Machine learning-based parameter optimization for M\"untz spectral methods

TL;DR

This paper introduces a machine learning framework using neural networks to optimize parameters in M"untz spectral methods, significantly enhancing accuracy for solving fractional PDEs with low-regularity solutions.

Contribution

It presents the first ML-based parameter optimization approach for M"untz spectral methods, demonstrating improved accuracy and generalization across dimensions for fractional PDEs.

Findings

ANN-based parameter prediction outperforms traditional methods

The trained ANN generalizes from 1D to 2D problems

The framework improves spectral method accuracy for low-regularity solutions

Abstract

Spectral methods employing non-standard polynomial bases, such as M\"untz polynomials, have proven effective for accurately solving problems with solutions exhibiting low regularity, notably including sub-diffusion equations. However, due to the absence of theoretical guidance, the key parameters controlling the exponents of M\"untz polynomials are usually determined empirically through extensive numerical experiments, leading to a time-consuming tuning process. To address this issue, we propose a novel machine learning-based optimization framework for the M\"untz spectral method. As an illustrative example, we optimize the parameter selection for solving time-fractional partial differential equations (PDEs). Specifically, an artificial neural network (ANN) is employed to predict optimal parameter values based solely on the time-fractional order as input. The ANN is trained by minimizing solution errors on a one-dimensional time-fractional convection-diffusion equation featuring manufactured exact solutions that manifest singularities of varying intensity, covering a comprehensive range of sampled fractional orders. Numerical results for time-fractional PDEs in both one and two dimensions demonstrate that the ANN-based parameter prediction significantly improves the accuracy of the M\"untz spectral method. Moreover, the trained ANN generalizes effectively from one-dimensional to two-dimensional cases, highlighting its robustness across spatial dimensions. Additionally, we verify that the ANN substantially outperforms traditional function approximators, such as spline interpolation, in both prediction accuracy and training efficiency. The proposed optimization framework can be extended beyond fractional PDEs, offering a versatile and powerful approach for spectral methods applied to various low-regularity problems.