What a frame's sphere-of-influence tests already know, so most ticks can skip them.
For each child of the frame, and for the frame's own boundary in the last
slot, the gap between the entity and the sphere as of at, measured from
origin. A gap is consumed by how far the entity has since moved from the
origin plus how far the child could have moved toward it — its periapsis
speed times the elapsed time — and while what is left is positive the
triangle inequality says the entity cannot be inside. That is a subtraction
and a multiply per child in place of a Kepler solve and a pose composition,
and it is what makes sixty-six children cost nothing on a tick that is not
near any of them.
Derived, never saved: a world that has none makes every test on its first tick and gets the answers the bounds were standing in for.
simulation/src/flight.ts:132Members
at
framereadonly
gapsreadonly
One per child, then the parent boundary. Mutated in place.