Abstract:Lawson's iteration is a classical and effective method for solving the linear (polynomial) minimax approximation in the complex plane. Extension of Lawson's iteration for the rational minimax approximation with both computationally high efficiency and theoretical guarantee is challenging. A recent work [L.-H. Zhang, L. Yang, W. H. Yang and Y.-N. Zhang, A convex dual programming for the rational minimax approximation and Lawson's iteration, 2023,
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the convergence analysis of Lawson's iterative method in polynomial and rational minimax approximation. Specifically, the article mainly solves the following two problems:
1. **Convergence analysis of linear minimax approximation**:
- For the linear (polynomial) minimax approximation problem, the author proves that the nearly optimal Lawson index \(\beta\) in Lawson's iteration is 1. That is, for the linear minimax approximation problem, \(\beta = 1\) in Lawson's iteration is an approximately optimal choice.
2. **Convergence analysis of rational minimax approximation**:
- For the rational minimax approximation problem, the author proposes an improved Lawson's iterative method and proves that this method converges monotonically on the dual objective function \(d(\mathbf{w})\). Moreover, they also prove that when \(d(\mathbf{w}^{(k + 1)})=d(\mathbf{w}^{(k)})\), the convergent solution satisfies the complementary slackness condition: any node associated with a positive weight is either an interpolation point or has a constant error.
### Formula summary
- **Lawson's iterative update formula**:
\[
w_j^{(k + 1)}=\frac{w_j^{(k)}|f_j - p^{(k)}(x_j)|^\beta}{\sum_{i = 1}^m w_i^{(k)}|f_i - p^{(k)}(x_i)|^\beta}, \quad \forall j
\]
where \(\beta>0\) is the Lawson index.
- **Dual objective function**:
\[
d_2(\mathbf{w})=\min_{\mathbf{a}\in\mathbb{C}^{n_1 + 1}, \mathbf{b}\in\mathbb{C}^{n_2 + 1}}\left\|\sqrt{\mathbf{W}}[\Psi, F\Phi]\begin{bmatrix}\mathbf{a}\\\mathbf{b}\end{bmatrix}\right\|_2^2\quad\text{s.t.}\quad\left\|\sqrt{\mathbf{W}}\Phi\mathbf{b}\right\|_2 = 1
\]
- **Complementary slackness condition**:
\[
w_j^*(\|\mathbf{f}-\xi^*(\mathbf{x})\|_\infty - |f_j-\xi^*(x_j)|)=0, \quad \forall j = 1, 2,\ldots, m
\]
### Conclusion
Through the above analysis, the author provides theoretical support for the application of Lawson's iteration in polynomial and rational minimax approximation problems and proposes an improved algorithm to ensure its efficiency and reliability in practical applications. These theoretical results provide a solid foundation for understanding and improving Lawson's iteration.