Numerical problems at "last bounce"

• Using only s ≤0:
  • event E1
    • (when s <= 0) triggers at s = -1e-10
    • v jumps to a positive value
    • new event → integration restarts → s rises again
  • event E2
    • jump of v is smaller than solver tolerance 1e-4
    • → no new event → fall through
• Using s ≤ 0 and v ≤ 0:
  • event E2
    • s falls through as before
  • event E3
    • v gets negative → flying is false → counterforce
    • v stays fixed at tiny negative value
    • s crawls downwards very slowly