Fig. 4: Persistent noise-induced asymmetry under optimal NPI initiation and relaxation.
From: Asymmetric limits on timely interventions from noisy epidemic data
a–f Repeat the simulations and analyses of Fig. 3 but under Ebola virus transmission and surveillance noise parameters (see “Methods” for details). a and c Provide the optimal control performance for initiating a lockdown as epidemics grow under no noise and practical surveillance respectively. e Shows the histograms of lockdown start, end, and duration times for the perfect (shaded) and noisy (unshaded) scenarios. b and d Present the optimal MPC performance for lifting a lockdown as epidemics wane under no noise and practical surveillance, respectively. f Plots corresponding histograms of timing and duration performance. We find the asymmetry in Fig. 3 therefore, persists.
