New constant dimension subspace codes from parallel linkage construction and multilevel construction (Q2130037)

Constant dimension subspace codes have efficient encoding and decoding algorithms and have important applications in network coding. Inspired by the results in [\textit{H. Chen} et al., IEEE Trans. Inf. Theory 66, No. 9, 5317--5321 (2020; Zbl 1448.94256); \textit{S. Liu} et al., IEEE Trans. Inf. Theory 66, No. 11, 6884--6897 (2020; Zbl 1453.94146)], The authors of this paper present an improved version of parallel construction and multilevel constructions, and are thus able to construct constant dimension codes with more codewords. More precisely, after giving some basic results on Ferrers diagram rank-metric codes, \(SC\)-representation, parallel linkage construction, the Delsarte theorem and so on, in section 3, the authors proved the following main results: Theorem 3.2. Let \(\mathcal{U} \subseteq \mathbb{F}_{q}^{k \times m}\) be SC-representation of constant dimension codes \(\mathcal{C}(\mathcal{U})\) with \(m \geq k,|\mathcal{U}|=N_{1}\), and \(d_{S}(\mathcal{C}(\mathcal{U}))=2 d\). Let \(\mathcal{P} \subseteq \mathbb{F}_{q}^{k \times n}\) be a rank-metric code with \(|\mathcal{P}|=N_{2}\) and \(d_{R}(\mathcal{P})=d\). Define a constant dimension code \[ \mathcal{C}_{1}=\{r s(U |P): U \in \mathcal{U}, P \in \mathcal{P}\}. \] Let \(\mathcal{V} \subseteq \mathbb{F}_{q}^{k \times n}\) be SC-representation of constant dimension codes \(\mathcal{C}(\mathcal{V})\) with \(n \geq k,|\mathcal{V}|=\) \(N_{3}\), and \(d_{S}(\mathcal{C}(\mathcal{V}))=2 d\). Let \(\mathcal{Q} \subseteq \mathbb{F}_{q}^{k \times m}\) be a rank-metric code with \(|\mathcal{Q}|=N_{4}\) and \(d_{R}(\mathcal{Q})=d\) such that the rank of each matrix in \(\mathcal{Q}\) is at most \(k-d\). Define a constant dimension code \[ \mathcal{C}_{2}=\{r s(Q |V): Q \in \mathcal{Q}, V \in \mathcal{V}\}. \] Consider the constant dimension code defined by \(\mathcal{C}=\mathcal{C}_{1} \cup \mathcal{C}_{2} \cup \mathcal{C}_{\mathcal{F}}\), where \(\mathcal{C}_{\mathcal{F}}=\) \(\cup_{1 \leq i, j \leq 3} \mathcal{C}_{\mathcal{F}_{i j}}\) is the same as Lemma 3.1. Then \(\mathcal{C}\) is an \((m+n,|\mathcal{C}|, 2 d, k)_{q}\) constant dimension code with \(|\mathcal{C}|=N_{1} N_{2}+N_{3} N_{4}+\sum_{1 \leq i, j \leq 3}\left|\mathcal{C}_{\mathcal{F}_{i j}}\right|\). Finally, let \(A_{q}(n, d, k)\) be the maximal sizes of an \((n, M, d, k)_q\) constant dimension code. We heve Corollary 3.3. Let \(m, n, k, d\) be integers with \(m, n \geq k \geq 3 d\). Then \[ A_{q}(m+n, 2 d, k) \geq A_{q}(m, 2 d, k) q^{m \times(k-d+1)}+\sum_{j=d}^{k-d} A_{j}(\mathcal{Q}) A_{q}(n, 2 d, k)+\sum_{1 \leq i, j \leq 3}\left|\mathcal{C}_{\mathcal{F}_{i j}}\right|. \] Specially, the maximal sizes \(A_{q}(18,6,9)\) and \(A_{q}(17,6,8)\) are discussed in the last two corollaries. From Table 2, we find these constant dimension codes own new lower bounds, which are better than the known lower bounds in [\textit{D. Heinlein} et al., Des. Codes Cryptography 87, No. 2--3, 375--391 (2019; Zbl 1409.51010)].