Inspired by biology and implemented using nanotechnology, the self-assembly of patchy particles has emerged as a pivotal mechanism for constructing complex structures that mimic natural systems with diverse functionalities. Here, we explore the dissipative self-assembly of patchy particles under nonequilibrium conditions, with the aim of overcoming the constraints imposed by equilibrium assembly. Utilizing extensive Monte Carlo (MC) and Molecular Dynamics (MD) simulations, we provide insight into the effects of external forces that mirror natural and chemical processes on the assembly rates and the stability of the resulting assemblies comprising $8$, $10$, and $13$ patchy particles. Implemented by a favorable bond-promoting drive in MC or a pulsed square wave potential in MD, our simulations reveal the role these external drives play in accelerating assembly kinetics and enhancing structural stability, evidenced by a decrease in the time to first assembly and an increase in the duration the system remains in an assembled state. Through the analysis of an order parameter, entropy production, bond dynamics, and interparticle forces, we unravel the underlying mechanisms driving these advancements. We also validated our key findings by simulating a larger system of $100$ patchy particles. Our comprehensive results not only shed light on the impact of external stimuli on self-assembly processes but also open a promising pathway for expanding the application by leveraging patchy particles for novel nanostructures.
{"title":"Dissipative self-assembly of patchy particles under nonequilibrium drive: a computational study","authors":"Shubhadeep Nag, Gili Bisker","doi":"arxiv-2409.04748","DOIUrl":"https://doi.org/arxiv-2409.04748","url":null,"abstract":"Inspired by biology and implemented using nanotechnology, the self-assembly\u0000of patchy particles has emerged as a pivotal mechanism for constructing complex\u0000structures that mimic natural systems with diverse functionalities. Here, we\u0000explore the dissipative self-assembly of patchy particles under nonequilibrium\u0000conditions, with the aim of overcoming the constraints imposed by equilibrium\u0000assembly. Utilizing extensive Monte Carlo (MC) and Molecular Dynamics (MD)\u0000simulations, we provide insight into the effects of external forces that mirror\u0000natural and chemical processes on the assembly rates and the stability of the\u0000resulting assemblies comprising $8$, $10$, and $13$ patchy particles.\u0000Implemented by a favorable bond-promoting drive in MC or a pulsed square wave\u0000potential in MD, our simulations reveal the role these external drives play in\u0000accelerating assembly kinetics and enhancing structural stability, evidenced by\u0000a decrease in the time to first assembly and an increase in the duration the\u0000system remains in an assembled state. Through the analysis of an order\u0000parameter, entropy production, bond dynamics, and interparticle forces, we\u0000unravel the underlying mechanisms driving these advancements. We also validated\u0000our key findings by simulating a larger system of $100$ patchy particles. Our\u0000comprehensive results not only shed light on the impact of external stimuli on\u0000self-assembly processes but also open a promising pathway for expanding the\u0000application by leveraging patchy particles for novel nanostructures.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204060","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Joanna Piper Morgan, Ilham Variansyah, Braxton Cuneo, Todd S. Palmer, Kyle E. Niemeyer
Finding a software engineering approach that allows for portability, rapid development, open collaboration, and performance for high performance computing on GPUs and CPUs is a challenge. We implement a portability scheme using the Numba compiler for Python in Monte Carlo / Dynamic Code (MC/DC), a new neutron transport application for rapid Monte Carlo methods development. Using this scheme, we have built MC/DC as a single source, single language, single compiler application that can run as a pure Python, compiled CPU, or compiled GPU solver. In GPU mode, we use Numba paired with an asynchronous GPU scheduler called Harmonize to increase GPU performance. We present performance results for a time-dependent problem on both the CPU and GPU and compare them to a production code.
