What problem does this paper attempt to address?

The problem that this paper attempts to solve is the challenges in low - rank tensor decomposition and approximation calculation. Specifically, the paper focuses on how to use generating polynomials to calculate the rank decomposition and low - rank approximation of tensors. The main problems can be summarized as follows: 1. **Low - rank tensor decomposition**: - How to effectively calculate the low - rank decomposition of a given tensor? - For asymmetric tensors, how to extend the generating polynomial method to calculate their rank decomposition? 2. **Low - rank tensor approximation**: - How to find a low - rank tensor close to a given tensor? - When the given tensor is close to a low - rank tensor, how to ensure that the calculated low - rank approximation is quasi - optimal? ### Specific problem description #### 1. Low - rank tensor decomposition The paper proposes a method based on generating polynomials to calculate the rank decomposition of tensors. The generating polynomials can be estimated by solving the linear least - squares problem, and then the approximate common zeros of these polynomials can be found by calculating the eigenvalue decomposition. Finally, the decomposition of the tensor is obtained by solving the linear least - squares problem through these common zeros. #### 2. Low - rank tensor approximation For the low - rank tensor approximation problem, the paper proposes an algorithm. First, a low - rank approximate tensor is obtained using the above - mentioned decomposition method, and then this approximation is further improved by a nonlinear optimization method. The paper proves that if the given tensor is close enough to a low - rank tensor, then the calculated low - rank approximation is quasi - optimal. ### Main contributions - **Extension of the generating polynomial method**: The paper extends the generating polynomial method to the rank decomposition and low - rank approximation calculation of asymmetric tensors. - **Quasi - optimal low - rank approximation**: The paper proves that when the given tensor is close enough to a low - rank tensor, the calculated low - rank approximation is quasi - optimal. - **Algorithm implementation**: The paper provides a series of algorithms, including the specific steps for calculating the tensor rank decomposition and low - rank approximation, and conducts numerical experiments for verification. ### Mathematical representation The mathematical formulas involved in the paper are as follows: - **Tensor rank decomposition**: \[ F=\sum_{s = 1}^r v_s^{(1)}\otimes v_s^{(2)}\otimes\cdots\otimes v_s^{(m)} \] where \(F\) is a \(n_1\times n_2\times\cdots\times n_m\) tensor, and \(v_s^{(i)}\in\mathbb{F}^{n_i}\). - **Low - rank tensor approximation**: \[ \min_{v_s^{(j)}\in\mathbb{F}^{n_j}}\|F - \sum_{s = 1}^r v_s^{(1)}\otimes v_s^{(2)}\otimes\cdots\otimes v_s^{(m)}\|^2 \] - **Generating polynomial**: \[ \langle pq,F\rangle = 0\quad\text{for all }q\in M_J^c \] where \(p\in M_J\) is the generating polynomial, and \(M_J\) and \(M_J^c\) represent certain specific polynomial sets respectively. ### Conclusion Through theoretical analysis and numerical experiments, the paper shows the effectiveness and accuracy of the proposed generating polynomial method in calculating low - rank tensor decomposition and approximation. These methods are applicable not only to symmetric tensors but also to asymmetric tensors, and have broad application prospects.

Low Rank Tensor Decompositions and Approximations

Low Rank Symmetric Tensor Approximations

Decomposing tensors via rank-one approximations

On the Global Convergence of the Alternating Least Squares Method for Rank-One Approximation to Generic Tensors

Decomposition of linear tensor transformations

Desingularization of bounded-rank tensor sets

Low-Rank Tensor Decomposition over Finite Fields

Low-Rank Tensor Completion Based on Log-Det Rank Approximation and Matrix Factorization

Low-rank optimization on Tucker tensor varieties

On The Tensor Svd And The Optimal Low Rank Orthogonal Approximation Of Tensors

The Computation of Low Multilinear Rank Approximations of Tensors Via Power Scheme and Random Projection

Robust Approximations of Low-Rank Minimization for Tensor Completion

On low-rank approximability of solutions to high-dimensional operator equations and eigenvalue problems

A low-rank approximation of tensors and the topological group structure of invertible matrices

Low-rank approximate inverse for preconditioning tensor-structured linear systems

Subrank and optimal reduction of scalar multiplications to generic tensors

A Non-Convex Tensor Rank Approximation for Tensor Completion

Optimal low rank tensor recovery

Uniqueness of Nonnegative Tensor Approximations

Randomized Algorithms for the Low Multilinear Rank Approximations of Tensors

Best subspace tensor approximations