motivation

  1. https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.126.148001: Vicsek model with metric distance has a discontinuous onset to order that is density-dependent and makes inhomogeneous polar bands. Vicsek model with topological (KNN or Voronoi) distance has a continuous onset to order that’s density independent.
  2. https://arxiv.org/abs/2208.06597: Parameter space exploration shows that there is the continuous transition as a linear function of the Peclet (diffusion/transport factor) and the alignment strength. Additionally, there are already inhomogeneous bands, although they are more unstable, when you increase the alignment strength past a clustering transition.
  3. https://www.nature.com/articles/s42005-021-00708-y: But, if we remove homogeneous media assumption — to be more realistic in the wild— topological VM also makes inhomogeneous bands.
  4. https://www.nature.com/articles/nature03236: What if we remove the closed social information assumption — to be more realistic in the wild? What kinds of phase transitions, including long-range order, response to fluctuation and phase separations, come up?
  5. https://royalsocietypublishing.org/doi/10.1098/rstb.2018.0378: One hypothesis from the static network limit of analysis is that collective decision accuracy is concave to the size of groups.
    1. I have yet to coin a specific in the wild case that demonstrates topological/metric and informed individual/homogeneous. It is probably gonna be a species that has two very distinct collective motion tasks, such as foraging vs migrating.
  6. We investigate the problem of (4) & (5) through the perception/interaction mechanism choices of (1)-(3) to expand the relationship space.

approach

  • exp 1-2: transferred and organized simulation sets for metric Vicsek Model varied clustering, KNN VM varied clustering. → to plot prettier.
    • maybe vary system size, fluctuation size, Pe value to get the dots fit by curve showing scale-invariance plots from the PRL papers?
  • exp 2: animated metric VM swarm with progressive injection. maybe separately plot informed vs uninformed individuals to check cluster membership and polarization speed?
  • exp 3: debugged information reversal. debug workflow = container py, container bash, test job, batch job, transfer, analysis pipeline.
  • exp 4: for theoretical analysis, still have to numerically implement the Fokker-Planck.