Hadamard matrix of order 16 multiplied with a vectorNaturally ordered Hadamard matrix permuted into sequency-ordered Walsh matrix. The number of sign changes per row in the naturally ordered matrix is (0, 15, 7, 8, 3, 12, 4, 11, 1, 14, 6, 9, 2, 13, 5, 10), in the sequency-ordered matrix the number of sign changes is consecutive.LDU decomposition of a Hadamard matrix. The ones in thetriangular matrices formSierpinski triangles. The entries of thediagonal matrix are values fromGould's sequence, with the minus signs distributed like the ones inThue–Morse sequence.
The Walsh matrices are a special case ofHadamard matrices where the rows are rearranged so that the number of sign changes in a row is in increasing order. In short, a Hadamard matrix is defined by therecursive formula below and isnaturally ordered, whereas a Walsh matrix issequency-ordered.[1] Confusingly, different sources refer to either matrix as the Walsh matrix.
The Walsh matrix (and Walsh functions) are used in computing theWalsh transform and have applications in the efficient implementation of certainsignal processing operations.
We can obtain a Walsh matrix from a Hadamard matrix. For that, we first generate the Hadamard matrix for a given dimension. Then, we count the number of sign changes of each row. Finally, we re-order the rows of the matrix according to the number of sign changes in ascending order.
For example, let us assume that we have a Hadamard matrix of dimension
,
where the successive rows have 0, 3, 1, and 2 sign changes (we count the number of times we switch from a positive 1 to a negative 1, and vice versa). If we rearrange the rows in sequency ordering, we obtain:
where the successive rows have 0, 1, 2, and 3 sign changes.
The sequency ordering of the rows of the Walsh matrix can be derived from the ordering of the Hadamard matrix by first applying thebit-reversal permutation and then theGray-codepermutation:[2]
where the successive rows have 0, 1, 2, 3, 4, 5, 6, and 7 sign changes.