Simultaneous integer values of pairs of quadratic forms (Q2397737)

While there is an extensive theory concerning integers represented by integral quadratic forms, there is much less known about pairs of integers \((n_1,n_2)\) represented by pairs \((Q_1,Q_2)\) of integral quadratic forms \(Q_i\in\mathbb{Z}[x_1,\ldots,x_k]\), i.e., where there exists an \( {\mathbf{x}}\in \mathbb{Z}^k\) with \(Q_i({\mathbf{x}})=n_i\), \(i=1,2\). In the present paper, the authors address this question. Here, an integral quadratic form \(Q\) is understood to be of the form \(Q( {\mathbf{x}})=b( {\mathbf{x}},{\mathbf{x}})\) for some symmetric bilinear form \(b\) with coefficients in \(\mathbb{Z}\). The motivating idea is that if \(k\) is large enough and under a certain nonsingularity assumption on the pair of quadratic forms, then at least all but finitely many pairs \((n_1,n_2)\) should be representable provided the necessary local conditions are satisfied. The nonsingularity condition that the authors impose is that the projective variety defined by the equations \(Q_1=Q_2=0\) over the algebraic closure \(\overline{\mathbb{Q}}\) is nonsingular in the sense that for every \( {\mathbf{x}}\in\overline{\mathbb{Q}}^k\setminus\{ 0\}\) with \(Q_1({\mathbf{x}})=Q_2( {\mathbf{x}})=0\), one has \[ \text{rk}\begin{pmatrix} \nabla Q_1(\mathbf{x})\\ \nabla Q_2( \mathbf{x})\end{pmatrix}=2. \] Under this nonsingularity assumption and provided \(k\geq 5\), the authors show that if \({\mathcal E}(N)\) (\(N>0\)) denotes the number of pairs of integers \((n_1,n_2)\) with \(|n_i|\leq N\) and for which \(Q_i({\mathbf{x}})=n_i\), \(i=1,2\), has solutions over \(\mathbb{R}\) and each \(\mathbb{Z}_p\), then \[ {\mathcal E}(N)\ll_{Q_1,Q_2}N^{2-m}\quad\text{ with }m=1/(8k^3). \] They show that if in addition the \(Q_i\) represent simultaneously some positive value over \(\mathbb{R}\), and that if for each prime \(q\) there exists an \( {\mathbf{x}}_q\in \mathbb{Z}^k\) such that \(q\) does not divide \(Q_1( {\mathbf{x}}_q)Q_2({\mathbf{x}}_q)\), then there are infinitely many primes represented simultaneously by \(Q_1\) and \(Q_2\) over \(\mathbb{Z}\). As an example, the authors point out that for two diagonal forms \(Q_1=\sum_{m=1}^ka_mx_m^2\) and \(Q_1=\sum_{m=1}^kb_mx_m^2\), the nonsingularity condition is equivalent to all ratios \(a_m/b_m\) being distinct, so that for example the two forms \(Q_1=\sum_{m=1}^5x_m^2\) and \(Q_2=\sum_{m=1}^5mx_m^2\) represent simultaneously infinitely many primes. The authors point out that their result on simultaneous representation of primes fails if the nonsingularity assumption is dropped, but they conjecture that the result may still hold if the nonsingularity condition is replaced by the assumption that neither \(Q_1\) nor \(Q_2\) factors in \(\mathbb{Z}[x_1,\ldots,x_k]\). The proofs are long and technical. They are based on the two-dimensional circle method involving as novel ingredient a two-dimensional Kloosterman refinement applied to the contribution of the minor arcs.