“…Note that since C 1 c (Ω) is dense in W 1,p 0 (Ω) with respect to W 1,p norm, the infimum in (1.4) can be taken over C 1 c (Ω). When Ω is an arbitrary (possibly unbounded) domain, following Berestycki et al [7,8,9,11,28], we define This type of eigenvalue was first introduced in a celebrated work of BerestyckiNirenberg-Varadhan [8] for second order operators in bounded (not necessarily smooth) domains, and then was developed to second order operators in unbounded domains [7,9,11]. An important feature of the notion of generalized principal eigenvalue is that if Ω is a smooth and bounded domain, λ(K V , Ω) coincides with the principal eigenvalue λ 1,V (Ω), while if Ω is unbounded λ(K V , Ω) is well defined and can be expressed by a variational formula.…”