Abstract:This article study the average conditioning for a random underdetermined polynomial system. The expected value of the moments of the condition number are compared to the moments of the condition number of random matrices. An expression for these moments is given by studying the kernel finding problem for random matrices. Furthermore, the second moment of the Frobenius condition number is computed.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to study the average condition number of random under - determined polynomial systems. Specifically, the author compares the condition number moments of random under - determined polynomial systems with those of random matrices, and gives expressions for these moments by studying the kernel - finding problem of random matrices. In addition, the second - order moment of the Frobenius condition number is also calculated.
### Main Contributions
1. **Definition of Condition Number**:
- The condition number is used to measure the sensitivity of a computational problem to input changes. For a polynomial system \(h\), the Frobenius condition number at point \(x\) is defined as:
\[
\mu_r^F(h, x) := \|h\| \|Dh(x)^\dagger \Delta(d^{1/2}_i \|x\|^{d_i - 1})\|_F
\]
where \(\|\cdot\|_F\) is the Frobenius norm, and \(Dh(x)^\dagger\) is the Moore - Penrose pseudo - inverse of \(Dh(x)\).
2. **Main Results**:
- The main result of the paper is the expected value of the second - order moment of the Frobenius condition number. Specifically, for a random polynomial system \(h\) under the standard Gaussian distribution, the expected value of the second - order moment of its Frobenius condition number is:
\[
E_{h\in H_{r,n}^{(d)}}(\mu_{r,2}^F,\text{Av}(h))=\frac{(N - 1)r}{n - r+ 1}
\]
where \(N\) is the complex dimension of \(H_{r,n}^{(d)}\).
3. **Corollaries**:
- When \(r = n\), the expected value of the second - order moment of the Frobenius condition number in the deterministic case can be recovered:
\[
E_{g\in H_{r,r}^{(d)}}(\mu_2^F,\text{Av}(g))=(\tilde{N}-1)r
\]
where \(\tilde{N}\) is the complex dimension of \(H_{r,r}^{(d)}\).
### Methods and Techniques
1. **Geometric Framework**:
- The solution variety \(V\) and the projection mapping on it, as well as the critical point set \(\Sigma'\) and the discriminant variety \(\Sigma\) are defined.
2. **Double - Fiber Technique**:
- The double - fiber technique is used to handle integration problems, especially to convert the integral from \(H_{r,n}^{(d)}\) to \(P(C^{n + 1})\) through the smoothed co - area formula.
3. **Linear Solution Variety**:
- The linear solution variety \(V_{\text{lin}}\) and the projection mapping on it, as well as the properties of the critical point set are studied.
4. **Solution Variety of Kernel - Finding Problem**:
- The solution variety \(V_{\text{ker}}\) of the kernel - finding problem and the projection mapping on it, as well as the properties of the critical point set are studied.
### Conclusions
Through the above methods and techniques, the paper successfully calculates the expected value of the second - order moment of the Frobenius condition number of random under - determined polynomial systems and gives its specific expression. This result not only extends the existing theory but also provides a new perspective for understanding the stability of random polynomial systems.