framework for the detection of aerosol fluorescence with integration spheres. Utilizing a Monte Carlo based ray-tracing approach, aerosol fluorescence within integrating sphere setups is simulated from photon generation through laser excitation over interactions with the setup components to losses and finally detection. Through modular design, the position and number of openings, sensors, etc. can be freely configured. Therefore, potential experimental setups can be evaluated with regard to overall performance, bottlenecks can be identified and the impact of different component parameters determined.</div><div><strong>PROGRAM SUMMARY</strong></div><div><em>Program Title:</em> AFIS - <strong>A</strong>erosol <strong>F</strong>luorescence in <strong>I</strong>ntegrating <strong>S</strong>pheres</div><div><em>CPC Library link to program files:</em> <span><span>https://doi.org/10.17632/nj9dg3tr6d.1</span><svg><path></path></svg></span></div><div><em>Licensing provisions:</em> BSD 3-clause</div><div><em>Programming language:</em> Python</div><div><em>Nature of problem:</em> Measuring (bio-)aerosol fluorescence is a complex task, especially for thin aerosols. In order to evaluate new experimental setups utilizing an integrating sphere, simulation data is essential to asses which system configurations yield promising results. Therefore, a simulation environment capable of calculating the different interactions within the setup is necessary, ideally providing a high level of customizability for the simulated setups.</div><div><em>Solution method:</em> The AFIS simulation framework utilizes a ray-tracing approach based on a classical Monte Carlo description of the involved processes. Through batch-wise processing and penalization the computational efficiency is increased.</div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"321 ","pages":"Article 110008"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145974179","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-16DOI: 10.1016/j.cpc.2026.110038
Li Huang
<div><div>Analytic continuation is an essential step in quantum Monte Carlo calculations. We present version 2.0 of the ACFlow package, a full-fledged open source toolkit for analytic continuation of quantum Monte Carlo simulation data. The new version adds support for three recently developed analytic continuation methods, namely the barycentric rational function approximation method, the stochastic pole expansion method, and the Nevanlinna analytical continuation method. The well-established maximum entropy method is also enhanced with the Bayesian reconstruction entropy algorithm. Furthermore, a web-based graphical user interface and a testing toolkit for analytic continuation methods are introduced. In this paper, we at first summarize the basic principles of the newly implemented analytic continuation solvers, and the most important improvements of ACFlow 2.0. Then a representative example is provided to demonstrate the new usages and features.</div><div>PROGRAM SUMMARY <em>Program Title:</em> ACFlow <em>CPC Library link to program files:</em> <span><span>https://doi.org/10.17632/th6w74gwjm.2</span><svg><path></path></svg></span> <em>Developer’s repository link:</em> <span><span>https://github.com/huangli712/ACFlow</span><svg><path></path></svg></span> <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> Julia <em>Journal reference of previous version:</em> Computer Physics Communications 292, 108,863 (2023) <em>Does the new version supersede the previous version?:</em> Yes <em>Reasons for the new version:</em> Many features, including new analytic continuation solvers, a web-based graphical user interface, and a benchmark toolkit, are implemented. The user’s manual and internal tests are greatly enhanced as well. <em>Summary of revisions:</em> (1) The barycentric rational function approximation method is implemented, which is extremely fast and accurate. (2) The stochastic pole expansion method is implemented. It is a new variation of the stochastic analytic continuation method. (3) The Nevanlinna analytical continuation method is implemented. If the input Matsubara data is noise-free, this method can reach unprecedented accuracy. (4) The traditional maximum entropy method is enhanced by the Bayesian reconstruction entropy algorithm. Then it is extended to implement the positive-negative entropy formalism to support analytic continuation for off-diagonal Green’s function. (5) A web-based graphical user interface, namely ACGui, is developed. (6) A benchmark toolkit for testing various analytic continuation methods and codes, namely ACTest, is developed. (7) The documentation is polished. More examples and tests are included. <em>Nature of problem:</em> Most of the quantum Monte Carlo algorithms work on the imaginary axis. In order to extract physical observables and compare them with the experimental results, analytic continuation must be done in the post-processing stage to convert the quantum Monte Carlo simulated data from
