Extraction of material properties through multi-fidelity deep learning from molecular dynamics simulation

Islam, Mahmudul; Thakur, Shajedul Hoque; Mojumder, Satyajit; Hasan, Mohammad Nasim

doi:10.1016/j.commatsci.2020.110187

Cited by 23 publications

(14 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in recent years, physics-informed neural networks have shown great potential in function approximation and have been able to develop relationships between viscosity and loading wt%. [372,373] Thus, developing a deep neural network or an analytical model that can anticipate the rheological properties of the ink from its constituents and other process parameters will accelerate the ink preparation and further extend the library of printable materials. • Numerical modeling and simulation of the printing process may provide great insights into the physics of printing.…”

Section: Summary and Future Prospectsmentioning

confidence: 99%

Direct Ink Writing: A 3D Printing Technology for Diverse Materials

et al. 2022

Self Cite

View full text Add to dashboard Cite

Additive manufacturing (AM) has gained significant attention due to its ability to drive technological development as a sustainable, flexible, and customizable manufacturing scheme. Among the various AM techniques, direct ink writing (DIW) has emerged as the most versatile 3D printing technique for the broadest range of materials. DIW allows printing of practically any material, as long as the precursor ink can be engineered to demonstrate appropriate rheological behavior. This technique acts as a unique pathway to introduce design freedom, multifunctionality, and stability simultaneously into its printed structures. Here, a comprehensive review of DIW of complex 3D structures from various materials, including polymers, ceramics, glass, cement, graphene, metals, and their combinations through multimaterial printing is presented. The review begins with an overview of the fundamentals of ink rheology, followed by an in‐depth discussion of the various methods to tailor the ink for DIW of different classes of materials. Then, the diverse applications of DIW ranging from electronics to food to biomedical industries are discussed. Finally, the current challenges and limitations of this technique are highlighted, followed by its prospects as a guideline toward possible futuristic innovations.

show abstract

Section: Summary and Future Prospectsmentioning

confidence: 99%

Direct Ink Writing: A 3D Printing Technology for Diverse Materials

et al. 2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…Symbolic and numerical methods like finite differentiation perform very badly when applied to complex functions; automatic differentiation (AD), on the other hand, overcomes numerous restrictions as floating-point precision errors, for numerical differentiation, or [41], Schiassi et al [153] 2-4 layers 32 neurons per layer He et al [60] 50 neurons per layer Tartakovsky et al [170] 5-8 layers 250 neurons per layer Zhu et al [199] 9+ layers Cheng and Zhang [31] Waheed et al [175] Sparse Ramabathiran and Ramachandran [148] multi FC-DNN Amini Niaki et al [6] Islam et al [69] CNN plain CNN Gao et al [48] Fang [44] AE CNN Zhu et al [200], Geneva and Zabaras [51] Wang et al [177] RNN RNN Viana et al [174] LSTM Zhang et al [197] Yucesan and Viana [194] Other BNN Yang et al [190] GAN Yang et al [189] We summarize Sect. 2 by showcasing some of the papers that represent each of the many Neural Network implementations of PINN.…”

Section: Injection Of Physical Lawsmentioning

confidence: 99%

“…Long-range molecular dynamics simulation is addressed with multi-fidelity PINN (MPINN) by estimating nanofluid viscosity over a wide range of sample space using a very small number of molecular dynamics simulations Islam et al [69]. The authors were able to estimate system energy per atom, system pressure, and diffusion coefficients, in particular with the viscosity of argon-copper nanofluid.…”

Section: Molecular Dynamics and Materials Related Applicationsmentioning

confidence: 99%

Scientific Machine Learning Through Physics–Informed Neural Networks: Where we are and What’s Next

et al. 2022

View full text Add to dashboard Cite

Physics-Informed Neural Networks (PINN) are neural networks (NNs) that encode model equations, like Partial Differential Equations (PDE), as a component of the neural network itself. PINNs are nowadays used to solve PDEs, fractional equations, integral-differential equations, and stochastic PDEs. This novel methodology has arisen as a multi-task learning framework in which a NN must fit observed data while reducing a PDE residual. This article provides a comprehensive review of the literature on PINNs: while the primary goal of the study was to characterize these networks and their related advantages and disadvantages. The review also attempts to incorporate publications on a broader range of collocation-based physics informed neural networks, which stars form the vanilla PINN, as well as many other variants, such as physics-constrained neural networks (PCNN), variational hp-VPINN, and conservative PINN (CPINN). The study indicates that most research has focused on customizing the PINN through different activation functions, gradient optimization techniques, neural network structures, and loss function structures. Despite the wide range of applications for which PINNs have been used, by demonstrating their ability to be more feasible in some contexts than classical numerical techniques like Finite Element Method (FEM), advancements are still possible, most notably theoretical issues that remain unresolved.

show abstract

“…Long-range molecular dynamics simulation is addressed with multi-fidelity PINN (MPINN) by estimating nanofluid viscosity over a wide range of sample space using a very small number of molecular dynamics simulations Islam et al (2021). The authors were able to estimate system energy per atom, system pressure, and diffusion coefficients, in particular with the viscosity of argoncopper nanofluid.…”

Section: Molecular Dynamics and Materials Related Applicationsmentioning

confidence: 99%