Abstract:Given samples of a real or complex-valued function on a set of distinct nodes, the traditional linear Chebyshev approximation is to compute the best minimax approximation on a prescribed linear functional space. Lawson's iteration is a classical and well-known method for that task. However, Lawson's iteration converges linearly and in many cases, the convergence is very slow. In this paper, by the duality theory of linear programming, we first provide an elementary and self-contained proof for the well-known Alternation Theorem in the real case. Also, relying upon the Lagrange duality, we further establish an $L_q$-weighted dual programming for the linear Chebyshev approximation. In this framework, we revisit the convergence of Lawson's iteration, and moreover, propose a Newton type iteration, the interior-point method, to solve the $L_2$-weighted dual programming. Numerical experiments are reported to demonstrate its fast convergence and its capability in finding the reference points that characterize the unique minimax approximation.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is the slow convergence rate of the Lawson iteration method in linear Chebyshev approximation. Specifically, for a given set of real - or complex - valued function samples on a set of discrete nodes, the traditional linear Chebyshev approximation aims to calculate the best minimax approximation in a predefined linear function space. The Lawson iteration method is a classical method for solving this problem, but it usually has a linear convergence rate and converges very slowly in many cases.
To solve this problem, the author improves the existing methods in the following ways:
1. **Duality theory**: Using the duality theory of linear programming, the author provides an elementary and self - contained proof of the Alternation Theorem in the real - valued case.
2. **Lq - weighted dual programming**: Based on Lagrange duality, an Lq - weighted dual programming framework for linear Chebyshev approximation is established.
3. **Newton - type iteration and interior - point method**: Under the L2 - weighted dual programming framework, the convergence of the Lawson iteration method is re - examined, and an interior - point method based on Newton iteration is proposed to solve the L2 - weighted dual programming. This method shows a faster convergence rate and can find the reference points characterizing the unique minimax approximation in many cases.
These improvements not only accelerate the convergence of weight updates but also improve the efficiency and accuracy of the algorithm, especially when the nodes are relatively dense or there are too many nodes close to some reference points.
### Formula summary
- Objective function of the Chebyshev approximation problem:
\[
p^*=\arg\min_{p\in P_n}\|f - p(x)\|_\infty
\]
where \(\|f - p(x)\|_\infty=\max_{1\leq j\leq m}|f_j - p(x_j)|\).
- Lq - weighted dual programming:
\[
\eta^*_\infty=\left(\max_{w\in S}d_q(w)\right)^{1/q}
\]
where \(d_q(w)=\min_{p\in P_n}\sum_{j = 1}^m w_j|f_j - p(x_j)|^q\), and \(S\) is the probability simplex \(S = \{w = [w_1,\dots,w_m]^T\in\mathbb{R}^m:w\geq0,w^T e = 1\}\).
- Objective function of the interior - point method:
\[
\min_{w,s}-d(w)-\mu\sum_{i = 1}^m\log s_i,\quad\text{s.t., }w^T e = 1,\;w - s = 0
\]
Through these improvements, the method proposed in the paper can solve the linear Chebyshev approximation problem more efficiently, especially when dealing with large - scale data.