On the Colmez conjecture for non-abelian CM fields

The Colmez conjecture relates the Faltings height of an abelian variety with complex multiplication by the ring of integers of a CM field $E$ to logarithmic derivatives of certain Artin $L$--functions at $s=0$. In this paper, we prove that if $F$ is any fixed totally real number field of degree $[F:\mathbb{Q}] \geq 3$, then there are infinitely many CM extensions $E/F$ such that $E/\mathbb{Q}$ is $\textit{non-abelian}$ and the Colmez conjecture is true for $E$. Moreover, these CM extensions are explicitly constructed to be ramified at "arbitrary" prescribed sets of prime ideals of $F$. We also prove that the Colmez conjecture is true for a generic class of non-abelian CM fields called Weyl CM fields, and use this to develop an arithmetic statistics approach to the Colmez conjecture based on counting CM fields of fixed degree and bounded discriminant. We illustrate these results by evaluating the Faltings height of the Jacobian of a genus 2 hyperelliptic curve with complex multiplication by a non-abelian quartic CM field in terms of the Barnes double Gamma function at algebraic arguments. This can be seen as an explicit non-abelian Chowla-Selberg formula. A crucial input to the proofs is an averaged version of the Colmez conjecture which was recently proved independently by Andreatta-Goren-Howard-Madapusi Pera and Yuan-Zhang.