为 GPU 和 CPU 上的高性能计算寻找一种可移植性、快速开发、开放协作和性能的软件工程方法是一项挑战。我们在蒙特卡洛/动态代码（MC/DC）中使用Python的Numba编译器实现了一种可移植性方案，MC/DC是一种用于快速蒙特卡洛方法开发的新型中子传输应用程序。利用这一方案，我们将 MC/DC 构建成了一个单源代码、单语言、单编译器的应用程序，可以作为纯 Python、编译 CPU 或编译 GPU 的求解器运行。在 GPU 模式下，我们使用 Numba 搭配称为 Harmonize 的异步 GPU 调度来提高 GPU 性能。我们展示了 CPU 和 GPU 上一个随时间变化的问题的性能结果，并将其与生产代码进行了比较。
{"title":"Performance Portable Monte Carlo Neutron Transport in MCDC via Numba","authors":"Joanna Piper Morgan, Ilham Variansyah, Braxton Cuneo, Todd S. Palmer, Kyle E. Niemeyer","doi":"arxiv-2409.04668","DOIUrl":"https://doi.org/arxiv-2409.04668","url":null,"abstract":"Finding a software engineering approach that allows for portability, rapid\u0000development, open collaboration, and performance for high performance computing\u0000on GPUs and CPUs is a challenge. We implement a portability scheme using the\u0000Numba compiler for Python in Monte Carlo / Dynamic Code (MC/DC), a new neutron\u0000transport application for rapid Monte Carlo methods development. Using this\u0000scheme, we have built MC/DC as a single source, single language, single\u0000compiler application that can run as a pure Python, compiled CPU, or compiled\u0000GPU solver. In GPU mode, we use Numba paired with an asynchronous GPU scheduler\u0000called Harmonize to increase GPU performance. We present performance results\u0000for a time-dependent problem on both the CPU and GPU and compare them to a\u0000production code.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We propose a computation of curvature of arbitrary two-dimensional surfaces of three-dimensional objects, which is a contribution to discrete gravity with potential applications in network geometry. We begin by linking each point of the surface in question to its four closest neighbors, forming quads. We then focus on the simplices of $d=2$, or triangles embedded in these quads, which make up a pure simplicial complex with $d=2$. This allows us to numerically compute the local metric along with zweibeins, which subsequently leads to a derivation of discrete curvature defined at every triangle or face. We provide an efficient algorithm with $mathcal{O}(N log{N})$ complexity that first orients two-dimensional surfaces, solves the nonlinear system of equations of the spin-connections resulting from the torsion condition, and returns the value of curvature at each face.
{"title":"Curvature of an Arbitrary Surface for Discrete Gravity and for $d=2$ Pure Simplicial Complexes","authors":"Ali H. Chamseddine, Ola Malaeb, Sara Najem","doi":"arxiv-2409.04375","DOIUrl":"https://doi.org/arxiv-2409.04375","url":null,"abstract":"We propose a computation of curvature of arbitrary two-dimensional surfaces\u0000of three-dimensional objects, which is a contribution to discrete gravity with\u0000potential applications in network geometry. We begin by linking each point of\u0000the surface in question to its four closest neighbors, forming quads. We then\u0000focus on the simplices of $d=2$, or triangles embedded in these quads, which\u0000make up a pure simplicial complex with $d=2$. This allows us to numerically\u0000compute the local metric along with zweibeins, which subsequently leads to a\u0000derivation of discrete curvature defined at every triangle or face. We provide\u0000an efficient algorithm with $mathcal{O}(N log{N})$ complexity that first\u0000orients two-dimensional surfaces, solves the nonlinear system of equations of\u0000the spin-connections resulting from the torsion condition, and returns the\u0000value of curvature at each face.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204104","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Understanding the combustion chemistry of acetaldehyde is crucial to developing robust and accurate combustion chemistry models for practical fuels, especially for biofuels. This study aims to reevaluate the important rate and thermodynamic parameters for acetaldehyde combustion chemistry. The rate parameters of 79 key reactions are reevaluated using more than 100,000 direct experiments and quantum chemistry computations from >900 studies, and the thermochemistry ({Delta}hf(298K), s0(298K) and cp) of 24 key species are reevaluated based on the ATCT database, the NIST Chemistry WebBook, the TMTD database, and 35 published chemistry models. The updated parameters are incorporated into a recent acetaldehyde chemistry model, which is further assessed against available fundamental experiments (123 ignition delay times and 385 species concentrations) and existing chemistry models, with clearly better performance obtained in the high-temperature regime. Sensitivity and flux analyses further highlight the insufficiencies of previous models in representing the key pathways, particularly the branching ratios of acetaldehyde- and formaldehyde-consuming pathways. Temperature-dependent and temperature-independent uncertainties are statistically evaluated for kinetic and thermochemical parameters, respectively, where the large differences between the updated and the original model parameters reveal the necessity of reassessment of kinetic and thermochemical parameters completely based on direct experiments and theoretical calculations for rate and thermodynamic parameters.
