Abstract:We propose a new stopping criterion for Krylov subspace iterative regularization of large-scale ill-posed inverse problems. Our stopping criterion accurately filters the data using a generalization of the Picard parameter that was originally introduced for direct regularization of small-scale problems. In the one dimension we filter the data in the discrete Fourier transform (DFT) basis using the Picard parameter, which separates noise-dominated Fourier coefficients from the signal-dominated ones. For two-dimensional problems we propose a novel vectorization scheme of the Fourier coefficients of the data based on the Kronecker product structure of the two-dimensional DFT matrix, which effectively reduces the problem to one dimension. At each iteration we compute the distance between the data reconstructed from the iterated solution and the filtered data, terminating the iterations once this distance begins to increase or to level off. The accuracy and robustness of the proposed method is demonstrated by several numerical examples and a MATLAB-based implementation is provided.

What problem does this paper attempt to address?

This paper attempts to solve the problem of how to determine appropriate stopping criteria when using the Krylov subspace iterative regularization method to solve large - scale ill - posed inverse problems. Specifically, the author proposes a new stopping criterion based on the Picard parameter to filter data more accurately and ensure that the iteration is terminated before the noise starts to dominate the solution. ### Main problems 1. **Solving large - scale ill - posed problems**: - The paper focuses on how to effectively solve large - scale linear ill - posed problems, which usually appear in fields such as image deblurring and machine learning. - Due to the ill - posed nature of the problem, direct solution is usually infeasible, so iterative methods are required for approximate solution. 2. **Influence of noise**: - In practical applications, the observed data usually contains noise, which will lead to a decline in the quality of the solution. Therefore, how to effectively handle noise during the iterative process is a key issue. 3. **Deficiencies of existing stopping criteria**: - Existing stopping criteria such as the L - curve method, generalized cross - validation (GCV) and the discrepancy principle perform poorly in some cases, especially when the noise level is unknown or varies greatly. - These methods may not be able to accurately determine the optimal number of iteration stops, resulting in unstable solution quality. ### Proposed new method The author proposes a stopping criterion based on the Picard parameter, which is implemented through the following steps: 1. **Data filtering**: - Use the discrete Fourier transform (DFT) basis instead of the singular value decomposition (SVD) basis to decompose the data into periodic and smooth components. - By detecting the Picard parameter \(k_0\), separate the noise - dominated frequency components from the signal - dominated frequency components. 2. **Iteration termination conditions**: - After each iteration, calculate the distance between the reconstructed data and the filtered data \(\hat{f}(k)=\|\hat{b}-A x^{(k)}\|_2\). - When the relative change \(\frac{\hat{f}(k)-\hat{f}(k + 1)}{\hat{f}(k)}\leq\delta\) persists for \(p\) iterations, terminate the iteration. 3. **Handling two - dimensional problems**: - For two - dimensional problems, a new vectorization scheme based on the Kronecker product structure is proposed to effectively simplify the problem into a one - dimensional problem. - This method can estimate the Picard parameter more accurately and improve the filtering effect. ### Conclusion Verified by numerical experiments, the stopping criterion based on the Picard parameter proposed by the author performs excellently in dealing with large - scale ill - posed inverse problems, can terminate the iteration more stably, and avoid the pollution of the solution by noise. In addition, this method is simple and easy to implement and is suitable for various application scenarios.

Stopping criterion for iterative regularization of large-scale ill-posed problems using the Picard parameter

A preconditioned Krylov subspace method for linear inverse problems with general-form Tikhonov regularization

A fast iterative regularization method for ill-posed problems

Scalable iterative data-adaptive RKHS regularization

Early Stopping for Ensemble Kalman-Bucy Inversion

A relaxed iterated Tikhonov regularization for linear ill-posed inverse problems

Hanke-Raus heuristic rule for iteratively regularized stochastic gradient descent

Early stopping for statistical inverse problems via truncated SVD estimation

Distributed Tikhonov regularization for ill-posed inverse problems from a Bayesian perspective

Stopping criteria for nonlinear variational problems: iterative approach

A Joint Bidiagonalization Based Algorithm for Large Scale Linear Discrete Ill-posed Problems in General-Form Regularization

Convergence analysis and parameter estimation for the iterated Arnoldi-Tikhonov method

Subspace projection regularization for large-scale Bayesian linear inverse problems

Iterative regularization for low complexity regularizers

Generalized hybrid iterative methods for large-scale Bayesian inverse problems

Projected Newton method for large-scale Bayesian linear inverse problems

Early stopping for conjugate gradients in statistical inverse problems

Uncertainty quantification for iterative algorithms in linear models with application to early stopping

A new stopping criterion for Krylov solvers applied in Interior Point Methods

Modified iterated Tikhonov methods for solving systems of nonlinear ill-posed equations

Parameter Selection by GCV and a $χ^2$ test within Iterative Methods for $\ell_1$-regularized Inverse Problems