All \(s\)-sum sets of type 2 or 3 are triple-sum sets (Q1193129)

The difference sets have been considered and studied by several authors, for example, \textit{I. M. Chakravarti} and \textit{K. V. Suryanarayana} [J. Comb. Theory, Ser. A 13, 426-431 (1972; Zbl 0251.05016)], \textit{J. Wolfmann} [J. Comb. Theory, Ser. A 23, 208-222 (1977; Zbl 0364.94020)], and \textit{P. Camion} [Difference sets in elementary Abelian groups (1979; Zbl 0455.94018)]. A natural extension is the concept of triple-sum sets, which was introduced by \textit{B. Courteau} and \textit{J. Wolfmann} [Discrete Math. 50, 179-191 (1984; Zbl 0541.94027)]. A further generalization of this is the notion of \(s\)-sum-sets, that appears in \textit{M. Griera}, the author and \textit{L. Huguet} [Algebraic algorithms and error-correcting codes, Proc. 3rd Int. Conf., Grenoble/France 1985, Lect. Notes Comput. Sci. 229, 135-142 (1986; Zbl 0625.94015)] and in the author's doctoral thesis (Barcelona, 1987). Let \(F\) be the Galois field with \(q\) elements \((q=p^ m\), \(p\) being a prime number), and let \(\Omega'\) be a subset of \(F^ k\). One can say that \(\Omega'\) is an \(s\)-sum-set with parameters \(\mu_ 0\), \(\mu_ 1\), \(\mu_ 2\) if: (a) \(F\cdot\Omega'=\Omega'\). (b) If \(R=\text{Card}\{(a_ 1,a_ 2,\dots,a_ s)\mid h=\sum a_ i, a_ i\in\Omega'\}\), then the value of \(R\) does not depend on \(h\), but only on the fact that \(h\) belongs or does not belong to \(\Omega'\). More precisely, \(R=\mu_ 0\), if \(h=0\), \(R=\mu_ 1\), if \(h\in\Omega'-0\), \(R=\mu_ 2\), if \(h\in \Omega'{}^ c-0\) \([\Omega'{}^ c=F^ k-\Omega']\). Let \(\Omega'\) be an \(s\)-sum-set. Starting from \(\Omega'\) a set \(\Omega\) by taking only one representative vector from each one-dimensional \(GF(q)\)-subspace contained in \(\Omega'\) can be build. \textit{M. Griera}, the author and \textit{L. Huguet} (loc. cit.) showed that if \(\Omega\) is the set of coordinate forms of the linear code \(C(n,k)\) and if \(X=F^*\cdot\Omega\) in an \(s\)-sum-set, then \(C(n,k)\) has at most three non-zero weights. The authors also characterized the \(s\)-sum-sets according to that error-correcting capability \(e\) of the orthogonal code \(C^ \perp\), where \(C\) is the code associated with the \(s\)-sum-set. After the above-mentioned paper, the central question about \(s\)-sum-sets is the following: does an \(s\)-sum-set that is not already a 3-sum-set exist? The main result of the present paper gives a negative response to the question in the particular case where the \(s\)-sum-set is of type 2 or 3. According to the error-correcting capability \(e\) of their associated code the \(s\)-sum-sets are characterized. It is proved that there exist only two \(s\)-sum-sets of type \(e=3\). Their associated orthogonal codes are the binary Golay code (23,12) and the repetition code of length 7. Further the characterization of \(s\)-sum-sets of type \(e=2\) is being discussed. It was mentioned that the problem of classification for the \(s\)-sum-sets of type \(e=1\) is still an open problem, the case of the 3-sum-sets included. It is also an open problem to know if there exist \(s\)-sum-sets for which \(C^ \perp\) is not a completely regular code. All the \(s\)-sum- sets that we know give completely regular codes. And it is an open question to verify the existence of some odd exponent \(s>3\), for which there exist non-zero integers \(x\), \(y\), \(z\), such that they satisfy the equation \(x^ s\cdot(y-z)+y^ s\cdot(z-x)+z^ s\cdot(x-y)=0\).