Figure 3: Effect of the length of time series and network size.
From: Reconstructing propagation networks with natural diversity and identifying hidden sources
(a,b) Success rate as a function of the relative length nt of time series for SIS (a) and CP (b) in combination with Newman–Watts (NW) and Barabási–Albert (BA) networks. The dashed lines represent 95% success rate. (c,d) the minimum relative length 
that assures at least 95% success rate as a function of the network size N for SIS (c) and CP (d) dynamics on NW and BA networks. Here the success rate is the geometric average over SREL and SRNC. For SIS dynamics, λε(0.1, 0.3) and δε(0.2, 0.4). For CP dynamics, λε(0.7, 0.9) and δε(0.3, 0.5). In a and b, the network size N is 500. The other parameters are the same as in Fig. 2. Note that nt≡t/N, where t is the absolute length of time series, and 
, where tmin is the minimum absolute length of time series required for at least 95% success rate.
