Degree reduction under specialization (Q5951562): Difference between revisions

The main application of this paper is the resolution of some parametric systems of polynomial equations. This is done looking for which information remains in a Gröbner basis under specialization. Let \(R\) be a ring and let \(\varphi:R[x] \rightarrow R[x]\) be a specialization, i.e. a ring homomorphism such that \(\varphi(x) = x\). For any \(f \in R[x]\) such that \(\varphi(f) \neq 0\), \(\delta(f)\) is the integer indicating how much the degree of \(f\) decreases under \(\varphi\), i.e. \(\delta(f) = \deg f - \deg \varphi(f)\). The authors prove that for any pair \(f,g \in R[x]\) with \(\delta(g) < \delta(f)\) there exists \(h \in (f,g)\) such that \(\delta(h) \leq \delta(g)\) and \(\varphi(h) = c \varphi(f)\) for a suitable \(c \in \varphi(R) \setminus \{0\}\). Let \(k\) be a field, \(\alpha \in K^r\) and let \(\varphi_\alpha\) be the ring endomorphism on \(k[x,T]\), where \(T = (t_1, \dots, t_r)\), that evaluates \(T\) at \(\alpha\). If \(I\) is an ideal of \(k[x,T]\) and \(\delta_I=\min \{\delta(h)\mid h \in I, \varphi_\alpha(h) \neq 0 \}\), a direct consequence of the previous statement is that every element in \(\varphi_\alpha(I)\) is the image under \(\varphi_\alpha\) of an element in \(I\) whose degree decrease is \(\delta_I\). This is used to prove the main theorem of section 2: If \(G\) is a Gröbner basis of \(I\) with respect to a block order \(>\) for \(x,T\), and \(g_m\) is the smallest (with respect to \(>\)) polynomial in \(G\) such that \(\varphi_\alpha(g_m) \neq 0\) then \(\varphi_\alpha(g_m)\) generates \(\varphi_\alpha(I)\) and \(\delta(g_m) = \delta_I\). The applications of the previous theorem are in section 3. The parametric GCD of a set of polynomials in \(k[x,T]\) can be easily computed via a Gröbner basis. If \(a\) is a zero of the generator of \(\varphi_\alpha(I)\) referred to in the previous theorem, where \(\alpha \in V(I\cap k[T])\) and \(\varphi_\alpha(I) \neq 0\), then \((a,\alpha)\) belongs to \(V(I)\). This new theorem is the key to solve systems of parametric polynomial equations in several variables \(X=(x_1, \dots, x_n)\). If a block order in \(X,T\) with the \(X\)-variables lexicographically ordered (\(x_1 > \dots > x_n\)) is chosen the elements of a Gröbner basis \(G\) for the ideal generated by the equations can be collected in blocks \(B_1, \dots, B_n\) [they exist thanks to \textit{W. Trinks}, J. Number Theory 10, 475-488 (1978; Zbl 0404.13004)] which allows the extensions of solutions if these blocks do not vanish under intermediate specializations. In the two variables case some solutions can be also found even if they vanish.