Journal of the London Mathematical Society · 2006 · 30 citations · 9 references
Natural SciencesMathematicsOptimizationProbability TheoryRepresentation TheoryFunctional AnalysisIntegral EquationsOperator TheoryOperator Inequalities
We give necessary and sufficient conditions in order that inequalities of the type ‖ T K f ‖ L q ( d μ ) ⩽ C ‖ f ‖ L p ( d σ ) , f ∈ L p ( d σ ) , hold for a class of integral operators TK f(x) = ∫Rn K(x, y) f(y) d σ(y) with nonnegative kernels, and measures d μ and d σ on ℝn, in the case where p > q > 0 and p > 1. An important model is provided by the dyadic integral operator with kernel K𝔇(x, y) = ∑Q∈𝔇 K(Q)χQ(x) χQ(y), where 𝔇 = {Q} is the family of all dyadic cubes in ℝn, and K(Q) are arbitrary nonnegative constants associated with Q ∈ 𝔇. The corresponding continuous versions are deduced from their dyadic counterparts. In particular, we show that, for the convolution operator Tk f = k ⋆ f with positive radially decreasing kernel k(|x − y|), the trace inequality ‖ T K f ‖ L q ( d μ ) ⩽ C ‖ f ‖ L p ( d x ) , f ∈ L p ( d x ) , holds if and only if 𝒲k[μ] ∈ Ls (dμ), where s = q(p − 1)/(p − q). Here 𝒲k[μ] is a nonlinear Wolff potential defined by W k [ μ ] ( x ) = ∫ 0 + ∞ k ( r ) k *¯ ( r ) / ( p − 1 ) μ ( B ( x , r ) ) 1 / ( p − 1 ) r n − 1 d r , and k *¯ ( r ) = ( 1 / r n ) ∫ 0 r k ( t ) t n − 1 d t . Analogous inequalities for 1 ⩽ q < p were characterized earlier by the authors using a different method which is not applicable when q < 1.
9