Sharp bounds for the number of roots of univariate fewnomials (Q533812): Difference between revisions

Let \(K\) be a field and \(t\geq 0\) be an integer. Define \(B_1(K,t)\) (resp. \(B_m(K,t)\)) the supremum of the number of distinct (resp. counted with multiplicity) roots in \(K^*\) that a polynomial with coefficients in \(K\) and \(t+1\) terms can have. The authors give a number of lower/upper bounds for \(B_1(K,t)\) and \(B_m(K,t)\) under some hypotheses on \(K\). For instance, if \(p\) is an odd prime number and \(K\) is a finite extension of \({\mathbb Q}_p\) with residue field \({\mathbb F}_q\) and ramification index \(e\), they prove that \(B_1(t,K)\geq (2t-1)(q-1)\) and, under the assumption \(p\geq e +t\), that \(B_m(t,K)\leq (t^2-t+1)(q-1)\). This settles the problem completely for \(t=2\) and \(K={\mathbb Q}_p\) with \(p\geq 5\). The main tool in the paper is the study of the properties of generalized Vandermonde determinants.