A polynomial Zsigmondy theorem (Q661902): Difference between revisions

Let \(k\) be a field, and let \((F_n)_{n\geq 1}\) be a sequence of elements of \(k[T]\). Note that \(k[T]\) is a unique factorization domain. An irreducible factor that divides a term in the sequence but does not divide an earlier term is called a primitive prime divisor. In particular, the authors consider the sequence \(F_n=f^n-g^n\), where \(f\) and \(g\) are nonzero coprime elements of \(k[T]\) that are not both units. The main result of this paper is the following. If \(\text{char}(k)=0,\) then each term of \((F_n)_{n\geq 1}\) beyond the second term has a primitive prime divisor. If \(\text{char}(k)=p>0,\) let \(F'\) be the sequence obtained from \((F_n)_{n\geq 1}\) by deleting the terms with \(p\mid n.\) Then each term of \(F'\) beyond the second term has a primitive prime divisor. This is an analogue of work of \textit{A. S. Bang} [Zeuthen Tidskr. (5) 4, 70--80, 130--137 (1886; JFM 19.0168.02)] and \textit{K. Zsigmondy} [``Zur Theorie der Potenzreste'', Monatsh. Math. 3, 265--284 (1892; JFM 24.0167.02)] on the sequence \((a^n-b^n)_{n\geq 1},\) where \(a>b>0\) are coprime integers. (In that case, every term beyond the sixth has a primitive prime divisor.)