SPiFOL: A Spectral-based physics-informed finite operator learning for prediction of mechanical behavior of microstructures
Source abstract
Abstract
A novel physics-informed operator learning technique based on spectral methods is introduced to model the complex behavior of heterogeneous materials. The Lippmann–Schwinger operator in Fourier space is employed to construct physical constraints with minimal computational overhead, effectively eliminating the need for automatic differentiation. The introduced methodology accelerates the training process by enabling gradient construction on a fixed, finite discretization in Fourier space. Later, the spectral physics-informed finite operator learning (SPiFOL) framework is built based on this discretization and trained to map the arbitrary shape of microstructures to their mechanical responses (strain fields) without relying on labeled data. The training is done by minimizing equilibrium in Fourier space concerning the macroscopic loading condition, which also guarantees the periodicity. SPiFOL, as a physics-informed operator learning method, enables rapid predictions through forward inference after training. To ensure accuracy, we incorporate physical constraints and diversify the training data. However, performance may still degrade for out-of-distribution microstructures. SPiFOL is further enhanced by integrating a Fourier Neural Operator (FNO). Compared to the standard data-driven FNO, SPiFOL shows higher accuracy in predicting stress fields and provides nearly resolution-independent results. Additionally, its zero-shot super-resolution capabilities are explored in heterogeneous domains. Finally, SPiFOL is extended to handle 3D problems and further adapted to finite elasticity, demonstrating the robustness of the framework in handling nonlinear mechanical behavior. The framework shows great potential for efficient and scalable prediction of mechanical responses in complex material systems while also reducing the training time required for training physics-informed neural operators.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Model Reduction and Neural Networks · Non-Destructive Testing Techniques · Ultrasonics and Acoustic Wave Propagation
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (73)
Show source-provided references
1. Akiba, Takuya, Sano, Shotaro, Yanase, Toshihiko, Ohta, Takeru, Koyama, Masanori, 2019. Optuna: A Next-generation Hyperparameter Optimization Framework. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.10.1145/3292500.3330701 ↗
2. Asl · 2025
3. Bahmani · 2024
4. Bassey · 2021
5. Chapter 3 - A mathematical guide to operator learning
6. Bradbury · 2018
7. Calvello · 2024
8. Physics-informed neural networks for inverse problems in nano-optics and metamaterials10.1364/OE.384875 ↗
9. CAN-PINN: A fast physics-informed neural network based on coupled-automatic–numerical differentiation method10.1016/j.cma.2022.114909 ↗
10. Challenges in two-scale computational homogenization of mechanical metamaterials10.1002/pamm.202200139 ↗
11. FFT-based surrogate modeling of auxetic metamaterials with real-time prediction of effective elastic properties and swift inverse design10.1016/j.matdes.2024.113491 ↗
12. Finite strain FFT-based non-linear solvers made simple10.1016/j.cma.2016.12.032 ↗
13. Physics-guided, physics-informed, and physics-encoded neural networks and operators in scientific computing: Fluid and solid mechanics10.1115/1.4064449 ↗
14. Transfer learning enhanced physics informed neural network for phase-field modeling of fracture10.1016/j.tafmec.2019.102447 ↗
15. A physics-informed variational DeepONet for predicting crack path in quasi-brittle materials10.1016/j.cma.2022.114587 ↗
16. En-DeepONet: An enrichment approach for enhancing the expressivity of neural operators with applications to seismology10.1016/j.cma.2023.116681 ↗
17. Mixed formulation of physics-informed neural networks for thermo-mechanically coupled systems and heterogeneous domains10.1002/nme.7388 ↗
18. Geom-DeepONet: A point-cloud-based deep operator network for field predictions on 3D parameterized geometries10.1016/j.cma.2024.117130 ↗
19. Novel DeepONet architecture to predict stresses in elastoplastic structures with variable complex geometries and loads10.1016/j.cma.2023.116277 ↗
20. Neural tangent kernel: Convergence and generalization in neural networks
21. Jiang · 2025
22. Efficient fixed point and Newton–Krylov solvers for FFT-based homogenization of elasticity at large deformations10.1007/s00466-014-1071-8 ↗
23. Kapoor · 2022
24. Physics-informed machine learning10.1038/s42254-021-00314-5 ↗
25. Khorrami · 2024
