A generalized Gleason-Pierce-Ward theorem (Q735040)

The authors generalize the Gleason-Pierce-Ward theorem on linear divisible codes to additive divisible codes. Here, a divisible code is defined as a code whose codewords all have Hamming weights divisible by some integer \(\Delta>1\). An additive code over GF\((q)\) of length \(n\) is a subset of GF\((q)^n\) closed under vector addition, that is a GF\((p)\)-linear code of GF\((q)^n\) for \(q=p^ m\) and prime \(p\). The main result is the following: Suppose that \(\mathcal{C}\) is an additive divisible code over GF\((q)^n\) (with \(q=p^m\)) of length \(n\) and size \(p^k\) with \(k=\frac{m\cdot n}{2}\) and divisor \(\Delta>1\). Then one or more of the following holds: (I) \(q=2\) and \(\Delta =2\) (II) \(q=2\), \(\Delta=4\) and \(\mathcal{C}\) is self-dual. (III) \(q=3\), \(\Delta=3\) and \(\mathcal{C}\) is self-dual. (IV) \(q=4\), \(\Delta =2\) and \(\mathcal{C}\) is trace self-dual (see below) and (V) \(\Delta=2\) and \(\mathcal{C}\) is equivalent to the linear code over GF\((p^m)\) with generator matrix \([I_{n/2},I_{n/2}]\). An additive code \(\mathcal{C}\) over GF\((4)\) is called trace self-dual if \(\mathcal{C}=\mathcal{C}^\bot\) where the dual is formed by the trace inner product \(\langle(x_1,\dots, x_n),(y_1,\dots,y_n)\rangle=\sum^n_{i=1}(x_iy^2_i+x^2_iy_i)\). Two additive codes over GF\((q)\) are called equivalent if one can be obtained from the other by rearranging and/or re-scaling (that is applying an additive automorphism of GF\((q)\) on the coordinates). The codes described in type (V) are not necessarily linear because re-scaling only keeps GF\((p)\)-linearity.