Abstract:Fourier neural operators (FNOs) can learn highly nonlinear mappings between function spaces, and have recently become a popular tool for learning responses of complex physical systems. However, to achieve good accuracy and efficiency, FNOs rely on the Fast Fourier transform (FFT), which is restricted to modeling problems on rectangular domains. To lift such a restriction and permit FFT on irregular geometries as well as topology changes, we introduce domain agnostic Fourier neural operator (DAFNO), a novel neural operator architecture for learning surrogates with irregular geometries and evolving domains. The key idea is to incorporate a smoothed characteristic function in the integral layer architecture of FNOs, and leverage FFT to achieve rapid computations, in such a way that the geometric information is explicitly encoded in the architecture. In our empirical evaluation, DAFNO has achieved state-of-the-art accuracy as compared to baseline neural operator models on two benchmark datasets of material modeling and airfoil simulation. To further demonstrate the capability and generalizability of DAFNO in handling complex domains with topology changes, we consider a brittle material fracture evolution problem. With only one training crack simulation sample, DAFNO has achieved generalizability to unseen loading scenarios and substantially different crack patterns from the trained scenario. Our code and data accompanying this paper are available at <a class="link-external link-https" href="https://github.com/ningliu-iga/DAFNO" rel="external noopener nofollow">this https URL</a>.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the limitations of existing Fourier neural operators (FNOs) when dealing with non - rectangular domains and complex topological changes. Specifically, traditional FNOs rely on the fast Fourier transform (FFT), which requires that the computational domain must be rectangular and the grid be evenly distributed. This limitation makes it difficult for FNOs to be applied to problems with irregular geometries or topological changes, such as material contact, crack propagation, etc. To this end, the authors propose the Domain Agnostic Fourier Neural Operator (DAFNO), a new neural operator architecture that can explicitly encode the boundary information of irregular domains, thereby expanding the application range of FNOs and enabling them to maintain high efficiency and accuracy when dealing with complex geometries and topological changes.
### Main Contributions
1. **Proposing DAFNO**: A new type of Fourier neural operator architecture that can explicitly encode the boundary information of irregular domains, enabling the learned operator to recognize domain boundaries and generalize to domains with different complex geometries and topological structures.
2. **Improved Integral Layer**: By introducing a smooth domain feature function in the integral layer, the layer update of DAFNO is similar to that of non - local models, which can collect the interactions between material points within the domain while cutting off non - physical influences outside the domain. In this way, the model retains both the fidelity of the domain boundary and the form of the convolution kernel, thereby retaining the computational efficiency of the FFT.
3. **Extensive Experimental Verification**: Demonstrates the expressive and generalization abilities of DAFNO in a variety of scientific problems, including hyperelastic material constitutive modeling, airfoil design, and crack propagation in brittle materials. The experimental results show that DAFNO can not only handle irregular domains but also cope with topological changes during the evolution process.
### Background and Related Work
- **Neural Operator Learning**: A neural operator is a deep - learning method for learning infinite - dimensional mappings from input functions to output functions. Compared with traditional neural networks, neural operators are insensitive to resolution and can be generalized to different input instances.
- **Fourier Neural Operator**: FNO learns mappings in the spectral domain through convolution operations and uses FFT to improve computational efficiency. However, FNO is limited to rectangular domains and uniform grids.
- **Limitations of Existing Solutions**: Existing solutions such as zero - padding or extrapolation techniques may introduce numerical instabilities and errors, while methods based on continuous mappings cannot handle topological changes.
### DAFNO Architecture
- **Domain Feature Function**: Defines a domain feature function \(\chi(x)\) for encoding domain information of different geometries.
- **Integral Layer Modification**: Introduce \(\chi(x)\) and \(\chi(y)\) in the integral layer to make the integral term disappear between points inside and outside the domain, thereby eliminating unnecessary interactions.
- **Smoothing Technique**: To overcome the Gibbs phenomenon, a smooth domain feature function \(\tilde{\chi}(x)\) is introduced, which is implemented through the hyperbolic tangent function.
### Experimental Results
- **Hyperelastic Material Modeling**: DAFNO performs well under different dataset scales, especially on large datasets, where its performance is significantly better than other baseline models.
- **Airfoil Design**: When learning transonic flow fields, DAFNO has the lowest test error, which is 24.9% lower than the best non - DAFNO baseline model and 63.9% lower than Geo - FNO.
- **Crack Propagation**: Using only one training sample, DAFNO can generalize to unseen loading scenarios and different crack patterns.
In conclusion, by explicitly encoding domain boundary information, DAFNO successfully solves the limitations of traditional FNOs in dealing with complex geometries and topological changes, demonstrating its superior performance in a variety of scientific problems.