Fig. 8

The performance of the algorithms LS, LS+ and random+DMP on the DCS model with the Tubingen dynamics if the metric is the probability of finding the source (solid curves) or the first symptomatic patient (dashed curves), together with the theoretical results (dash-dotted lines), as shown in Fig. 7. The simulations were computed on a population of \(n=9054\) individuals, and each datapoint is the average of 2400 independent realizations for the LS/LS+/LSv2/LS+v2 algorithms, and 48 independent realizations for the random+DMP algorithm. The default population and infection parameters were selected to match the population and COVID-19 infection datasets of Tubingen. As in all of the experiments, the algorithms were allowed to test 1% of the population (in this case 90 individuals) each day, and the algorithms finished (successfully or not) after 3–8 days after the first hospitalization occured