Gröbner bases and the Nagata automorphism (Q1295695)

This paper contains an assortment of ideas circling around problems of automorphisms of polynomial rings and in particular the Nagata automorphism. Recall that the Nagata automorphism, given by \textit{M. Nagata} [``On automorphism group of \(k[x,y]\)'', Dep. Math., Kyoto Univ., Lect. Math. 5 (1972; Zbl 0306.14001)], is the automorphism \(\nu\) of \(K[x,y,z]\) defined by \[ \nu(x) = x - 2 y (y^2 + x z) - z (y^2 + x y)^2, \quad \nu(y) = y + z (y^2 + x z), \quad \nu(z) = z. \] Nagata conjectured that \(\nu\) is not tame, i.e., it cannot be written as a composition of affine automorphisms and triangular automorphisms, defined in the obvious way. This would provide a counter example to the conjecture that every automorphism of \(K[x_1,\ldots,x_n]\) is tame (which is true for \(n = 2\)). In the paper under review, two approaches to Nagata's conjecture are suggested. The first one considers reduced Gröbner bases of face functions of an automorphism of \(K[x_1,\ldots,x_n]\). For details see the original paper. The authors prove that for \(n = 2\) such Gröbner bases give rise to automorphisms of \(K[x_1,x_2]\). They conjecture that the same is true for any \(n\) and any tame automorphism, for at least one choice of an admissible monomial ordering. This would imply that \(\nu\) is not tame. The authors have some experimental evidence in favor of their conjecture. The second approach is by trying to lift the Nagata automorphism to an automorphism of the free associative algebra \(K\langle x,y,z \rangle\). Since every tame automorphism can be lifted to an automorphism of \(K\langle x,y,z \rangle\), the impossibility to lift \(\nu\) would imply Nagata's conjecture. The paper contains some more interesting ideas and many conjectures and problems.