Unified Treatment of Dynamical Processes on Generalized Networks: Higher-Order, Multilayer, and Temporal Interactions

The paper explicitly shows that for nonintertwined clusters the generalized Laplacians L(k), together with the cluster indicator matrices D(m), suffice for the stability analysis and for obtaining a finest simultaneous block diagonalization (SBD) that separates parallel and transverse perturbations (see Eq. (4) and the SBD algorithm steps; proofs and details are given in the main text and Supplemental Material) . The paper also provides a proof sketch of the optimality of the discovered block structure (with probability 1) and demonstrates the automatic separation between the manifold-parallel and transverse modes under SBD . The candidate solution matches these high-level conclusions and the SBD workflow, but it makes two key incorrect claims: (i) that the coupling matrices C(k) and degree matrices preserve the cluster-constant subspace S (false in general unless intracluster degrees are uniform), and (ii) that alg{D(m), C(k)} = alg{D(m), L(k)} because Ddeg lies in span{D(m)} (generally false). The paper does not rely on these claims; instead it derives a representation of the variational equation directly in terms of D(m) and L(k) for nonintertwined clusters (Supplemental Eqs. S1–S5) , and then applies SBD to that set .