R. de la Madrid
Abstract: The rigged Hilbert space of the algebra of the one-dimensional rectangular barrier potential is constructed. The one-dimensional rectangular potential provides another opportunity to show that the rigged Hilbert space fully accounts for Dirac's bra-ket formalism. The analogy between Dirac's formalism and Fourier methods is pointed out.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is **how to overcome the numerical instability of the divided differences algorithm**. Specifically, the author focuses on the numerical instability problem in the divided differences algorithm, which is widely used in polynomial interpolation and approximation problems. The divided differences algorithm is used to calculate the coefficients of the Newton interpolation polynomial, which is the first step in the Björck and Pereyra algorithm for solving the Vandermonde system of equations.
### Main problem description
1. **Numerical instability of the divided differences algorithm**:
- The divided differences algorithm may produce large numerical errors in some cases, especially when the interpolation points are unevenly distributed or the data is complex.
- This instability will affect the accuracy of the Newton interpolation coefficients and, in turn, the precision of the entire interpolation polynomial.
2. **Limitations of the Aitken algorithm**:
- The Aitken algorithm is another commonly used method for calculating interpolation polynomials, but its numerical stability also has problems, especially when dealing with complex data.
### Goals of the paper
The main goal of the paper is to propose a new algorithm to ensure higher numerical stability and accuracy when calculating the Newton coefficients of the interpolation polynomial. Specifically:
- **Propose a new algorithm**: The author proposes a new algorithm that is always numerically stable in floating - point operations and is more accurate than the existing divided differences algorithm and Aitken algorithm, even for complex data.
- **Analyze existing algorithms**: Through detailed error analysis of the Aitken algorithm and the divided differences algorithm, their numerical stability performance under different conditions is revealed.
- **Verify the effectiveness of the new algorithm**: Through numerical experiments in MATLAB, the superior performance of the new algorithm in practical applications is verified.
### Key formulas
The key formulas involved in the paper include:
- **Lagrange interpolation polynomial**:
\[
w_N(z)=\sum_{j = 0}^{N}f_j\prod_{\substack{i = 0\\i\neq j}}^{N}\frac{z - z_i}{z_j - z_i}
\]
- **Newton interpolation polynomial**:
\[
w_N(z)=\sum_{j = 0}^{N}c_j\prod_{i = 0}^{j - 1}(z - z_i)
\]
- **Definition of divided differences**:
\[
f(z_j,z_{j + 1},\dots,z_{j + n})=\frac{f(z_j,z_{j + 1},\dots,z_{j + n- 1})-f(z_{j + 1},z_{j + 2},\dots,z_{j + n})}{z_j - z_{j + n}}
\]
### Conclusion
The paper shows through theoretical analysis and numerical experiments that the newly proposed algorithm is superior to the traditional divided differences algorithm and Aitken algorithm in terms of numerical stability, and especially shows better precision when dealing with complex data. This provides an effective solution to the numerical instability of the divided differences algorithm.