Primitive divisors of Lucas and Lehmer sequences. II (Q679090)

[For Part I, see Math. Comput. 64, 869--888 (1995; Zbl 0832.11009).] Let \(\alpha, \beta\) be quadratic numbers such that \(\alpha+\beta\) and \(\alpha \beta\) are relatively prime rational integers and \(\alpha/\beta\) is not a root of unity. The sequence of rational integers defined by \[ u_n= {\alpha^n- \beta^n \over\alpha -\beta},\;n= 0,1,2, \ldots \] is called a Lucas sequence. If we assume the weaker condition \((\alpha+ \beta)^2 \in\mathbb{Z}\) and leave all other conditions as above then we can define another sequence of rational integers, namely \[ u_n= \begin{cases} {\alpha^n- \beta^n \over\alpha- \beta} \quad & \text{if } n \text{ is odd} \\ {\alpha^n- \beta^n \over \alpha^2 -\beta^2} \quad & \text{if } n\text{ is even,} \end{cases} \quad n=0,1,2, \dots \] which is called a Lehmer sequence. A prime number \(p\) is called a primitive divisor of a Lucas number \(u_n\) if \(p\) divides \(u_n\) but does not divide \((\alpha- \beta)^2 u_2 \cdots u_{n-1}\); \(p\) is called a primitive divisor of a Lehmer number \(u_n\) if it divides \(u_n\) but not \((\alpha^2- \beta^2)^2 u_3 \cdots u_{n-1}\). As a consequence of Theorem 2 of his paper, the author proves: If \(\alpha\) and \(\beta\) generate a Lucas or Lehmer sequence, then the \(n\)-th term of this sequence has a primitive divisor for all \(n>2 \cdot 10^{10}\). This considerably improves on an earlier result of similar type due to C. Stewart, in which the lower bound for \(n\) was \(e^{452} 2^{67}\) for Lucas sequences and \(e^{452} 4^{67}\) for Lehmer sequences. The other main result of this paper is: If \(\alpha\) and \(\beta\) generate a Lucas or Lehmer sequence and \(h(\beta/ \alpha) \leq 4\), then for all \(n>30\) the \(n\)-th term of this sequence has a primitive divisor. The proofs are based on the theory of linear forms in logarithms of algebraic numbers, making use of a sharp lower bound for such linear forms in two logarithms, due to Laurent, Mignotte and Nesterenko.