“…For the second direction (in generalizing the Laplacian to some other operator L) we cite the body of work in [13,12,14,15,16,19,25,26,27]. The starting point here is to replace the semigroup e −t 2 ∆ in (i) and (ii) by some other semigroup e −t 2 L , but one can define an adaptation of (iii) by encoding the cancellation of atoms using L in a certain way (see [25] and also Definition 2.1 below).…”