Fig. 3: Evolution of cooperation on temporal networks generated from synthetic data.

Here we generate \(M\) sparse snapshots based on: a a base Erdős-Rényi (ER) random network⁵⁵; and b a base scale-free (SF) network with degree exponent \(2.5\) constructed by the static model⁷⁰, choosing a fraction \(p\) of edges to be active within each snapshot. Note that when \(p\) is bigger, there are more links being active in each snapshot, which reduces the gap of the results obtained from temporal and static networks, while large \(g\) does not necessarily reduce the gap (Supplementary Fig. 9). Here \(M = 100\), the network size \(N = 1000\), and average degree \(\langle k\rangle = 10\). The robustness of the corresponding results for other parameters and other methods of generating synthetic temporal networks has been verified (see Supplementary Fig. 7).