Abstract: Given a Banach space , for and we investigate the smallest constant for which every satisfy int_{{-1,1}^n}Bigg|sum_{j=1}^n partial_jf_j(varepsilon)Bigg|^pdmu(varepsilon) leq mathfrak{P}^pint_{{-1,1}^n}int_{{-1,1}^n}Bigg|sum_{j=1}^n d_jDelta f_j(varepsilon)Bigg|^pdmu(varepsilon) dmu(delta), where is the uniform probability measure on the discrete hypercube and and are the hypercube partial derivatives and the hypercube Laplacian, respectively. Denoting this constant by , we show that for every Banach space . This extends the classical Pisier inequality, which corresponds to the special case for some . We show that if either the dual is a Banach space, or for some we have , where is a Hilbert space and is an arbitrary Banach space. It follows that if is a Banach lattice of finite cotype.
Recommendations
Cited in
(11)- Polynomial inequalities on the Hamming cube
- Rademacher type and Enflo type coincide
- Bounds for partial derivatives: necessity of UMD and sharp constants
- Metric \(X_{p}\) inequalities
- Remarks on non linear type and Pisiers inequality
- On Pisier's inequality for UMD targets
- Modified Paouris Inequality
- Comparison of metric spectral gaps
- Tail spaces estimates on Hamming cube and Bernstein-Markov inequality
- Discrete logarithmic Sobolev inequalities in Banach spaces
- Dimensionality of Hamming metrics and Rademacher type
This page was built for publication: Pisier's inequality revisited
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4927469)