Spectra of linear codes and error probability of decoding (Q4027641)

[For the entire collection see Zbl 0745.00062.] Let \(C\) be a linear \([n,k,d]_ q\)-code and \(A_ i\) be the number of codewords of weight \(i\). Let \(d^ \perp\) the dual distance of \(C\). Its spectrum \(\{A_ j\}\) is defined as \(A_ j=A_ j(C)=|\{x\in C: \| x\|=j\}|\). The enumerator is given by \(W_ C(x:y)=\sum^ n_{i=0} A_ i x^{n-i} y^ i=x^ n+\sum^{n-d}_{j=0} B_ j(x- y)^ j y^{n-j}\), where \(B_ i=\sum^{n-i}_{j=d}{n-j\choose i} A_ j\). The main idea of that paper was to redevelop the enumerator \(W_ C(x:y)\) over the basis \(\bigl(x^ j(x-y)^{n-j}\bigr)\) and to estimate the corresponding coefficients \(B_ j\) which have a lucid algebraic- geometric interpretation. Finally, using algebraic-geometric codes and estimates for their spectra the authors ameliorate the Gallager bound [\textit{R. G. Gallager}, ``Information theory and reliable communications'' (1968; Zbl 0198.522)] in many cases. The details are too involved to be described here.