Supplement - Modelling Populations of Path-observed Networks - ISBA World Meeting 2022

George Bolt - g.bolt@lancaster.ac.uk - https://gmbolt.github.io

• Animation shows multiset of paths corresponding to marker on traceplots (big gaps separate distinct paths);
• Green entries are common with mode, whilst magenta entries highlight errors;
• Simulation parameters
  • $\mathcal{E}^m$ = {(1,2,1), (1,3,2,1,4), (1,2,4), (1,5,1), (1,5), (2,1)} (mode)
  • $\gamma=3.5$ (dispersion)
  • $n=25$ (sample size)
  • $d_E(\cdot, \cdot)$ taken as matching distance (defined on posterior), with distance between paths based upon common subsequences
• Initialised MCMC chain at $\mathcal{E}_0 = \{(10,10)\}$ and $\gamma_0=3.0$.

Data

• Below is the $n=25$ observations $\{\mathcal{E}^{(i)}\}_{i=1}^n$ used in this simulated example, drawn from an $\text{SIM}(\mathcal{E}^m, \gamma)$ model (approximately via MCMC algorithm);
• Green entries are common with the mode $\mathcal{E}^m$, whilst pink entries highlight errors.