Abstract:Direct solution of simultaneous linear equations is regarded to be slow for large systems of equations and requires special treatment to avoid numerical instability. A new method is proposed that addresses the numerical instability without any special treatments. The method uses the orthogonal matrices to achieve numerical stability.
What problem does this paper attempt to address?
### What problems does this paper attempt to solve?
This paper aims to solve two main problems encountered in the direct solution of large - scale linear equation systems:
1. **Low computational efficiency**: For large - scale linear equation systems, direct solution methods (such as the Gaussian elimination method) are generally considered to be slow in terms of computational speed.
2. **Numerical instability**: Numerical instability is likely to occur during the solution process, resulting in a decrease in the accuracy of the solution or the inability to obtain the correct solution.
To solve these problems, the paper proposes a new direct solution method. By using orthogonal matrices (especially Householder matrices), it improves numerical stability and does not require special processing steps. This method can not only effectively reduce the amount of computation but also ensure the stability of numerical calculations.
### Overview of the specific method
The core idea of the paper is to use the Householder matrix \( H \) to transform the coefficient matrix \( A \), so that the system is gradually simplified into smaller - scale sub - problems, and finally solved by the back - substitution method. The specific steps are as follows:
1. **Introducing the Householder matrix**:
- Select a Householder matrix \( H \) such that the first row of \( C = AH \) is proportional to the unit vector.
- Through the formula \( c_1 = a_1H = |a_1|e_1 \), where \( a_1 \) is the first row of matrix \( A \), \( |a_1| \) is its 2 - norm, and \( e_1 \) is the unit vector.
2. **Gradually simplifying the system**:
- By gradually applying the Householder matrix, the original \( n\times n \) system is gradually simplified into an \((n - 1)\times(n - 1) \) system until finally a \( 2\times2 \) system remains.
- At each step, update the right - hand - side vector \( b \), and record the intermediate results \( y \) and the Householder vector \( V \).
3. **Solving by back - substitution**:
- After solving the final \( 2\times2 \) system, gradually solve the solution \( x \) of the entire system by the back - substitution method.
### Numerical stability and computational complexity
- **Numerical stability**: Since the Householder matrix is an orthogonal matrix, its inverse is equal to its transpose, which avoids complex matrix inversion operations and thus improves numerical stability.
- **Computational complexity**: The computation at each step mainly involves matrix - vector multiplication, and the total computational complexity is \( O(n^2) \), which significantly reduces the amount of computation compared with traditional direct solution methods.
### Summary
This paper proposes a direct solution method based on the Householder matrix, which effectively solves the computational efficiency and numerical stability problems in the solution of large - scale linear equation systems, and provides an efficient and stable solution.