{"title":"Comprehensive reevaluation of acetaldehyde chemistry and the underlying uncertainties","authors":"Xinrui Ren, Hongqing Wu, Ruoyue Tang, Yanqing Cui, Mingrui Wang, Song Cheng","doi":"arxiv-2409.04015","DOIUrl":"https://doi.org/arxiv-2409.04015","url":null,"abstract":"Understanding the combustion chemistry of acetaldehyde is crucial to\u0000developing robust and accurate combustion chemistry models for practical fuels,\u0000especially for biofuels. This study aims to reevaluate the important rate and\u0000thermodynamic parameters for acetaldehyde combustion chemistry. The rate\u0000parameters of 79 key reactions are reevaluated using more than 100,000 direct\u0000experiments and quantum chemistry computations from >900 studies, and the\u0000thermochemistry ({Delta}hf(298K), s0(298K) and cp) of 24 key species are\u0000reevaluated based on the ATCT database, the NIST Chemistry WebBook, the TMTD\u0000database, and 35 published chemistry models. The updated parameters are\u0000incorporated into a recent acetaldehyde chemistry model, which is further\u0000assessed against available fundamental experiments (123 ignition delay times\u0000and 385 species concentrations) and existing chemistry models, with clearly\u0000better performance obtained in the high-temperature regime. Sensitivity and\u0000flux analyses further highlight the insufficiencies of previous models in\u0000representing the key pathways, particularly the branching ratios of\u0000acetaldehyde- and formaldehyde-consuming pathways. Temperature-dependent and\u0000temperature-independent uncertainties are statistically evaluated for kinetic\u0000and thermochemical parameters, respectively, where the large differences\u0000between the updated and the original model parameters reveal the necessity of\u0000reassessment of kinetic and thermochemical parameters completely based on\u0000direct experiments and theoretical calculations for rate and thermodynamic\u0000parameters.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204107","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Kanka Ghosh, Oguz Umut Salman, Sylvain Queyreau, Lev Truskinovsky
We use molecular dynamics to show that plastic slip is a crucial component of the transformation mechanism of a square-to-triangular structural transition. The latter is a stylized analog of many other reconstructive phase transitions. To justify our conclusions we use a novel atomistically-informed mesoscopic representation of the field of lattice distortions in molecular dynamics simulations. Our approach reveals a hidden alternating slip distribution behind the seemingly homogeneous product phase which points to the fact that lattice invariant shears play a central role in this class of phase transformations. While the underlying pattern of anti-parallel displacements may also be interpreted as microscopic shuffling, its precise crystallographic nature strongly suggests the plasticity-centered interpretation.
{"title":"Slip-dominated structural transitions","authors":"Kanka Ghosh, Oguz Umut Salman, Sylvain Queyreau, Lev Truskinovsky","doi":"arxiv-2409.04066","DOIUrl":"https://doi.org/arxiv-2409.04066","url":null,"abstract":"We use molecular dynamics to show that plastic slip is a crucial component of\u0000the transformation mechanism of a square-to-triangular structural transition.\u0000The latter is a stylized analog of many other reconstructive phase transitions.\u0000To justify our conclusions we use a novel atomistically-informed mesoscopic\u0000representation of the field of lattice distortions in molecular dynamics\u0000simulations. Our approach reveals a hidden alternating slip distribution behind\u0000the seemingly homogeneous product phase which points to the fact that lattice\u0000invariant shears play a central role in this class of phase transformations.\u0000While the underlying pattern of anti-parallel displacements may also be\u0000interpreted as microscopic shuffling, its precise crystallographic nature\u0000strongly suggests the plasticity-centered interpretation.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Lattice thermal conductivity (kL) is a crucial physical property of crystals with applications in thermal management, such as heat dissipation, insulation, and thermoelectric energy conversion. However, accurately and rapidly determining kL poses a considerable challenge. In this study, we introduce an formula that achieves high precision (mean relative error=8.97%) and provides fast predictions, taking less than one minute, for kL across a wide range of inorganic binary and ternary materials. Our interpretable, dimensionally aligned and physical grounded formula forecasts kL values for 4,601 binary and 6,995 ternary materials in the Materials Project database. Notably, we predict undiscovered high kL values for AlBN2 (kL=101 W/ m/ K) and the undetectedlow kL Cs2Se (kL=0.98 W/ m/ K) at room temperature. This method for determining kL streamlines the traditionally time-consuming process associated with complex phonon physics. It provides insights into microscopic heat transport and facilitates the design and screening of materials with targeted and extreme kL values through the application of phonon engineering. Our findings offer opportunities for controlling and optimizing macroscopic transport properties of materials by engineering their bulk modulus, shear modulus, and Gruneisen parameter.
