The Manin-Peyre formula for a certain biprojective threefold (Q681634)

In [Proc. Lond. Math. Soc. (3) 108, No. 4, 911--964 (2014; Zbl 1338.11058)], the authors confirmed the predictions of Manin and Peyre for the distribution of rational points on the cubic fourfold in \(\mathbb{P}^{5} \) defined by \[ x_1y_2y_3+x_2y_1y_3+x_3y_1y_2=0. \tag{1} \] In the paper under review, the authors study the same equation from a different point of view, resulting in a considerably harder problem. Regarding the polynomial on the left hand side as a linear form in \(\textbf{x} = (x_1 , x_2 , x_3 )\) and a quadratic form in \(\textbf{y} = (y_1 , y_2 , y_3 )\), the Eq. (1) defines a biprojective variety \(V\) in the product space \(\mathbb{P}^2 \times \mathbb{P}^2\). For a \(\mathbb{Q}\)-rational point on \(V\) there are representatives \(\textbf{x}, \textbf{y} \in \mathbb{Z}^3\) with \((x_1 ; x_2 ; x_3 ) = (y_1 ; y_2 ; y_3 ) = 1\), both unique up to sign. An anticanonical height function on \(V\) is then given by \[ H(\textbf{x}, \textbf{y}) = \max_{1\le i,j\le 3}|x_i|^{2}|y_j|. \tag{2} \] Rational points on \(V\) ordered with respect to the height in Eq. (2) accumulate on the subvariety cut out from \(V\) by the additional equation \(x_1 x_2 x_3 y_1 y_2 y_3 = 0\). To see this, note that the choices \(\textbf{x} = (0, 1, 1)\) and \(\textbf{y} = (y_1 , y_2 , -y_2 )\) with \((y_1 ; y_2 ) = 1\) produce more than \(B^2\) rational points of height at most \(B\) on this subvariety, while on the Zariski-open subset \(V^\circ\) of \(V\) where \(x_1 x_2 x_3 y_1 y_2 y_3 \neq 0\) the rational points are much sparser. This is a consequence of the following asymptotic formula, which is the main result in this paper. Theorem. Let \(N(B)\) denote the number of rational points on \(V^\circ \) with height not exceeding \(B\), and let \[ C=\prod_{p}\left(1-\frac{1}{p}\right)^{5}\left(1+\frac{5}{p}+\frac{5}{p^2}+\frac{1}{p^{3}}\right). \tag{3} \] Then \[ N(B)=\dfrac{\pi^{2}-3+24\log 2}{144}CB(\log B)^{4}+O\left(B(\log B)^{4-\frac{1}{480}}\right). \tag{4}\] While a simple argument is sufficient to show that \(N(B) \asymp B(\log B)^4\), the precise asymptotic formula confirming the Manin-Peyre conjecture requires a new approach, different from the ones used previously to study other contexts. With more work it is possible to show that there is a polynomial \(P\) of degree four and a positive number \( \delta \) with the property that \[ N(B)=BP(\log B)+O\left(B^{1-\delta}\right). \tag{5}\] In order to keep the paper at reasonable length, the authors leave the proof of Eq. (5) as an exercise to the reader and they only restrict themselves to the detailed proof of Eq. (4). The techniques presented in the paper should also be applicable to more general contexts. In fact, the variety studied in this paper is nothing more than the case \(n = 3\) of the family of varieties defined by \[ \dfrac{x_1}{y_1}+\dfrac{x_2}{y_2}+\cdots+\dfrac{x_n}{y_n}=0. \] The authors hope that the present example spurs further work on higher dimensional cases of the Manin-Peyre formula.