“…In his letter to Smyth dated 24 February 1998, Serre showed that, for any t > 0 and γ ∈ [0, 1], the inequality However, we know that this measure is not perfectly optimized for the trace problem. Indeed, the 14 totally positive algebraic integers α with tr(α)/ deg(α) at most 1.793 have all conjugates lying in the support of µ [22], while the support of an optimal measure would contain none of these points by Proposition 5.8.…”