Publications

Stable recovery of regularized linear inverse problems

Tran T. A. Nghia, Huy N. Pham, and Nghia V. Vo

Inverse Problems 41(6), 065018 (2025)

Abstract

We study when a low-complexity signal (sparse, group-sparse, or with sparse gradient) can be recovered stably from noisy linear measurements by convex regularization: the regularized solutions stay within a distance of the true signal proportional to the noise level. We give a characterization of stable recovery in finite dimensions that uses nonsmooth second-order information of the regularizer, and we show that for piecewise linear-quadratic regularizers stable recovery is equivalent to unique recovery. The theory yields new sufficient conditions for analysis group sparsity, including group Lasso and isotropic total variation, which we test numerically.

Cite

Tran T. A. Nghia, Huy N. Pham, and Nghia V. Vo. Stable recovery of regularized linear inverse problems. Inverse Problems 41(6), 065018, 2025. DOI: 10.1088/1361-6420/addffb.

@article{NghiaPhamVo2025Stable,
  title   = {Stable recovery of regularized linear inverse problems},
  author  = {Nghia, Tran T. A. and Pham, Huy N. and Vo, Nghia V.},
  journal = {Inverse Problems},
  volume  = {41},
  number  = {6},
  pages   = {065018},
  year    = {2025},
  doi     = {10.1088/1361-6420/addffb}
}