Equidistribution of Heegner points and ternary quadratic forms (Q634796)

The authors prove new equidistribution results for Galois orbits of Heegner points with respect to single reduction maps at inert primes. Let \(N\) be a positive integer and \(X_0(N)_{/\mathbb Q}\) be the modular curve associated to the congruence subgroup \(\Gamma_0(N)\). Let \(\ell\) be a prime such that \((\ell,N)=1\) and let \(\mathcal D_N\) be the set of all fundamental discriminants \(D<0\) such that every prime factor of \(N\) splits in \(K_D:=\mathbb Q(\sqrt D)\) and such that \(\ell\) is inert in \(K_D\). Let also \(\Omega_N\) be the set of pairs \((D,c)\) with \(D\in\mathcal D_N\) and \((c,N)=1\). For \((D,c)\in\Omega_N\), the Heegner points \(x\) of discriminant \(c^2D\) on \(X_0(N)\) are defined over the ring class field \(K_D[c]\). Let \(\Gamma_{D,c}:=\{\sigma x: \sigma\in\mathrm{Gal}(K_D[c]/K_D)\}\) be the Galois orbit. Since \(\ell\) is inert in \(K_D\), the Heegner points of discriminant \(c^2D\) reduce to supersingular points modulo \(\ell\). The uniform distribution of Galois orbits of Heegner points with respect to these reduction maps was the key step in the argument of Cornut and Vatsal for the proof of Mazur's conjecture on the non-triviality of Heegner points over the \(p\)-adic anticyclotomic tower. To state the new equidistribution result in the paper, let \(X_0(N)^{\mathrm{SS}}_{/\mathbb F_{\ell^2}}\) be the set of supersingular points on \(X_0(N)\) modulo \(\ell\). Recall that each point \(s\) in \(X_0(N)^{\mathrm{SS}}_{/\mathbb F_{\ell^2}}\) represents a pair \(\mathbb E=(E,C)\) of a superelliptic elliptic curve \(E_{/\mathbb F_{\ell^2}}\) and a cyclic subgroup \(C\) of \(E\) of order \(N\). The endomorphism ring of \(\mathbb E\) is an Eichler order \(R_s\) of level \(N\) in the definite quaternion algebra over \(\mathbb Q\) ramified precisely at \(\ell\) and \(\infty\). Let \(w_s\) denote the number of units in \(R_s\) and define the canonical measure on \(X_0(N)^{\mathrm{SS}}_{/\mathbb F_{\ell^2}}\) by \[ \mu_{\mathrm{can}}(s):=\frac{1/w_s}{\sum_{s'\in X_0(N)^{\mathrm{SS}}_{/\mathbb F_{\ell^2}}}1/w_s'}. \] Then the main result of the paper states that if we let \[ \mu_{D,c}(s):=\frac{\#\{x\in\Gamma_{D,c}:\text{ the reduction of }x \text{ is }s\}}{\#\Gamma_{D,c}}, \qquad s\in X_0(N)^{\mathrm{SS}}/\mathbb F_{\ell^2}, \] then the weak-\(^*\) limit \(\lim_{-c^2D\to\infty,(D,c)\in\Omega_N}\mu_{D,c}\) exists and equals \(\mu_{\mathrm{can}}\). Here to say that the weak-\(^*\) limit of a sequence of measure \(\{\mu_n\}\) on a finite set \(X\) exists and converges to a measure \(\mu\) on \(X\) means that for each function \(f:X\to\mathbb R\), the limit \(\lim_{n\to\infty}\int_Xf\,d\mu_n\) exists and equals \(\int_Xf\,d\mu\). The authors' results generalize the results of Cornut and Vatsal in the sense that the authors allow both the fundamental discriminant and the conductor to grow.