Abstract:In this paper, we solve a classical counting problem for non-degenerate forms of symplectic and hermitian type defined on a vector space: given a subspace $\pi$, we find the number of non-singular subspaces that are trivially intersecting with $\pi$ and span a non-singular subspace with $\pi$. Lower bounds for the quantity of such pairs where $\pi$ is non-singular were first studied in ``Glasby, Niemeyer, Praeger (Finite Fields Appl., 2022)'', which was later improved in ``Glasby, Ihringer, Mattheus (Des. Codes Cryptogr., 2023)'' and generalised in ``Glasby, Niemeyer, Praeger (Linear Algebra Appl., 2022)''. In this paper, we derive explicit formulae, which allow us to give the exact proportion and improve the known lower bounds.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: Given a vector space $\mathbb{F}_q^n$ under a symplectic form or a Hermitian form, for a fixed subspace $\pi$, find the number of non - singular subspaces that have a trivial intersection with it (i.e., intersect only at the zero vector) and jointly generate a non - singular subspace. Specifically, the authors aim to determine the proportion of these specific pairs of subspaces and improve the known lower bounds.
### Main contributions of the paper
1. **Explicit formula**: The author derives a specific formula to calculate the exact number of non - singular subspaces that have a trivial intersection with a given subspace $\pi$ and jointly generate a non - singular subspace.
2. **Improved lower bound**: Through these formulas, the author is able to improve the lower bounds on the proportion of such pairs of subspaces given in previous studies.
3. **Anzahl theorem**: This problem is regarded as an Anzahl problem, similar to Segre's result, but for the cases of Hermitian and symplectic forms.
### Background and motivation
The proposal and discussion of this problem stem from the algorithmic requirements in computational group theory, especially algorithms for identifying classical groups. Specifically, understanding the proportion of these pairs of subspaces helps to optimize and improve the performance of these algorithms.
### Main results
- **Hermitian case**: In Chapter 3, the author presents the main results under the Hermitian form, including Theorem 3.7 and Theorem 3.9.
- **Symplectic case**: In Chapter 4, the author discusses similar results under the symplectic form, including Theorem 4.7.
### Mathematical expressions
The following are some key formulas in the paper:
- For the Hermitian form, given $0\leq i\leq\min\{j, n - j\}$ and $j\leq n - 1$, we have:
\[
\gamma_{i,j,n}=q^{2j(n - j)-\binom{j + 1}{2}}\phi^+_{1,i}(q)\sum_{m = 0}^{j - i}(- 1)^m(n - j)\phi^+_{i + 1,j - m}(q)\left[\begin{array}{c}j - i\\m\end{array}\right]^-_q q^{\binom{m}{2}-m(n - j - i)}
\]
- The expression for the proportion $\rho_{j,k,n}$ is:
\[
\rho_{j,k,n}=q^{jk-\binom{j + 1}{2}}\frac{\phi^-_{n - j - k + 1,n - k}(q)}{\phi^-_{n - j + 1,n}(q)}\sum_{m = 0}^{j}(-1)^m k\phi^+_{1,j - m}(q)\left[\begin{array}{c}j\\m\end{array}\right]^-_q q^{\binom{m}{2}-mk}
\]
These formulas not only provide the exact number but also improve the previous lower - bound estimates.