Mathematics

Erdős problem 521 is resolved for random polynomials

The number of real roots of a random polynomial of degree \(n\), divided by \(\log n\), almost surely has no limit; its limit points fill an explicit interval.

Filed under Mathematics, Probability

Erdős problem 521 concerns whether \(N_n/\log n\), the normalised number of real roots of a polynomial with independent random signs as coefficients, converges almost surely. Chen et al. show that it does not.

For Kac polynomials \(P_n(x)=\sum_{j=0}^n \xi_j x^j\) with i.i.d. coefficients of mean zero, variance one and a finite exponential moment, they prove that almost surely \[ \liminf_{n\to\infty} \frac{N_n}{\log n} = \frac{1}{\pi}, \qquad \limsup_{n\to\infty} \frac{N_n}{\log n} = \frac{1}{\pi} + C_*, \] where \(C_* > 1/\pi\) is fixed by an equation, and that the set of subsequential limits is exactly \([1/\pi, 1/\pi + C_*]\). The lower limit holds under only a \(2+\delta\) moment.