On the weight distribution of convolutional codes (Q2568379)

It is known that there do not exist MacWilliams relations for the weight enumerator of atomic codewords for convolutional codes. The author examines a different type of weight enumerator, namely the adjacency matrix of the state diagram associated with a minimal realization of a minimal encoder. This enumerator induces an invariant of the code, namely the generalized adjacency matrix. The author uses this to prove that two codes with the same generalized adjacency matrix will have the same dimension and Forney indices and that two one dimensional binary codes with the same generalized adjacency matrix are monomially equivalent. As a corollary, it is shown that if two binary one-dimensional convolutional codes have the same generalized adjacency matrix then their dual codes have the same generalized adjacency matrix as well. A MacWilliams duality is given for codes with overall constraint length 1 but a general duality remains elusive.