MacWilliams identities

From Encyclopedia of Mathematics
Revision as of 17:19, 7 February 2011 by (talk) (Importing text file)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

A -dimensional subspace of the linear space of all -tuples over the finite field is called an linear code over . The vectors in are called code words. The weight of a vector is the number of its non-zero components. A generator matrix for a code is any -matrix whose rows form a basis of . The ordinary inner product of vectors , in is

The dual of is the linear code defined by

The weight distribution of a code of length gives the number of vectors of each weight , , denoted by . For example, let be the linear code over with generating matrix

The only non-zero weights in are , .

While the weight distribution does not, in general, uniquely determine a code, it gives important information that has both practical and theoretical significance. The most fundamental result about weight distributions are the MacWilliams equations, a set of linear relations between the weight distributions of and . These equations imply that the weight distribution of is uniquely determined by the weight distribution of . These relations have been the most significant tool available for investigating and calculating weight distributions.

Let denote the number of vectors of weight in . If is an linear code over , then there are two equivalent sets of MacWilliams equations [a1]:


where .

If , both sets of equations say that , which states the obvious result that an code over contains codewords. The other equations are not so obvious. One can solve either set of equations for the weight distribution of , obtaining and .

An elegant and much used way to express these relations is in terms of polynomials. The weight enumerator of a code of length is:

The following equation is equivalent to either (a1) or (a2) for a code over :


Two other sets of equations equivalent to either (a1), (a2) or (a3) and more convenient for some calculations are called the power moment identities [a3]. They involve the Stirling numbers of the second kind (cf. also Combinatorial analysis). One such set is


where .

The power moments are particularly useful for giving conditions under which the weight distributions of a code and its dual are uniquely determined. This is illustrated in the following theorem: A unique solution to the power moments exists under the conditions that or fewer 's are unknown and that are known [a3]. In this situation, the weight distributions of both and are uniquely determined.

There are also MacWilliams equations for non-linear codes [a2].


[a1] F.J. MacWilliams, "A theorem on the distribution of weights in a systematic code" Bell System Techn. J. , 42 (1963) pp. 79–94
[a2] F.J. MacWilliams, N.J.A. Sloane, J.M. Goethals, "The MacWilliams identities for nonlinear codes" Bell System Techn. J. , 51 (1972) pp. 803–819
[a3] V. Pless, "Power moment identities on weight distributions in error-correcting codes" Inform. and Control , 6 (1963) pp. 147–152
How to Cite This Entry:
MacWilliams identities. Encyclopedia of Mathematics. URL:
This article was adapted from an original article by V. Pless (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article