L^p-improving properties of averages on polynomial curves and related integral estimates

Abstract: In the combinatorial method proving of

L^{p}

-improving estimates for averages along curves pioneered by Christ (IMRN, 1998), it is desirable to estimate the average modulus (with respect to some uniform measure on a set) of a polynomial-like function from below using only the value of the function or its derivatives at some prescribed point. In this paper, it is shown that there is always a relatively large set of points (independent of the particular function to be integrated) for which such estimates are possible. Inequalities of this type are then applied to extend the results of Tao and Wright (JAMS, 2003) to obtain endpoint restricted weak-type estimates for averages over curves given by polynomials.

Review text

The optimal range for weak-type \(L^p\)-improving properties of averaging operators along curves is obtained, including the complete range just missed in the earlier work of \textit{T. Tao} and \textit{J. Wright} [J. Am. Math. Soc. 16, No. 3, 605--638 (2003; Zbl 1080.42007)]. One starts with an open ball \(U\subset \mathbb{R}^{d+1}\) and projections \(\pi_i:U\to\mathbb{R}^d\), \(i=1,2\) such that \(d\pi_i\) are surjective at every point. One defines \(R\) via \(\int_{\mathbb{R}^d} R f(y)\, g(y)\, dy=\int_U f(\pi_1(x))g(\pi_2(x))\psi(x)dx\) with \(\psi\) a suitable cutoff supported in \(U\). Let \(X_i\), \(i=1,2\) be nonvanishing vector fields on \(U\) such that \(d\pi_i(X_i)=0\) and suppose the \(k\)-fold commutator \(X_\epsilon=[X_{\epsilon_1}[X_{\epsilon_2}[\cdots[X_{\epsilon_{k-1}} X_{\epsilon_k}]\cdots]]]\) vanishes on \(U\) for large enough \(k\) and all \(\epsilon=(\epsilon_1,\dots,\epsilon_k)\in\{0,1\}^k\). Then \(R\) and \(R^\ast\) are given by averages over polynomial curves. The main theorem is stated as follows. Suppose that \(X_1,\, X_2,\dots \) are as above with \(X_{j+2}=X_j\), \(x_0\in U\) and \(\Phi_{x_0} (t_1,\dots, t_{d+1})=\exp (t_1X_1)\circ \cdots \circ \exp (t_{d+1} X_{d+1})(x_0)\). Let \(J_{x_0}(t)\) be the Jacobian determinant of \(\Phi_{x_0}\). If \(\partial_t^\alpha J_{x_0}(t)\neq 0\) at \(t=0\) for some multi-index \(\alpha\) then \(R\) satisfies the restricted weak-type estimate \[ \Bigl|\int_U \chi_F (\pi_1(x))\chi_G(\pi_2(x)) \psi(x)\, dx\Bigr| \leq C |F|^{1/{p_1}}|G|^{1/{p_2}} \] when \(\psi\) is supported in a small enough neighborhood of \(x_0\) and \(p_1=(A_1+A_2-1)/A_1\) and \(p_2=(A_1+A_2-1)/A_2\), where \[ A_1=\bigl\lceil{d+1\over 2}\bigr\rceil+\alpha_1+\cdots+\alpha_{2\lceil{d+1\over 2}\rceil-1};\quad A_2=\bigl\lceil{d+1\over 2}\bigr\rceil+\alpha_2+\cdots+\alpha_{2\lceil{d+1\over 2}\rceil}. \] An important ingredient in the proof is a uniform estimate for polynomials stated as follows: If \(K\subset\mathbb{R}\) is measurable then for any \(n=1,2,\dots\) and \(\epsilon\in (0,1)\) there is an interval \(I\), \(|K\cap I|\geq (1-\epsilon)|K|/n\) and \(c_{n,\epsilon}\) such that \[ \int_K |p(t)|\, dt\geq c_{n,\epsilon} |K|^{j+1}\sup_{t\in I}|p^{j}(t)| \] for any polynomial \(p\) of degree at most \(n\) and each \(k=0,\dots, n\). A general version of this result is established involving averages with respect to probability measures and functions of {\textit{polynomial type}}. The combinatorial proof reduces to the idea of estimating higher derivatives via sampling at a finite number of points. Extension from one dimension to higher dimensions involves counting solutions of iterated flows in which the vector fields must be lifted to a nilpotent Lie group in order to produce a setting in which flows correspond to polynomial mappings.

Recommendations

Cited in

(23)

This page was built for publication: \(L^p\)-improving properties of averages on polynomial curves and related integral estimates

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q971493)