26. Learning nonlinear operators in latent spaces for real-time predictions of complex dynamics in physical systems10.1038/s41467-024-49411-w ↗
27. Deep learning operator network for plastic deformation with variable loads and material properties10.1007/s00366-023-01822-x ↗
28. Kovachki · 2024
29. Neural operator: Learning maps between function spaces with applications to PDEs
30. Characterizing possible failure modes in physics-informed neural networks
31. An assessment of numerical techniques to find energy-minimizing microstructures associated with nonconvex potentials10.1002/nme.6280 ↗
32. Analysis of inhomogeneous materials at large strains using fast Fourier transforms
33. Phase-Field DeepONet: Physics-informed deep operator neural network for fast simulations of pattern formation governed by gradient flows of free-energy functionals10.1016/j.cma.2023.116299 ↗
34. Fourier neural operator with learned deformations for PDEs on general geometries
35. Li · 2020
36. Li · 2020
37. Li · 2023
38. Physics-informed neural operator for learning partial differential equations
39. Eighty years of the finite element method: Birth, evolution, and future10.1007/s11831-022-09740-9 ↗
40. Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators10.1038/s42256-021-00302-5 ↗
41. A comprehensive and fair comparison of two neural operators (with practical extensions) based on FAIR data10.1016/j.cma.2022.114778 ↗
42. Physics-informed neural networks with hard constraints for inverse design10.1137/21M1397908 ↗
43. An FFT-based crystal plasticity phase-field model for micromechanical fatigue cracking based on the stored energy density10.1016/j.ijfatigue.2023.107670 ↗
44. FFT based approaches in micromechanics: fundamentals, methods and applications10.1088/1361-651X/ac34e1 ↗
45. Learning the stress-strain fields in digital composites using Fourier neural operator
46. A numerical method for computing the overall response of nonlinear composites with complex microstructure10.1016/S0045-7825(97)00218-1 ↗
47. Solving the discretised neutron diffusion equations using neural networks10.1002/nme.7321 ↗
48. A review of the FE 2 method for composites
49. Physics-informed deep learning for computational elastodynamics without labeled data10.1061/(ASCE)EM.1943-7889.0001947 ↗
50. Revealing the predictive power of neural operators for strain evolution in digital composites10.1016/j.jmps.2023.105444 ↗
51. PhyCRNet: Physics-informed convolutional-recurrent network for solving spatiotemporal PDEs10.1016/j.cma.2021.114399 ↗
52. A finite operator learning technique for mapping the elastic properties of microstructures to their mechanical deformations10.1002/nme.7637 ↗
53. Rezaei · 2024
54. A mixed formulation for physics-informed neural networks as a potential solver for engineering problems in heterogeneous domains: Comparison with finite element method10.1016/j.cma.2022.115616 ↗
55. FFT-based computational micromechanics with Dirichlet boundary conditions on the rotated staggered grid
56. Applications of physics informed neural operators
57. A review of nonlinear FFT-based computational homogenization methods10.1007/s00707-021-02962-1 ↗
58. On polarization-based schemes for the FFT-based computational homogenization of inelastic materials10.1007/s00466-019-01694-3 ↗
59. Shen · 2011
60. Sinha · 2024
61. A survey of several finite difference methods for systems of nonlinear hyperbolic conservation laws10.1016/0021-9991(78)90023-2 ↗
62. A multiscale approach for modeling progressive damage of composite materials using fast Fourier transforms10.1016/j.cma.2013.10.017 ↗
63. Tran · 2021
64. Wang · 2024
65. Learning the solution operator of parametric partial differential equations with physics-informed DeepONets10.1126/sciadv.abi8605 ↗
66. Improved architectures and training algorithms for deep operator networks10.1007/s10915-022-01881-0 ↗
67. Modeling spatio-temporal dynamical systems with neural discrete learning and levels-of-experts10.1109/TKDE.2024.3363711 ↗
68. When and why PINNs fail to train: A neural tangent kernel perspective10.1016/j.jcp.2021.110768 ↗
69. Xiao · 2023
70. A finite element-based physics-informed operator learning framework for spatiotemporal partial differential equations on arbitrary domains
71. Learning deep Implicit Fourier Neural Operators (IFNOs) with applications to heterogeneous material modeling10.1016/j.cma.2022.115296 ↗
72. A physics-informed convolutional neural network for the simulation and prediction of two-phase Darcy flows in heterogeneous porous media10.1016/j.jcp.2023.111919 ↗
73. Physics-informed convolutional neural networks for temperature field prediction of heat source layout without labeled data10.1016/j.engappai.2022.105516 ↗