Majority vote model with ancillary noise in complex networks

We analyze the properties of the majority-vote (MV) model with an additional noise in which a local spin can be changed independently of its neighborhood. In the standard MV, one of the simplest nonequilibrium systems exhibiting an order-disorder phase transition, spins are aligned with their local majority with probability $1-f$, and with complementary probability $f$, the majority rule is not followed. In the noisy MV (NMV), a random spin flip is succeeded with probability $p$ (with complementary $1-p$ the usual MV rule is accomplished). Such extra ingredient was considered by Vieira and Crokidakis [Physica A {\bf 450}, 30 (2016)] for the square lattice. Here, we generalize the NMV for arbitrary networks, including homogeneous [random regular (RR) and Erd\"os Renyi (ER)] and heterogeneous [Barabasi-Albert (BA)] structures, through mean-field calculations and numerical simulations. Results coming from both approaches are in excellent agreement with each other, revealing that the presence of additional noise does not affect the classification of phase transition, which remains continuous irrespective of the network degree and its distribution. The critical point and the threshold probability $p_t$ marking the disappearance of the ordered phase depend on the node distribution and increase with the connectivity $k$. The critical behavior, investigated numerically, exhibits a common set of critical exponents for RR and ER topologies, but different from BA and regular lattices. Finally, our results indicate that (in contrary to a previous proposition) there is no first-order transition in the NMV for large $k$.