{"title":"An interpretable formula for lattice thermal conductivity of crystals","authors":"Xiaoying Wang, Guoyu Shu, Guimei Zhu, Jiansheng Wang, Jun Sun, Xiangdong Ding, Baowen Li, Zhibin Gao","doi":"arxiv-2409.04489","DOIUrl":"https://doi.org/arxiv-2409.04489","url":null,"abstract":"Lattice thermal conductivity (kL) is a crucial physical property of crystals\u0000with applications in thermal management, such as heat dissipation, insulation,\u0000and thermoelectric energy conversion. However, accurately and rapidly\u0000determining kL poses a considerable challenge. In this study, we introduce an\u0000formula that achieves high precision (mean relative error=8.97%) and provides\u0000fast predictions, taking less than one minute, for kL across a wide range of\u0000inorganic binary and ternary materials. Our interpretable, dimensionally\u0000aligned and physical grounded formula forecasts kL values for 4,601 binary and\u00006,995 ternary materials in the Materials Project database. Notably, we predict\u0000undiscovered high kL values for AlBN2 (kL=101 W/ m/ K) and the undetectedlow kL\u0000Cs2Se (kL=0.98 W/ m/ K) at room temperature. This method for determining kL\u0000streamlines the traditionally time-consuming process associated with complex\u0000phonon physics. It provides insights into microscopic heat transport and\u0000facilitates the design and screening of materials with targeted and extreme kL\u0000values through the application of phonon engineering. Our findings offer\u0000opportunities for controlling and optimizing macroscopic transport properties\u0000of materials by engineering their bulk modulus, shear modulus, and Gruneisen\u0000parameter.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142269567","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The application of machine learning methodologies for predicting properties within materials science has garnered significant attention. Among recent advancements, Kolmogorov-Arnold Networks (KANs) have emerged as a promising alternative to traditional Multi-Layer Perceptrons (MLPs). This study evaluates the impact of substituting MLPs with KANs within three established machine learning frameworks: Allegro, Neural Equivariant Interatomic Potentials (NequIP), and the Edge-Based Tensor Prediction Graph Neural Network (ETGNN). Our results demonstrate that the integration of KANs generally yields enhanced prediction accuracies. Specifically, replacing MLPs with KANs in the output blocks leads to notable improvements in accuracy and, in certain scenarios, also results in reduced training times. Furthermore, employing KANs exclusively in the output block facilitates faster inference and improved computational efficiency relative to utilizing KANs throughout the entire model. The selection of an optimal basis function for KANs is found to be contingent upon the particular problem at hand. Our results demonstrate the strong potential of KANs in enhancing machine learning potentials and material property predictions.
