In a recent letter [1] a model of attractive spinless fermions on the honeycomb lattice at half filling has been studied by mean-field theory, where distinct homogenous phases at rather large attraction strength V > 3.36, separated by (topological) phase transitions, have been predicted. In this comment we argue that without additional interactions the ground states in these phases are not stable against phase separation. We determine the onset of phase separation at half filling V ps ≈ 1.7 by means of infinite projected entangled-pair states (iPEPS) [2][3][4] and exact diagonalization (ED).The Hamiltonian of the model readŝ(1) with t = 1 the hopping amplitude, and V > 0 the attraction strength. We work in the grand-canonical ensemble, i.e. we use a chemical potential µ to control the particle density n(µ) in the system. Setting µ = 0 corresponds to a half-filled state, n = 0.5, if the state at half filling is stable towards phase separation.Intuitively, at half filling, if the attraction V is much stronger than the hopping t, the fermions can minimize their energy by clustering, leading to phase separation, where half of the system is empty and the other half is occupied by the fermions. In the grand-canonical ensemble, such an instability can be identified as a discontinuity (a jump) in the particle density n(µ) at µ = 0.Figure 1(a) summarizes our numerical results obtained with ED on finite systems and with iPEPS, a tensor network ansatz to simulate the model directly in the thermodynamic limit. The accuracy of the iPEPS can be systematically controlled by the so-called bond dimension D. Details on the method can be found in refs. [4,5].Figure 1(a) shows an example of a jump in n(µ) for V = 1.72 between the two densities n 1 ≈ 0.11 and n 2 ≈ 0.89 obtained with iPEPS. For densities in between these two values there is no stable homogenous solution, because it is energetically favorable for the system to split into two regions, one with density n 1 and the other one with n 2 . Since iPEPS is an ansatz for a homogeneous phase, we either obtain a state with density n 1 or a state with density n 2 for µ = 0, if there is no homogenous solution at half filling. For very large attraction, V 1.9, the system splits into a completely empty and a completely filled region, i.e. n 1 = 0, n 2 = 1. The dependence of n 1 as a function of V is shown in fig. 1(c).The full symbols in fig. 1(b) for V < 1.71 show the iPEPS energy of stable solutions at half filling for µ = 0, whereas the open symbols for V > 1.71 correspond to states away from half-filling, where the state at half-filling arXiv:1204.4791v1 [cond-mat.str-el]