“…The scalar product fixes the norm that quantifies the notion of "big" or "small" perturbations ||δL||. On physical grounds, and following [37,45], we argue in favor of the system's energy as the most adequate norm to control the pseudospectra and assess spectral instability, because it conveniently encodes the size of the physical perturbation with respect to the nature of the problem (see [80] for a systematic discussion of this point). On the computational side, calculating the pseudospectrum requires the numerical evaluation of the norm of the resolvent, according to the previous definition.…”