On the equivalence of certain constant weight codes and combinatorial designs (Q1361726): Difference between revisions

Let \(F\) be the field of two elements. Denote by \(F^{n,w}\) the subset of all vectors of Hamming weight \(w\) of \(F^n\). A subset \(C \subseteq F^{n,w}\) is called a constant weight code with parameters \(n,w,d,N_C\), where \(d\) denotes the minimum distance and \(N_C\) the size of \(C.\) Such a code is denoted by \(C:(n,w,d,N_C)\). A constant weight code \(C:(b,r,2(r-\lambda),\nu)\) is called a \((\nu,b,r,k,\lambda)\) design if (i) one has \(d(x,y)=2(r-\lambda)\) for all distinct \(x,y \in C\), and (ii) any column of the matrix whose rows are the code vectors of \(C,\) has the same weight \(k.\) The first theorem of the present paper states: Let \(E:(n=4s+1,d=2s,N_E)\) be an equidistant code, \(C:(n=4s+1,w,d=2s,N_C)\) with \(w^2>n(w-d/2)\) a constant weight code, and \(D:(n,n,k,k,\lambda)\) a design. Then any one of the conditions (i) \(N_E = n,\) (ii) \(N_C = n,\) (iii) \(n = 2u^2+2u+1, k=u^2, \lambda=(u^2-u)/2\) implies the other two conditions. Another remarkable result is the following: A Hadamard matrix of order \(n\) exists if and only if a constant weight code \(C:(n,n/2,n/2,2n-2)\) exists.