Chapter 3 · Section 13
The Inspiral: Matter Spiraling In
The BPT law
BPT Part 3.10 (After the Outward Expansion: the Pulsing Decay): "The End of the Outward Expansion closes the census; it does not end the universe, and it does not end its wells' making. What follows is its decay." "Stars and remnants are flung out of their galaxies or spiral into the central wells as their orbits shed energy," and "the heaviest wells go last." Before that, in the same Part: "every bit returns through a singularity on its loop home, the End of the Outward Expansion is the last of those returns, and a universe ends when all its bits have returned" (quoted at matter.js:146–148). BPT Part 3.14 adds that "Ascendant Wells grow in the decay, when the wells take in all that is left." What the orbits shed goes to the space layer: BPT Part 3.13's Gravitational-Wave Power, P_GW = (32/5) · G⁴ · m₁² · m₂² · (m₁ + m₂) / (c⁵ · a⁵), "the power two orbiting masses give the space layer."
The established physics
A circular binary shrinks by gravitational radiation as d(a⁴)/dt = −(256/5)·G³·m·M·(m + M)/c⁵ (Peters 1964). The energy given up at the last orbit follows the Barausse, Morozova & Rezzolla 2012 fit (Part 3.10).
The game's math
Matter.inspiral(wells, dtau) matter.js:152, called each decay tick in the matter stage (stepDecay() app.js:2672), with dtau the tick's growing span of flips. For each body:
- Its well is the non-site well that pulls it hardest, M/d², with d at least 10⁻⁶.
- a⁴ ← a⁴ − (256/5)·G³·m·M·(m + M)/𝒞→⁵ · dtau.
- If a⁴ ≤ r_s⁴, the body is in. The well gains m − E, with E = gwFraction(m, M)·(m + M), and E is booked to that well's
gwByand then toradiate('gwIn'). - Otherwise the body is placed at a_new = (a⁴)^(1/4) on the same line, on a circular orbit of speed min(𝒞→, √(GM/a_new)), keeping its sense of turn.
A ripple is drawn for the largest shed if it passes 10⁻⁴ of the matter (app.js:2673).
The waves and the radiation, lightly (v8.6, 15 November 2026). The author: "what is happening to all the grav waves and radiation? we hid the activity but it seems like sometimes I can still see where they would have been, id like to add them lightly especially at the end of the universe as all the radiation is happening." The merger rings had been switched off on 14 November because, in the paced mode, a whole step's mergers fired their rings in the one frame after a hold; what could still be seen were the supernova, planetary-nebula and evaporation flashes, which never stopped. Through the afternoon of 15 November three drawings were tried in turn: the rings back, queued and dimmer; a shimmer and then thin crests on the tube driven by the ledger's gravitational-wave power, running to the inner equator; and a disc of contracting rings at the centre. The author: "the grav waves and radiation appears too thick and doesnt filter properly to the direct center like it should"; "the waves are not filtering to the direct center!!!!! that should be clearly going to the direct center"; and finally, "its the grav waves and radiation that needs to go straight to the center you split them up and now the radiation at the end looks weird." All three are gone (the ring code stays dormant behind RIPPLES_HIDDEN; the shader's uGWin/uRadIn terms are held at zero; the disc is hidden).
Every wave and every light goes straight to the centre (v8.6, 15 November 2026, evening). One drawing for all of it, which is Non-Resonant Return (BPT Part 3.13: what is physical, the wells' glow, starlight and gravitational waves, goes home through the wells to the Source) drawn as it reads. Every radiation event launches a pulse from its own place on the torus that runs in a straight line to the centre (radiatePulse(kind, r, c, k) scene3d.js; stepPulses() each frame):
- The events and their energies k (0–1): a merger in the life, k = min(1, 300 · shed / M_ref), where shed is the space layer's share by
gwFraction(app.js mergeWellsOnce); each well's inspiral in the decay's matter stage, k = min(1, 2,500 · E / M_ref), the forty largest a tick (was one ring for the largest); a supernova, k = 0.15, thirty a step at most, in the life and the decay; a well's glow as it evaporates, k = min(1, 300 · Δm / M_wells) from up to 24 wells a tick, and its last flash k = min(1, 1,000 · m₀ / M_wells). Gravitational waves are cyan (0.45, 0.95, 1.0), light warm (1.0, 0.62, 0.3). - The flight. A step's pulses leave spread over 0–500 ms so they stream rather than burst. Each runs from its surface point to the centre in 1.1–1.6 s along the straight line, its position (1 − s²) of the way out, so it falls faster as it nears the centre; it is drawn as a head point and a tail behind it (0.1 + 0.08 s² of the path, the tail at 0.35 of the head's light), brightening from 0.35 of k at launch to k at arrival.
