Abstract:The affine matrix rank minimization (AMRM) problem is to find a matrix of minimum rank that satisfies a given linear system constraint. It has many applications in some important areas such as control, recommender systems, matrix completion and network localization. However, the problem (AMRM) is NP-hard in general due to the combinational nature of the matrix rank function. There are many alternative functions have been proposed to substitute the matrix rank function, which lead to many corresponding alternative minimization problems solved efficiently by some popular convex or nonconvex optimization algorithms. In this paper, we propose a new nonconvex function, namely, $TL_{\alpha}^{\epsilon}$ function (with $0\leq\alpha<1$ and $\epsilon>0$), to approximate the rank function, and translate the NP-hard problem (AMRM) into the $TL_{p}^{\epsilon}$ function affine matrix rank minimization (TLAMRM) problem. Firstly, we study the equivalence of problem (AMRM) and (TLAMRM), and proved that the uniqueness of global minimizer of the problem (TLAMRM) also solves the NP-hard problem (AMRM) if the linear map $\mathcal{A}$ satisfies a restricted isometry property (RIP). Secondly, an iterative thresholding algorithm is proposed to solve the regularization problem (RTLAMRM) for all $0\leq\alpha<1$ and $\epsilon>0$. At last, some numerical results on low-rank matrix completion problems illustrated that our algorithm is able to recover a low-rank matrix, and the extensive numerical on image inpainting problems shown that our algorithm performs the best in finding a low-rank image compared with some state-of-art methods.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the **Affine Matrix Rank Minimization (AMRM) problem**. Specifically, the goal of this problem is to find a matrix with the minimum rank under the given linear system constraints. This problem has applications in several important fields, such as control theory, recommendation systems, matrix completion, and network localization, etc. However, due to the combinatorial nature of the matrix rank function, the AMRM problem is NP - hard, which means that all known finite - time algorithms have at least double - exponential running time both theoretically and practically. To solve this difficult problem, researchers have proposed many surrogate functions to approximate the matrix rank function, thus deriving many surrogate minimization problems that can be effectively solved by popular convex or non - convex optimization algorithms. In this paper, the author proposes a new non - convex function, namely the TLǫα function (where 0 ≤ α < 1 and ǫ > 0), to approximate the rank function and transform the NP - hard AMRM problem into the TLǫα - function Affine Matrix Rank Minimization (TLAMRM) problem. The main contributions of the paper include: 1. **Equivalence study**: The equivalence between the AMRM problem and the TLAMRM problem is studied, and it is proved that if the linear mapping A satisfies a certain Restricted Isometry Property (RIP), then the global minimum solution of the TLAMRM problem is also a solution to the NP - hard AMRM problem. 2. **Iterative threshold algorithm**: An iterative threshold algorithm is proposed to solve the Regularized TLAMRM problem (RTLAMRM), which is applicable for all 0 ≤ α < 1 and ǫ > 0. 3. **Numerical experiments**: The effectiveness of the algorithm is demonstrated through the numerical results of the low - rank matrix completion problem, and extensive numerical experiments on the image inpainting problem show that the proposed algorithm is superior to some existing advanced methods in finding low - rank images. In conclusion, this paper aims to provide an effective solution to the NP - hard AMRM problem by introducing a new non - convex function and the corresponding algorithm.

A New Nonconvex Strategy to Affine Matrix Rank Minimization Problem

Promoting low rank non-convex fraction function penalty for affine matrix rank minimization

Exact Recovery Low-Rank Matrix Via Transformed Affine Matrix Rank Minimization

Large-Scale Affine Matrix Rank Minimization with a Novel Nonconvex Regularizer.

Adaptive iterative singular value thresholding algorithm to low-rank matrix recovery

Generalized Singular Value Thresholding Operator to Affine Matrix Rank Minimization Problem

DC algorithm for solving the transformed affine matrix rank minimization

A Nonconvex Method to Low-Rank Matrix Completion

Efficient Rank Minimization via Solving Non-convexPenalties by Iterative Shrinkage-Thresholding Algorithm

Affine Matrix Rank Minimization Problem Via $P$-Thresholding Function

Efficient Recovery of Low-Rank Matrix Via Double Nonconvex Nonsmooth Rank Minimization.

Exploring Algorithmic Limits of Matrix Rank Minimization under Affine Constraints

Equivalence between the affine matrix rank minimization problem and the unconstrained matrix rank minimization problem

Low-Rank Matrix Recovery Via Modified Schatten-$p$ Norm Minimization with Convergence Guarantees

Multitask Learning for Classification Problem Via New Tight Relaxation of Rank Minimization.

Learned Turbo-type Affine Rank Minimization

Guaranteed Minimum-Rank Solutions of Linear Matrix Equations via Nuclear Norm Minimization

A continuous-time neurodynamic approach in matrix form for rank minimization

Matrix Completion Via Minimizing an Approximate Rank

A Novel Method for a Class of Structured Low-Rank Minimizations with Equality Constraint.

Augmented Lagrangian Alternating Direction Method for Low-Rank Minimization Via Non-Convex Approximation.