Evaluation of the Jacobian of a rational transformation of \(\mathbb{C}^2\) and some applications (Q1972533)

Statement of the result: Consider \(\mathbb{C}^2\) with coordinates \(x, y\). Let us perform the change of variables \(x\mapsto x_0/y_0\), \(y\mapsto 1/y_0\). Take a positive integer \(l\) and a set \(p\) of \(l-1\) complex numbers: \(p=p_1\), \(p_2, \dots, p_{l-1}\). Let us successively perform the changes of variables \(x_0 \mapsto x_1y_1\), \(y_0\mapsto y_1\), \(x_i-p_i\mapsto x_{i+1} y_{i+1}\), and \(y_i \mapsto y_{i+1}\) \((i=1,2, \dots, l-1)\). The equalities \(y_i=0\) \((i=1,2, \dots, l)\) specify spheres denoted by \(\sigma_i\). By \(M(p)\) we denote the manifold formed by \(\mathbb{C}^2\) and these spheres. Let \(F\) and \(G\) be a pair of polynomials in \(x, y\). The changes of variables specified above transform them into the functions \(F_l= y_1^{-k} f_l(x_l,y_l)\) and \(G_l=y_l^{-k} g_l(x_l,y_l)\), where \(k\) is a nonnegative integer and \(f_l\) and \(g_l\) are polynomials in \(x_l, y_l\). Let \(J\) and \(J_l\) denote the Jacobians of the pairs of functions \(F,G\) and \(F_l,G_l\), respectively. It is easy to verify that if \(J(x,y)=1\), then \(J_l(x_l, y_l)= -y_l^{l-3}\). Clearly, if \(l<3\), then \(k\geq 1\); if \(l\geq 3\), then it is not inconceivable that \(k=0\), i.e., that both functions \(F_l\) and \(G_l\) are polynomials. Such a pair of polynomials \(F\) and \(G\) would give a counterexample to the Jacobian conjecture. Thus, for an arbitrary \(l\), the question on the existence of such a pair is open. For \(l=3\), we prove the following assertion in \S 5. Proposition 1. If \(J(x,y)= \text{const} \neq 0\) for a pair of polynomials \(F,G\), then, for any \(p_1\) and \(p_2\), at least one of the functions \(F_3(x_3, y_3)\) and \(G_3(x_3, y_3)\) is not a polynomial. This assertion can be reformulated as follows: If \(J(x,y) \equiv 1\) for a pair of polynomials \(F(x,y)\) and \(G(x,y)\), then, for any choice of numbers \(a\) and \(b\), the functions \(F(xy^2+ ay+ b,1/y)\) and \(G(xy^2+ ay+ b,1/y)\) cannot be polynomial simultaneously. This answers the question discussed earlier [\textit{A. G. Vitushkin}, Math. Notes 65, No. 5, 539-547 (1999); translation from Mat. Zametki 65, No. 5, 643-653 (1999; Zbl 0947.14031)]. If only a similar assertion was valid for \(l=4\), it would be possible to prove that the ``exotic'' coverings of \(\mathbb{C}^2\) constructed in relation to the Jacobian conjecture cannot be represented by polynomial maps; see the paper cited above for more details.