- The glow at the point. Each arrival adds 0.12 k to a glow at the centre, cyan or warm by its kind, which fades with a half-life of 0.7 s; the glow is a sprite at the torus's centre, opacity min(1, glow), scale 0.25 + 0.6 √glow. The budget is 1,500 pulses in flight; what is owed beyond it arrives as glow after 1.5 s.
- The physics is unchanged: the ledger's radiation is what it was; only what is drawn from it is new.
The rings, not arrows (15 November 2026, late). The author, on seeing the pulses: "what the fuck are the arrows?? the gravitation waves were good as is they just didnt flow through the exact center you need to fix that now." The pulses are retired after an hour (radiatePulseRetired, kept for the record). A wave event is a ring again, exactly the afternoon's rings (queued, 0.28 of their light), and the fault they had is fixed: a ring grows to the far side of the tube, H/2 rows, and on the horn torus the far side of every point is the pinch at the centre, but it faded as (1 − x)^0.7 the whole way and so was gone as it arrived. Now it holds its light to 0.82 of its run, brightening by (0.7 + 0.5 x) as it closes on the centre, and goes out only over the last 0.18 as it passes through; its arrival feeds the glow at the centre (0.12 k). The event energies k (0–1) are the ones listed above, so a ring's light is 0.45 + 0.55 k. A light event (a supernova, a well's glow or last flash) feeds the warm glow at the centre directly (0.08 k), with no ring. radiatePulse(kind, r, c, k) keeps its name as the one door for every radiation event.
Through the exact centre (15 November 2026, night). The author: "it should go toward the centre as it does now but exactly pass through the center." The ring is unchanged in look and in how it grows. Two things changed. Its run is H/2 rows as before, or the merger's distance to the centre plus 0.1 H when that is farther (a merger on the outer edge), so every ring reaches past the centre, the pinch at row H/2. It keeps its full light until 0.12 of its run after it has crossed the centre (capped at 0.85), then fades as before; it used to fade the whole way and was gone as it arrived. And a twin of the ring, sharing its shape and drawn without the depth test, carries its light only within 0.08 H rows of the centre row, so the stretch that dips into the funnel is seen crossing the point through the tube's top; everywhere else only the ordinary ring is drawn.
Closing on the exact centre (15 November 2026, night, the fix that held). Photographed one ring, frozen at stages of its life (harness __pcRing, __pcRingX): it reached the centre's row but smeared down the funnel as a pale sheet, because the rows are spread round the tube by equal area and near the inner equator a few rows cover the whole funnel wall (d(row)/dv = (R + r cos v)/(2πR) → (R − r)/(2πR), nearly zero on a horn torus). That sheet was also the blotch the author saw flickering. Now each ring vertex is drawn by its angle round the tube: on the inner half it is pulled toward the torus's centre by w = max(0, −cos v)⁶, all the way at v = π, so the ring's inner arc closes onto the centre point and opens out past it; the twin drawn through the body shows only where w > 0.2 (full at 0.6). In the frames the ring's tip reaches the point at x ≈ 0.5 and its lines pass through it and fan out by 0.66–0.85.
