English

On domains of elliptic operators with distributional coefficients

Analysis of PDEs 2025-09-30 v1 Functional Analysis Probability

Abstract

In this note we show how one can use recently gained insights from the study of singular SPDEs, more particularly the study of singular operators via the theory of Paracontrolled Distributions, to construct domains for (singular) elliptic operators. Formally we consider A(u)"=(1Δ)u+Vu+ξu+div(ρu), A (u) \text{"$=$''} (1 - \Delta) u + \nabla V \cdot \nabla u + \xi u + {{div} (\rho u)}, where VCδV \in \mathcal{C}^{\delta}, ξC2+δ\xi \in \mathcal{C}^{- 2 + \delta}, ρC1+δ,divρ=0\rho \in \mathcal{C}^{- 1 + \delta},{div}\rho = 0 and which satisfy a structural assumption that is notably satisfied when ξ\xi is a "sub-critical noise". We also show that under this assumption, one can construct a continuous change of variables Θ\Theta which satisfies AΘ(1Δ)L(H2;Hδ) A \Theta - (1 - \Delta) \in \mathcal{L} (H^2 ; H^{\delta'}) which allows us to define AA rigorously and parametrise a domain. Moreover, for suitably regularised operators Aε(u):=(1Δ)u+Vεu+(ξε+cε)u+div(ρεu), A_{\varepsilon} (u) := (1 - \Delta) u + \nabla V_{\varepsilon} \cdot \nabla u + (\xi_{\varepsilon} + c_{\varepsilon}) u + {{div} (\rho_{\varepsilon} \cdot u)}, we show that for a strongly converging regularised change of variables ΘεΘ\Theta_{\varepsilon} \rightarrow \Theta we have AεΘεAΘ in L(H2;L2) A_{\varepsilon} \Theta_{\varepsilon} \rightarrow A \Theta \text{ in } \mathcal{L} (H^2 ; L^2) which in particular implies norm resolvent convergence to a limiting closed operator.

Keywords

Cite

@article{arxiv.2509.24950,
  title  = {On domains of elliptic operators with distributional coefficients},
  author = {Immanuel Zachhuber},
  journal= {arXiv preprint arXiv:2509.24950},
  year   = {2025}
}

Comments

15 pages