{"title":"ACFlow 2.0 : An open source toolkit for analytic continuation of quantum Monte Carlo data","authors":"Li Huang","doi":"10.1016/j.cpc.2026.110038","DOIUrl":"10.1016/j.cpc.2026.110038","url":null,"abstract":"<div><div>Analytic continuation is an essential step in quantum Monte Carlo calculations. We present version 2.0 of the ACFlow package, a full-fledged open source toolkit for analytic continuation of quantum Monte Carlo simulation data. The new version adds support for three recently developed analytic continuation methods, namely the barycentric rational function approximation method, the stochastic pole expansion method, and the Nevanlinna analytical continuation method. The well-established maximum entropy method is also enhanced with the Bayesian reconstruction entropy algorithm. Furthermore, a web-based graphical user interface and a testing toolkit for analytic continuation methods are introduced. In this paper, we at first summarize the basic principles of the newly implemented analytic continuation solvers, and the most important improvements of ACFlow 2.0. Then a representative example is provided to demonstrate the new usages and features.</div><div>PROGRAM SUMMARY <em>Program Title:</em> ACFlow <em>CPC Library link to program files:</em> <span><span>https://doi.org/10.17632/th6w74gwjm.2</span><svg><path></path></svg></span> <em>Developer’s repository link:</em> <span><span>https://github.com/huangli712/ACFlow</span><svg><path></path></svg></span> <em>Licensing provisions:</em> GPLv3 <em>Programming language:</em> Julia <em>Journal reference of previous version:</em> Computer Physics Communications 292, 108,863 (2023) <em>Does the new version supersede the previous version?:</em> Yes <em>Reasons for the new version:</em> Many features, including new analytic continuation solvers, a web-based graphical user interface, and a benchmark toolkit, are implemented. The user’s manual and internal tests are greatly enhanced as well. <em>Summary of revisions:</em> (1) The barycentric rational function approximation method is implemented, which is extremely fast and accurate. (2) The stochastic pole expansion method is implemented. It is a new variation of the stochastic analytic continuation method. (3) The Nevanlinna analytical continuation method is implemented. If the input Matsubara data is noise-free, this method can reach unprecedented accuracy. (4) The traditional maximum entropy method is enhanced by the Bayesian reconstruction entropy algorithm. Then it is extended to implement the positive-negative entropy formalism to support analytic continuation for off-diagonal Green’s function. (5) A web-based graphical user interface, namely ACGui, is developed. (6) A benchmark toolkit for testing various analytic continuation methods and codes, namely ACTest, is developed. (7) The documentation is polished. More examples and tests are included. <em>Nature of problem:</em> Most of the quantum Monte Carlo algorithms work on the imaginary axis. In order to extract physical observables and compare them with the experimental results, analytic continuation must be done in the post-processing stage to convert the quantum Monte Carlo simulated data from ","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"321 ","pages":"Article 110038"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-18DOI: 10.1016/j.cpc.2026.110040
Aaron Nagel , Johannes Löwe
To simulate highly-resolved flow fields, we extend the Quantum Lattice Boltzmann Method (QLBM) to be able to simulate multiple time steps without state extraction or reinitialization. We adjust and extend given QLBM approaches from the literature to completely remove the need to measure or reinitialize the flow field in between the simulation time steps. Therefore, our algorithm does not require to sample the entire flow field at any time. We solve the linear advection-diffusion problem with periodic boundary conditions and derive all necessary equations and build the corresponding quantum circuit diagrams, including details on the QLBM blocks and explicitly drawing the circuit gates. We discuss the general decay of a QLBM step and how that effects our algorithm. The new algorithm is verified on 1D and 2D test cases using the shot method of IBMs Qiskit package. We show excellent agreement and convergence between our QLBM and the classical Lattice Boltzmann method. The conclusion section includes a discussion on the advantages of our algorithm as well as limitations and to what extent it is more efficient.
{"title":"Quantum lattice boltzmann method for multiple time steps without reinitialization for linear advection-Diffusion problems","authors":"Aaron Nagel , Johannes Löwe","doi":"10.1016/j.cpc.2026.110040","DOIUrl":"10.1016/j.cpc.2026.110040","url":null,"abstract":"<div><div>To simulate highly-resolved flow fields, we extend the Quantum Lattice Boltzmann Method (QLBM) to be able to simulate multiple time steps without state extraction or reinitialization. We adjust and extend given QLBM approaches from the literature to completely remove the need to measure or reinitialize the flow field in between the simulation time steps. Therefore, our algorithm does not require to sample the entire flow field at any time. We solve the linear advection-diffusion problem with periodic boundary conditions and derive all necessary equations and build the corresponding quantum circuit diagrams, including details on the QLBM blocks and explicitly drawing the circuit gates. We discuss the general decay of a QLBM step and how that effects our algorithm. The new algorithm is verified on 1D and 2D test cases using the <em>shot</em> method of IBMs <em>Qiskit</em> package. We show excellent agreement and convergence between our QLBM and the classical Lattice Boltzmann method. The conclusion section includes a discussion on the advantages of our algorithm as well as limitations and to what extent it is more efficient.</div></div>","PeriodicalId":285,"journal":{"name":"Computer Physics Communications","volume":"321 ","pages":"Article 110040"},"PeriodicalIF":3.4,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034921","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：