Spokes into the point (16 November 2026). On the author's screen the lines still came "towards the center but completely missing the actual center". In live play the pull, w = max(0, −cos v)^6, reached only about half strength over most of the inner wall, so the points it caught partway drew loops round the matter core instead of meeting. The pull is now short and steep, w = smoothstep(0.6, 0.75, −cos v): each line keeps to the skin until the inner wall, then runs straight, as a spoke, into the single centre point and out the other side. Seen live from the pole and from the side, the spokes cross at the point. The smoke on the skin is unchanged.
Along the torus's own curve (16 November 2026, later). The author: the spokes "still arent following the exact curve of the torus". The straight pull toward the centre cut through the air. Now each ring vertex keeps its column and has its tube angle drawn toward π, v' = π + (v − π)(1 − w), with w = smoothstep(0.5, 0.85, −cos v), and is placed back on the surface at that angle. The line slides down the inner wall on the torus itself into the pinch. On the horn torus the pinch is the centre point, so the lines meet there. Seen live, they curve down the funnel and gather at the point.
No straight chords down the funnel (16 November 2026, later still). The author: "they are still taking straight lines from the edges of the mouth". The ring's 96 points were spaced evenly round the ring, but near the inner equator a step of a few rows spans the whole funnel wall, so a single segment ran straight from the mouth's edge to the centre. Each crest is now traced finely (8·n samples), and its n = 192 drawn points are spaced evenly by distance on the surface. Where it dives down the funnel it gets many points and stays a curve on the skin. Frozen rings, photographed at stages, curve down the wall into the pinch and out the other side.
Show Gravitational Waves (16 November 2026). The author approved the rings as they now run, and asked for a toggle, on by default. It sits beside Grid lines and Auto rotate. Off, it draws no merger rings (those on screen are cleared, and new ones are not drawn) and hides the first plane waves over the young torus. The radiation's glow and smoke on the skin stay, being radiation, not the waves.
Toned down (15 November 2026, night). The author: "even now as is there is too much coloring from the radiation." The decay's crest gain is 600 with a cap of 0.3 (was 2,500 and 0.6), the life's cap 0.25 (was 0.35), and the wells' warm glow gain 200 with a cap of 0.1 (was 1,000 and 0.4); the glow no longer tints the outer face at all, only rising toward the centre. The first light is unchanged.
Withdrawn (15 November 2026, last). The through-the-body rings read as "random circles of lightning" and the author asked for everything after the approved version to be reverted: "I simply wanted ... the gravitational waves going through the center." The radiation drawing is back to the approved version exactly (the queued rings at 0.28 of their light and the thin crests on the tube, pushRadiation() driving them; the centre disc off; the pulses, the centre glow and the through-the-body rings gone from the code). The decay's well merging and the star budgets, asked for the same evening, stand. What "through the centre" should look like is still unresolved; it is on the open-loops list and nothing more will be drawn until the author has described a single wave's path.
| Number | Where | What it is | Kind |
|---|---|---|---|
| 256/5 | matter.js:161 | Peters coefficient | PHYSICS (Peters 1964) |
| 5, 3, 4 (powers) | matter.js:155, 161 | c⁵, G³, a⁴ | PHYSICS (Peters 1964) |
| 1e-6 | matter.js:158, 160 | least distance | METHOD (guard) |
| 1 | matter.js:160 | least well mass | BPT (one Zinf quantum, Part 1.8) |
| 10⁻⁴ | app.js:2673 | least shed to draw a ripple (the rings, dormant) | DISPLAY |
| 300, 2,500, 0.15, 300, 1,000; 40, 30, 24 | app.js (mergers, decay inspiral, supernovae, evaporation) | the pulses' energy gains by event, and how many a step or tick | DISPLAY (the events and their energies are the ledger's) |
| 0–500 ms; 1.1–1.6 s; s²; 0.1 + 0.08 s²; 0.35→1 | scene3d.js stepPulses() |
a pulse's spread at launch, its flight time, its fall, its tail and its brightening | DISPLAY |
| 0.12; 0.7 s; 0.25 + 0.6 √g; 1,500 | scene3d.js stepPulses() |
the glow fed per arrival, its half-life, the sprite's size, the budget in flight | DISPLAY |