Uniqueness of diffusion operators and capacity estimates
Analysis of PDEs
2014-01-03 v1
Abstract
Let Ω be a connected open subset of \Rid. We analyze L1-uniqueness of real second-order partial differential operators H=−∑k,l=1d∂kckl∂l and K=H+∑k=1dck∂k+c0 on Ω where ckl=clk∈Wloc1,∞(Ω),ck∈L∞,loc(Ω), c0∈L2,loc(Ω) and C(x)=(ckl(x))>0 for all x∈Ω. Boundedness properties of the coefficients are expressed indirectly in terms of the balls B(r) associated with the Riemannian metric C−1 and their Lebesgue measure ∣B(r)∣. \noindent\hspace{10mm}First we establish that if the balls B(r) are bounded, the T\"acklind condition ∫R∞drr(log∣B(r)∣)−1=∞ is satisfied for all large R and H is Markov unique then H is L1-unique. If, in addition, C(x)≥κ(cT⊗c)(x) for some κ>0 and almost all x∈Ω, \divvc∈L∞,loc(Ω) is upper semi-bounded and c0 is lower semi-bounded then K is also L1-unique. \noindent\hspace{10mm}Secondly, if the ckl extend continuously to functions which are locally bounded on ∂Ω and if the balls B(r) are bounded we characterize Markov uniqueness of H in terms of local capacity estimates and boundary capacity estimates. For example, H is Markov unique if and only if for each bounded subset A of Ω there exist ηn∈Cc∞(Ω) satisfying limn→∞∥\oneAΓ(ηn)∥1=0, where Γ(ηn)=∑k,l=1dckl(∂kηn)(∂lηn), and limn→∞∥\oneA(\oneΩ−ηn)φ∥2=0 for each φ∈L2(Ω) or if and only if \capp(∂Ω)=0.
Cite
@article{arxiv.1311.5281,
title = {Uniqueness of diffusion operators and capacity estimates},
author = {Derek W Robinson},
journal= {arXiv preprint arXiv:1311.5281},
year = {2014}
}