Deep learning and inverse discovery of polymer self-consistent field theory inspired by physics-informed neural networks
Authors:
- 25 July 2022
- journal article
- research article
- Published by American Physical Society (APS) in Physical Review E
- Vol. 106 (1) , 014503
Abstract
We devise a deep learning solver inspired by physics-informed neural networks (PINNs) to tackle the polymer self-consistent field theory (SCFT) equations for one-dimensional AB-diblock copolymers. The PINNs framework comprises two parallel feedforward neural networks that separately represent the segmental partition functions and self-consistent chemical potential fields. The two networks are coupled through a loss function incorporating the governing equation, initial and boundary conditions, and the incompressibility constraint. To avoid the metastable homogeneous solution, the network parameters are initialized based on known self-consistent fields obtained from the numerical pseudospectral method. For copolymers of length at a given volume fraction of A block () and the reduced Flory-Huggins interaction parameter (), the minimization of the loss function leads to the converged network parameters that successfully capture the stable lamellar phase. The periodicity of the lamellar structure is correctly reproduced for the explored sets of , irrespective of the presumed computational domain size for initialization. Moreover, the proposed PINNs are applicable to the inverse discovery of the interaction parameter and the embedded chemical potential fields for an observed structure. This capability of solving the inverse SCFT problem demonstrates the potential of using PINNs to accelerate the exploration of new polymeric materials.
Keywords
Funding Information
- Ministry of Science and Technology, Taiwan (MOST 105-2218-E-002-036-MY3, MOST 108-2628-E-002-009-MY3)
- National Taiwan University (NTU 108L891205, NTU-109L891105)
- Ulusal Yüksek Başarımlı Hesaplama Merkezi, Istanbul Teknik Üniversitesi
This publication has 42 references indexed in Scilit:
- Responsive Behaviors of Diblock Polyampholyte Brushes within Self-Consistent Field TheoryThe Journal of Physical Chemistry B, 2012
- Solving differential equations with unsupervised neural networksChemical Engineering and Processing - Process Intensification, 2003
- Improved numerical algorithm for exploring block copolymer mesophasesJournal of Polymer Science Part B: Polymer Physics, 2002
- Efficient computation of the structural phase behavior of block copolymersPhysical Review E, 2002
- The standard Gaussian model for block copolymer meltsJournal of Physics: Condensed Matter, 2001
- Autophobic dewetting of homopolymer on a brush and entropic attraction between opposing brushes in a homopolymer matrixThe Journal of Chemical Physics, 2001
- Solving-Body Problems with Neural NetworksPhysical Review Letters, 2001
- Neural Network Differential Equation and Plasma Equilibrium SolverPhysical Review Letters, 1995
- Stable and unstable phases of a diblock copolymer meltPhysical Review Letters, 1994
- Neural‐network‐based approximations for solving partial differential equationsCommunications in Numerical Methods in Engineering, 1994