Combining subspace codes (Q6112253)

For two subspaces \(U\) and \(U'\) of \(\mathbb F_q^n\), define \[ d(U,U')=\dim(U+U')-\dim(U\cap U')=\dim(U)+\dim(U')-2\dim(U\cap U'). \] The set of all subspaces of \(\mathbb F_q^n\) equipped with the distance function \(d(\cdot\,,\,\cdot)\) is a metric space. A set \(\mathcal C\) of \(k\)-dimensional subspaces of \(\mathbb F_q^n\) with \(\#\mathcal C=\Lambda\) is called a \((n,\Lambda,d;k)_q\) \textit{constant-dimension code} (CDC) if its minimum distance \[ d(\mathcal C)=\min\{d(U,U'):U,U'\in\mathcal C,\ U\ne U'\}=d. \] CDCs with \(d=2k\) are precisely partial \(k\)-spreads. Let \(A_q(n,d;k)\) denote the largest possible size of an \((n,\star,d;k)_q\) CDC. The main objective of the paper is to improve the existing lower bounds on \(A_q(n,d;k)\) through secondary constructions of CDCs. The ingredients of these constructions are CDCs, rank metric codes, and partial spreads in ambient spaces of smaller dimensions. The approach is to combine the ingredients in various ways to generate CDCs in ambient spaces of larger dimensions with a prescribed minimum distance. It is demonstrated that the new constructions provide improvements on the lower bounds on \(A_q(n,d;k)\) for many parameters, including, notably, \(A_q(10,4;5)\), \(A_q(12,4;4)\), \(A_q(12,6;6)\) and \(A_q(16,4;4)\).