Analogs of Wiener's ergodic theorems for semisimple Lie groups. II (Q1572888)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Analogs of Wiener's ergodic theorems for semisimple Lie groups. II
scientific article

    Statements

    Analogs of Wiener's ergodic theorems for semisimple Lie groups. II (English)
    0 references
    0 references
    0 references
    0 references
    30 January 2001
    0 references
    [Part I by \textit{A. Nevo} and \textit{E. M. Stein} in Ann. Math. (2) 145, 565-595 (1997; Zbl 0884.43004).] Let \((X,{\mathcal B},m)\) be a standard Borel space with a Borel measurable ergodic \(G\)-action preserving the probability measure \(m\). If \(\nu_t\), \(t\in \mathbb{R}_+\), is a one-parameter family of probability measures on \(G\) such that the map \(t\to\nu_t\) is continuous when the space of probability measures is equipped with the weak-\(*\)-topology, then \(\nu_t\) is a pointwise ergodic family in \(L^p(X)\) if, for every \(f\in L^p(X)\), \[ \lim_{t\to\infty}\pi(\nu_t)f(x)=\int_Xf dm, \] where \(\pi(\nu_t)f(x):=\int_G f(g^{-1}x) d\nu_t(g)\) and the convergence is pointwise almost everywhere and in the \(L^p\) norm. The authors prove that if \(G\) is a connected semisimple group with finite center and no compact factors and if \(\nu_t\) is the family of ball averages on \(G\) defined by \[ \nu_t={1\over{m_G(B_t)}}\int_{B_t}\delta_g dm_G(g), \] where \(m_G\) is the Haar measure on \(G\), \(B_t\) is the ball of radius \(t\) with respect to the \(G\)-invariant Riemannian metric on \(G/K\), and \(\delta_g\) is the Dirac measure at \(g\), then \(\nu_t\) is a pointwise ergodic family in \(L^p(X)\) for every ergodic \(G\)-space \((X,{\mathcal B},m)\) as above. Moreover, again for every ergodic \((X,{\mathcal B},m)\), the family of ball averages satisfies the strong maximal inequality in \(L^p(X)\), \(1<p\leq\infty\), namely, \[ \left\|\sup_{t\geq 0}|\pi(\nu_t)f|\right\|_p\leq C_p\|f\|_p. \] The authors also show that for semisimple Lie groups satisfying property T, the convergence of the ball averages to the ergodic mean is at an exponential rate, namely there exists a positive constant \(b(G)\) such that for every \(\theta<b(G)\) and for all \(r\) and \(u\) such that \(0<u<1\) and \(2p=2r+pr\), then for almost every \(x\in X\) \[ \left|\pi(\nu_t)f(x)-\int_Gf dm\right|\leq B(x,f)\exp\left(-{{u\theta t}\over{4}}\right), \] where \(\|B(\cdot,f)\|_r\leq B\|f\|_p\) for all \(f\in L^p(X)\). The authors also generalize the above results to general radial averages and to actions with a spectral gap (where the group acting does not necessarily have property T).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    ergodic theorem
    0 references
    maximal inequality
    0 references
    Kazhdan property
    0 references
    semisimple Lie groups
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references