Abstract:We consider a setting in which it is desired to find an optimal complex vector $\mathbf{x}\in\mathbb{C}^N$ that satisfies $\mathcal{A}(\mathbf{x}) \approx \mathbf{b}$ in a least-squares sense, where $\mathbf{b} \in \mathbb{C}^M$ is a data vector (possibly noise-corrupted), and $\mathcal{A}(\cdot): \mathbb{C}^N \rightarrow \mathbb{C}^M$ is a measurement operator. If $\mathcal{A}(\cdot)$ were linear, this reduces to the classical linear least-squares problem, which has a well-known analytic solution as well as powerful iterative solution algorithms. However, instead of linear least-squares, this work considers the more complicated scenario where $\mathcal{A}(\cdot)$ is nonlinear, but can be represented as the summation and/or composition of some operators that are linear and some operators that are antilinear. Some common nonlinear operations that have this structure include complex conjugation or taking the real-part or imaginary-part of a complex vector. Previous literature has shown that this kind of mixed linear/antilinear least-squares problem can be mapped into a linear least-squares problem by considering $\mathbf{x}$ as a vector in $\mathbb{R}^{2N}$ instead of $\mathbb{C}^N$. While this approach is valid, the replacement of the original complex-valued optimization problem with a real-valued optimization problem can be complicated to implement, and can also be associated with increased computational complexity. In this work, we describe theory and computational methods that enable mixed linear/antilinear least-squares problems to be solved iteratively using standard linear least-squares tools, while retaining all of the complex-valued structure of the original inverse problem. An illustration is provided to demonstrate that this approach can simplify the implementation and reduce the computational complexity of iterative solution algorithms.

The differentiation of pseudo-inverses and nonlinear least squares problems whose variables separate

Least Squares Approximations in Linear Statistical Inverse Learning Problems

Proving Nonsingularity of Coefficient Matrix in Least Squares Normal Equation of Non-Linear Semiparametric Model

Least Squares Properties of Generalized Inverses

Convergence Analysis of a Variable Projection Method for Regularized Separable Nonlinear Inverse Problems

The Convergence of Least-Squares Progressive Iterative Approximation for Singular Least-Squares Fitting System.

A Levenberg-Marquardt Method for Nonsmooth Regularized Least Squares

Least squares formulation for ill-posed inverse problems and applications

A Riemannian under-determined BFGS method for least squares inverse eigenvalue problems

Optimal low-rank approximations of posteriors for linear Gaussian inverse problems on Hilbert spaces

A Note on Separable Nonlinear Least Squares Problem

Model Misfit Minimization

An Efficient Algorithm for Ill-Conditioned Separable Nonlinear Least Squares

Non-intrusive Least-Squares Functional A Posteriori Error Estimator: Linear and Nonlinear Problems with Plain Convergence

Variable projection methods for separable nonlinear inverse problems with general-form Tikhonov regularization

An Iterative Approach to a Constrained Least Squares Problem

Beyond Moore-Penrose Part I: Generalized Inverses that Minimize Matrix Norms

Stable gradient projection method for nonlinear conditionally well-posed inverse problems

Efficient Iterative Solutions to Complex-Valued Nonlinear Least-Squares Problems with Mixed Linear and Antilinear Operators

An Efficient Projection Method for Nonlinear Inverse Problems with Sparsity Constraints

Insights Into Algorithms for Separable Nonlinear Least Squares Problems