Some irreducibility results for truncated binomial expansions (Q616430): Difference between revisions

For positive integers \(n>k,\) let \(P_{n,k}(x)=\sum_{j=0}^k \binom{n}{j}x^j\) be the polynomial obtained by truncating the binomial expansion of \((1+x)^n.\) The authors prove the irreducibility of \(P_{n,k}(x)\) over the rational numbers when \(2\leq 2k\leq n<(k+1)^3.\) This paper builds on and extends results of \textit{M. Filaseta, A. Kumchev} and \textit{D. Pasechnik} [Rocky Mt. J. Math. 37, 455--464 (2007; Zbl 1211.12004)]. Let \(c_j=\sum_{i=j}^k\binom{n}{i}\binom{i}{j}(-1)^{i-j}.\) Then \(P_{n,k}(x-1)=\sum_{j=0}^k c_jx^j.\) More generally, the authors consider \(F_{n,k}(x)=\sum_{j=0}^k a_jc_jx^j,\) where \(a_0,a_1,\ldots,a_k\) are nonzero integers such that each \(a_i\) has all of its prime factors \(\leq k.\) They prove the following. Let \(k\) and \(n\) be positive integers such that \(8\leq 2k\leq n<(k+1)^3.\) Then \(F_{n,k}(x)\) is irreducible over \(\mathbb{Q}\) except possibly when \((n,k)\) is in the set \(\{(8,4),(10,5),(12,6),(16,8)\}.\) The proof relies on Newton polygons with respect to primes exceeding \(k\) that divide \(\binom{n}{k}\) and results on such primes from \textit{S. Laishram} and \textit{T. N. Shorey} [Acta Arith. 113, 4, 327--341 (2004; Zbl 1046.11004)].