Order parameter M(t), last 10 simulated minutes (ring: M = mean sense of rotation; 2-D: M = |P|)
Histogram of M (bimodal ⇒ the group switches between two directions)
Heading rose (all insects now)
Runs: mean |M| versus density ρ (dots), fitted transition
Phase diagram: order above the random-sign floor, noise η versus density ρ (run it on the Experiments tab)
Direction-flip dwell times: survival S(t) on a log axis vs the two-state Markov (exponential) fit
Batch experiments run in a Web Worker using the same model code. Settings η, r, paused-fraction are taken from the sliders (defaults reproduce the quoted numbers).
The science
Equations used
Gregarious vs solitarious phase
Flying swarms, wind, and swarm vs flock
What is simplified (honest limits)
How this differs from other locust / Vicsek labs here