机器学习方法在材料科学特性预测中的应用已引起广泛关注。在最近取得的进展中，Kolmogorov-Arnold 网络（KANs）已成为替代传统多层感知器（MLPs）的一种有前途的方法。本研究评估了在三个成熟的机器学习框架内用 KAN 替代 MLP 的影响：我们的研究结果表明，KANs 的集成通常能提高预测精度。具体来说，在输出块中用 KAN 替代 MLP，可以显著提高准确率，在某些情况下还能缩短训练时间。此外，与在整个模型中使用 KAN 相比，仅在输出块中使用 KAN 可以加快推理速度并提高计算效率。我们发现，为 KANs 选择最佳基函数取决于手头的特定问题。我们的研究结果证明了 KANs 在增强机器学习潜力和材料特性预测方面的强大潜力。
{"title":"Efficient prediction of potential energy surface and physical properties with Kolmogorov-Arnold Networks","authors":"Rui Wang, Hongyu Yu, Yang Zhong, Hongjun Xiang","doi":"arxiv-2409.03430","DOIUrl":"https://doi.org/arxiv-2409.03430","url":null,"abstract":"The application of machine learning methodologies for predicting properties\u0000within materials science has garnered significant attention. Among recent\u0000advancements, Kolmogorov-Arnold Networks (KANs) have emerged as a promising\u0000alternative to traditional Multi-Layer Perceptrons (MLPs). This study evaluates\u0000the impact of substituting MLPs with KANs within three established machine\u0000learning frameworks: Allegro, Neural Equivariant Interatomic Potentials\u0000(NequIP), and the Edge-Based Tensor Prediction Graph Neural Network (ETGNN).\u0000Our results demonstrate that the integration of KANs generally yields enhanced\u0000prediction accuracies. Specifically, replacing MLPs with KANs in the output\u0000blocks leads to notable improvements in accuracy and, in certain scenarios,\u0000also results in reduced training times. Furthermore, employing KANs exclusively\u0000in the output block facilitates faster inference and improved computational\u0000efficiency relative to utilizing KANs throughout the entire model. The\u0000selection of an optimal basis function for KANs is found to be contingent upon\u0000the particular problem at hand. Our results demonstrate the strong potential of\u0000KANs in enhancing machine learning potentials and material property\u0000predictions.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204108","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The Quantum Natural Gradient (QNG) method enhances optimization in variational quantum algorithms (VQAs) by incorporating geometric insights from the quantum state space through the Fubini-Study metric. In this work, we extend QNG by introducing higher-order integrators and geodesic corrections using the Riemannian Euler update rule and geodesic equations, deriving an updated rule for the Quantum Natural Gradient with Geodesic Correction (QNGGC). QNGGC is specifically designed for small, shallow quantum circuits. We also develop an efficient method for computing the Christoffel symbols necessary for these corrections, leveraging the parameter-shift rule to enable direct measurement from quantum circuits. Through theoretical analysis and practical examples, we demonstrate that QNGGC significantly improves convergence rates over standard QNG, highlighting the benefits of integrating geodesic corrections into quantum optimization processes. Our approach paves the way for more efficient quantum algorithms, leveraging the advantages of geometric methods.
{"title":"Quantum Natural Gradient with Geodesic Corrections for Small Shallow Quantum Circuits","authors":"Mourad Halla","doi":"arxiv-2409.03638","DOIUrl":"https://doi.org/arxiv-2409.03638","url":null,"abstract":"The Quantum Natural Gradient (QNG) method enhances optimization in\u0000variational quantum algorithms (VQAs) by incorporating geometric insights from\u0000the quantum state space through the Fubini-Study metric. In this work, we\u0000extend QNG by introducing higher-order integrators and geodesic corrections\u0000using the Riemannian Euler update rule and geodesic equations, deriving an\u0000updated rule for the Quantum Natural Gradient with Geodesic Correction (QNGGC).\u0000QNGGC is specifically designed for small, shallow quantum circuits. We also\u0000develop an efficient method for computing the Christoffel symbols necessary for\u0000these corrections, leveraging the parameter-shift rule to enable direct\u0000measurement from quantum circuits. Through theoretical analysis and practical\u0000examples, we demonstrate that QNGGC significantly improves convergence rates\u0000over standard QNG, highlighting the benefits of integrating geodesic\u0000corrections into quantum optimization processes. Our approach paves the way for\u0000more efficient quantum algorithms, leveraging the advantages of geometric\u0000methods.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204172","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Physics-Informed Neural Networks (PINNs) are regarded as state-of-the-art tools for addressing highly nonlinear problems based on partial differential equations. Despite their broad range of applications, PINNs encounter several performance challenges, including issues related to efficiency, minimization of computational cost, and enhancement of accuracy. Burgers' equation, a fundamental equation in fluid dynamics that is extensively used in PINNs, provides flexible results with the Adam optimizer that does not account for past gradients. This paper introduces a novel strategy for solving Burgers' equation by incorporating DiffGrad with PINNs, a method that leverages the difference between current and immediately preceding gradients to enhance performance. A comprehensive computational analysis is conducted using optimizers such as Adam, Adamax, RMSprop, and DiffGrad to evaluate and compare their effectiveness. Our approach includes visualizing the solutions over space at various time intervals to demonstrate the accuracy of the network. The results show that DiffGrad not only improves the accuracy of the solution but also reduces training time compared to the other optimizers.
