Stop 5 of 9 · What does a full region look like up close, and is the torus curved?

The dark foam — and why the torus is flat

A region filled to the top is not quiet. Look at one of its walls: every step is open half of the time and flips as much as a bit can, yet the average over time is perfectly flat. Curvature, read the way Bochner did, tests a landscape on the space: on T³ it never bends, the torus is flat, while the 16-cell of the fibred box has curvature 4.

Right now, and on average

frames averaged
0
largest slope in the average
–
flip per square (max ¼)
–
how full
–

Left: the wall now (orange = open). Right: the average of each square so far, from white (never open) to orange (always open). In the full state the right side settles to one even colour; with a slope it keeps a gradient, and the region is not full.

The torus is flat

Curvature, read the way Bochner did: put a landscape on the space and look, at every point, at how its steepness changes around that point. On a sphere the reading is always at least some positive amount K times the steepness; on a saddle it goes negative somewhere. On T³ Lean proves the reading is never negative, for any landscape, and a straight ramp gives exactly zero: no K > 0 works. The torus is flat. The closed space with a fourth, fibred axis (the 16-cell) reads K = 4 instead.

lowest reading on the slice
–
reading at the centre
–
steepness at the centre
–
best K for this landscape
–

Left: the landscape on one slice of T³ (it does not change along z). Right: Bochner's reading Γ₂, white = 0, blue = large. On a closed axis the ramp must jump back somewhere: the jump sits on the left and right edges, which are the same seam. Away from it the reading is exactly 0.

What is still open: the conjecture that curvature appears where the field moves away from the even, full state. It is not in Lean yet. Back to bits on the boundary · holography.

What is exact here

Theorem trail
StatementName
open with probability 1/2, variance 1/4TorusDarkFoam.openProb_uniform, TorusDarkFoam.bitVariance_uniform
the mean field is half the open fieldTorusDarkFoam.meanField_uniform
Bochner on T³; flatTorusBochner.gam2_eq, TorusBochner.torus_curvature, TorusBochner.not_curvature_pos
16-cell K = 4; path CD(0, 2)BochnerCurvature.cell16_curvature, BakryEmery.path_CD02

← Stop 4The pathStop 6 →