A bound for orders in differential Nullstellensatz (Q2267443)

Let \(\mathbf k\) be a \textit{differential field} of characteristic 0, i.e., a field of characteristic 0 endowed with a finite set of derivations \(\Delta = \{{\partial}_1, \dots ,{\partial}_m\}\). Let \({\mathcal R} = \mathbf k\{y_1, \dots ,y_n\}\) be the \textit{ring of differential polynomials} in the \textit{differential indeterminates} \(y_1, \dots ,y_n\). As a ring, \(\mathcal R\) is the polynomial ring over \(\mathbf k\) in the infinite set of indeterminates \({\partial}^{\alpha}y_j\), \(\alpha \in {\mathbb N}^m\), \(1 \leq j \leq n\). The derivations \({\partial}_1, \dots ,{\partial}_m\) can be extended to derivations on \(\mathcal R\) in an obvious manner. An ideal \(I\) of the ring \(\mathcal R\) is called a \textit{differential ideal} if \({\partial}_i(I) \subseteq I\), \(i = 1, \dots ,m\). If \(F\) is a subset of \(\mathcal R\) then the differential ideal \([F]\) generated by \(F\) is the union of the algebraic ideals \((F^{(\leq h)})\) generated by the sets \(F^{(\leq h)}\) consisting of the elements \({\partial}^{\alpha}f\), where \(f \in F\), \(\alpha \in {\mathbb N}^m\) and \(|\, \alpha \, | \leq h\), for \(h \geq 0\). A well-known lemma asserts that if \(g^d \in (F)\) then \(({\partial}_ig)^{2d-1} \in (F^{(\leq d)})\), \(i = 1, \dots , m\). It follows that the algebraic radical of a differential ideal is a differential ideal. Now, let \(H\) and \(D\) be positive integers, \(F\) a finite subset of \(\mathcal R\) and \(g\) an element of \(\mathcal R\) such that \(g\) and the elements of \(F\) are polynomials in the indeterminates \({\partial}^{\alpha}y_j\), \(1 \leq j \leq n\), \(\alpha \in {\mathbb N}^m\), \(| \, \alpha \, | \leq H\), of degrees \(\leq D\). The main result of the paper asserts that if \(g\) belongs to the radical of \([F]\) then there exists a positive integer \(t \leq A(m+8,\text{max}(n,H,D))\) such that \(g\) belongs to the radical of \((F^{(\leq t)})\). Here \(A\) is the \textit{Ackermann function} defined recursively by: \[ \begin{gathered} A(0,n) = n+1,\;A(m+1,0) = A(m,1),\\ A(m+1,n+1) = A(m,A(m+1,n))\, .\end{gathered} \] This result answers a question addressed by \textit{A. Seidenberg} [``An elimination theory for differential algebra,'' Univ. California Publ. Math., n. Ser. 3, 31--66 (1957; Zbl 0083.03302)]. The proof uses differential elimination techniques for which the basic reference is the book of [\textit{E.R. Kolchin}, Differential algebra and algebraic groups. Pure and Applied Mathematics, 54, Academic Press, New York (1973; Zbl 0264.12102)]. The Ackermann function appears through a result of [\textit{G. Moreno Socias}, An Ackermannian polynomial ideal. Applied algebra, algebraic algorithms and error-correcting codes, Proc. 9th Int. Symp., AAECC-9, New Orleans/LA (USA) 1991, Lect. Notes Comput. Sci. 539, 269--280 (1991; Zbl 0781.13017)] about the length of \textit{Dicksonian sequences} in \({\mathbb N}^n\).