Bernstein's inequality and the resolution of spaces of analytic functions (Q1913577)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Bernstein's inequality and the resolution of spaces of analytic functions
scientific article

    Statements

    Bernstein's inequality and the resolution of spaces of analytic functions (English)
    0 references
    0 references
    0 references
    19 January 1997
    0 references
    In Ann. Inst. Fourier 44, No. 4, 1091-1179 (1994; Zbl 0811.14046) the authors proved an extension theorem with bounds for polynomial functions in algebraic subvarieties of \(\mathbb{R}^n\) to rational functions in a neighborhood of the variety. Correspondingly, one obtains a priori bounds not just on the values of the extensions but also on their derivatives. For instance, one wants \[ \sup_{B (x, \rho)} |DF |\leq {C \over \rho} \sup_{B (x, \rho)} |F |\tag{*} \] where \(B(x, \rho)\) is the open ball of center \(x\) and radius \(\rho\) in \(\mathbb{R}^n\). When \(F\) is a polynomial in \(\mathbb{R}^n\), \(C\) depends only on the degree of \(F\), as shown by a classical theorem of S. Bernstein. Moreover, the algebraicity of the variety is tied to the existence of such bounds [\textit{L. Bos}, \textit{N. Levenberg}, \textit{P. Milman} and \textit{B. A. Taylor}, Indiana Univ. Math. J. 44, No. 1, 115-138 (1995; Zbl 0824.41015)]. The difficulty consists in obtaining some control of the constant \(C\) when \(F\) is only locally defined. In this paper the authors prove the following: Theorem. Let \(F_{1, \lambda}, \dots, F_{N, \lambda}\) be holomorphic functions in the complex ball \(B_c (0,1 + \varepsilon)\) of \(\mathbb{C}^n\), \(\varepsilon > 0\), depending real-analytically on \(\lambda \in U\), \(U\) open subset of \(\mathbb{R}^m\). Let \(V_\lambda = \text{span} \{F_{i, \lambda} : 1 \leq i \leq N\}\). Then, for any compact \(K \subseteq U\) there is a constant \(C\) such that for any \(\lambda \in K\), \(f \in V_\lambda\) and \(B(x, \rho) \subseteq B (0,1)\), the estimate (*) holds. The key technical lemma is a statement that appears naturally when one tries to relate the concept of multiplicity in the algebraic sense to that of counting roots of a system of equations [e.g. \textit{J. W. Milnor}, `Singular points of complex hypersurfaces' (1968; Zbl 0184.48405) and \textit{E. Bierstone} and \textit{P. Milman}, Publ. Math., Inst. Hautes Etud. Sci. 67, 5-42 (1988; Zbl 0674.32002)], namely; given a family \(F_{i, \lambda}\) as above, find a similar family \(G_{j, \lambda}\), \(1 \leq j \leq M\), such that (a) for every \(\lambda \in U\), \(\dim \text{span} \{G_{j, \lambda}\} = M\); (b) \(\text{span} \{F_{i, \lambda} \subseteq \text{span} \{G_{j, \lambda}\} \forall \lambda\;in U\); (c) there is a nowhere dense analytic subset \(A\) of \(U\) such that for all \(\lambda \in U \backslash A\), \(\text{span} \{F_{i, \lambda}\} = \text{span} \{G_{j, \lambda}\}\).
    0 references
    0 references
    resolution
    0 references
    spaces of analytic functions
    0 references
    Bernstein's inequality
    0 references
    0 references