On the nonvanishing of homogeneous product sums (Q5903124)

\textit{M. Ward} [Am. Math. Mon. 62, 155-160 (1955; Zbl 0064.04008)] conjectured that the diophantine equation \(H_n=0\), where \(H_n=H_n(x,y,z)\) is the homogeneous product sum \[\sum_{p+q+r=n}x^p y^q z^r, \] has only solutions with \(xyz=0\) if \(n>1\). Further, Ward proved that \(H_{2n}(x,y,z)\ne 0\) for all \(n\ge 1\) and \(H_n(x,y,z)\ne 0\) if \(n+2\) is a prime number \(>3\). Here and below it is always assumed that \(xyz\ne 0\). Ward's methods are based on algebraic number theory. The present paper studies \(H_n(x,y,z)\) by elementary methods and states some new results for it. For example: if exactly one of \(x,y,z\) is odd, then \(H_n\equiv 1\pmod 2\) for all \(n\ge 1\); if exactly two of \(x,y,z\) are odd, then \(H_n\equiv n+1 \pmod 2\); If all three of \(x,y,z\) are odd, then \(H_n\equiv [n/2]+1 \pmod 2\), and some other results. Thus, the paper adds considerably more evidence in support of Ward's conjecture.