Advances in Numerical Methods for Linear Inverse Problems and Matrix Approximations

The field of numerical methods for linear inverse problems and matrix approximations is witnessing significant developments, driven by the need for efficient and accurate solutions to complex problems. Researchers are exploring new approaches to improve the performance of existing methods, such as the quasi-minimal residual method, and developing innovative techniques like hierarchical Tucker low-rank matrices and adaptive randomized algorithms. These advancements have the potential to impact various applications, including image processing, information retrieval, and scientific simulations. Noteworthy papers in this area include: The paper On the Choice of Subspace for the Quasi-minimal Residual Method for Linear Inverse Problems, which analyzes the impact of subspace selection on solution quality and demonstrates the effectiveness of range restricted QMR. The paper Hierarchical Tucker Low-Rank Matrices: Construction and Matrix-Vector Multiplication, which proposes a new matrix format for approximating non-oscillatory kernel functions and achieves linear complexity. The paper Fast adaptive tubal rank-revealing algorithm for t-product based tensor approximation, which introduces an adaptive randomized algorithm for tubal rank revelation in data tensors and provides theoretical guarantees for its performance.

Sources

On the Choice of Subspace for the Quasi-minimal Residual Method for Linear Inverse Problems

Hierarchical Tucker Low-Rank Matrices: Construction and Matrix-Vector Multiplication

Does block size matter in randomized block Krylov low-rank approximation?

A Tight Lower Bound for the Approximation Guarantee of Higher-Order Singular Value Decomposition

Efficient adaptive randomized algorithms for fixed-threshold low-rank matrix approximation

A Note on Eigenvalues of Perturbed Hermitian Matrices

Fast adaptive tubal rank-revealing algorithm for t-product based tensor approximation

Output-Sparse Matrix Multiplication Using Compressed Sensing

Built with on top of