Condensation of the roots of real random polynomials on the real axis

Abstract

We introduce a family of real random polynomials of degree n whose coefficients a_k are symmetric independent Gaussian variables with variance = e^{-k^α}, indexed by a real α\geq 0. We compute exactly the mean number of real roots for large n. As αis varied, one finds three different phases. First, for 0 \leq α< 1, one finds that \sim (\frac{2}π) \log{n}. For 1 < α< 2, there is an intermediate phase where < N_n > grows algebraically with a continuously varying exponent, < N_n > \sim \frac{2}π \sqrt{\frac{α-1}α} n^{α/2}. And finally for α> 2, one finds a third phase where \sim n. This family of real random polynomials thus exhibits a condensation of their roots on the real line in the sense that, for large n, a finite fraction of their roots /n are real. This condensation occurs via a localization of the real roots around the values \pm \exp{[\fracα{2}(k+{1/2})^{α-1} ]}, 1 \ll k \leq n.