2007/05/07 by Hofmann, S.
#Classical Analysis and ODEs (math.CA) #FOS: Mathematics
paper · doi:10.48550/arxiv.0705.0839
We consider divergence form elliptic operators L=-\dv A(x)∇, defined in ℝn+1=\(x,t)∈ℝn×ℝ\, n ≥ 2, where the L∞ coefficient matrix A is (n+1)×(n+1), uniformly elliptic, complex and t-independent. Using recently obtained results concerning the boundedness and invertibility of layer potentials associated to such operators, we show that if Lu=0 in ℝn+1+, then for any vector-valued \bf v ∈ W1,2loc, we have the bilinear estimate |\iintℝn+1+ ∇ u ⋅ \bf v dx dt |≤ Csuptgt;0 ‖u(⋅,t)‖L2(ℝn)(‖|t ∇ \bf v‖| + ‖N_*\bf v‖L2(ℝn)), where ‖|F‖| ≡ (\iintℝn+1+ |F(x,t)|2 t-1 dx dt)1/2, and where N_* is the usual non-tangential maximal operator. The result is new even in the case of real symmetric coefficients, and generalizes the analogous result of Dahlberg for harmonic functions on Lipschitz graph domains.