{"title":"DiffGrad for Physics-Informed Neural Networks","authors":"Jamshaid Ul Rahman, Nimra","doi":"arxiv-2409.03239","DOIUrl":"https://doi.org/arxiv-2409.03239","url":null,"abstract":"Physics-Informed Neural Networks (PINNs) are regarded as state-of-the-art\u0000tools for addressing highly nonlinear problems based on partial differential\u0000equations. Despite their broad range of applications, PINNs encounter several\u0000performance challenges, including issues related to efficiency, minimization of\u0000computational cost, and enhancement of accuracy. Burgers' equation, a\u0000fundamental equation in fluid dynamics that is extensively used in PINNs,\u0000provides flexible results with the Adam optimizer that does not account for\u0000past gradients. This paper introduces a novel strategy for solving Burgers'\u0000equation by incorporating DiffGrad with PINNs, a method that leverages the\u0000difference between current and immediately preceding gradients to enhance\u0000performance. A comprehensive computational analysis is conducted using\u0000optimizers such as Adam, Adamax, RMSprop, and DiffGrad to evaluate and compare\u0000their effectiveness. Our approach includes visualizing the solutions over space\u0000at various time intervals to demonstrate the accuracy of the network. The\u0000results show that DiffGrad not only improves the accuracy of the solution but\u0000also reduces training time compared to the other optimizers.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204110","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Rate theory and DFT calculations of hydrogen evolution reaction (HER) on MoS2 with Co, Ni and Pt impurities show the significance of dihydrogen (H2*) complex where both hydrogen atoms are interacting with the surface. Stabilization of such a complex affects the competing Volmer-Heyrovsky (direct H2 release) and Volmer-Tafel (H2* intermediate) pathways. The resulting evolution proceeds with a very small overpotential for all dopants ($eta$ = 0.1 to 0.2 V) at 25% edge substitution, significantly reduced from the already low $eta$ = 0.27 V for the undoped edge. At full edge substitution, Co-MoS2 remains highly active ($eta$ = 0.18 V) while Ni- and Pt-MoS2 are deactivated ($eta$ = 0.4 to 0.5 V) due to unfavorable interaction with H2*. Instead of the single S-vacancy, the site of intrinsic activity in the basal plane was found to be the undercoordinated central Mo-atom in threefold S-vacancy configurations, enabling hydrogen evolution with $eta$ = 0.52 V via a H2* intermediate. The impurity atoms interact favorably with the intrinsic sulfur vacancies on the basal plane, stabilizing but simultaneously deactivating the triple vacancy configuration. The calculated shifts in overpotential are consistent with reported measurements, and the dependence on doping level may explain variations in experimental observations.
{"title":"Doping-Induced Enhancement of Hydrogen Evolution at MoS2 Electrodes","authors":"Sander Ø. Hanslin, Hannes Jónsson, Jaakko Akola","doi":"arxiv-2409.02749","DOIUrl":"https://doi.org/arxiv-2409.02749","url":null,"abstract":"Rate theory and DFT calculations of hydrogen evolution reaction (HER) on MoS2\u0000with Co, Ni and Pt impurities show the significance of dihydrogen (H2*) complex\u0000where both hydrogen atoms are interacting with the surface. Stabilization of\u0000such a complex affects the competing Volmer-Heyrovsky (direct H2 release) and\u0000Volmer-Tafel (H2* intermediate) pathways. The resulting evolution proceeds with\u0000a very small overpotential for all dopants ($eta$ = 0.1 to 0.2 V) at 25% edge\u0000substitution, significantly reduced from the already low $eta$ = 0.27 V for\u0000the undoped edge. At full edge substitution, Co-MoS2 remains highly active\u0000($eta$ = 0.18 V) while Ni- and Pt-MoS2 are deactivated ($eta$ = 0.4 to 0.5 V)\u0000due to unfavorable interaction with H2*. Instead of the single S-vacancy, the\u0000site of intrinsic activity in the basal plane was found to be the\u0000undercoordinated central Mo-atom in threefold S-vacancy configurations,\u0000enabling hydrogen evolution with $eta$ = 0.52 V via a H2* intermediate. The\u0000impurity atoms interact favorably with the intrinsic sulfur vacancies on the\u0000basal plane, stabilizing but simultaneously deactivating the triple vacancy\u0000configuration. The calculated shifts in overpotential are consistent with\u0000reported measurements, and the dependence on doping level may explain\u0000variations in experimental observations.","PeriodicalId":501369,"journal":{"name":"arXiv - PHYS - Computational Physics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142204114","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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：