Thinking in Waves Episode 50 / Part XVII — Looking for the Edge of the Axis
From Episode 1's single line to here
The Map —
Fifty episodes on, where did we arrive?
Episode 1's question was one line — "Differentiate a sound and it goes high, integrate it and it goes low. So what does the space in between — a fractional order — mean?"
★ Fifty episodes later: the order \(\alpha\) turned out to be the dB/oct, the phase, the dimension, the Hurst exponent, the AdS mass and the structure-function exponent. One and the same number.
★★ And honestly — no formula unknown to humanity came out of it.
Episodes 15 and 43 were scorecards; Episode 48 found the axis's edge; Episode 49 wrote the handbook.
★ This one is the map — where we walked, and what is left. And every number is recomputed here.
01One axis — everywhere an order appeared
| order \(\alpha\) | dB/oct | phase [deg] | Where it appeared |
|---|---|---|---|
| −1 | −6 | −90 | Ep. 30, \(d=1\) |
| −1/2 | −3 | −45 | ★ Ep. 30 \(d=2\); Ep. 38 \(d_s=2\) (UV gravity) |
| 0 | 0 | 0 | Ep. 30 \(d=3\); Ep. 39 BF bound; wherever a logarithm appears |
| 1/2 | 3 | 45 | ★ Ep. 2 1/f; Ep. 30 \(d=4\) |
| 5/6 | 5 | 75 | ★ Ep. 42 turbulence (K41) |
| 1 | 6 | 90 | Brownian motion; Ep. 30 \(d=5\) |
The point of §01 — the same number under different names
\(\text{dB/oct}=6\alpha\) (Ep. 1) \(\text{phase}=90\alpha\) degrees (Ep. 1)
dimension \(d=2\alpha+3\) (Ep. 30) Hurst \(H=\alpha-\frac12\) (Ep. 42)
AdS's \(\nu=\alpha/2\) (Ep. 39) spectral index \(\beta=2\alpha\)
★★ And the two most precisely measured "departures" also sit on this scale —
the universe's primordial fluctuations (Ep. 47): \(n_s=0.9649\) = \(-0.1053\) dB/oct
turbulent intermittency (Ep. 42): \(+0.0812\) dB/oct from K41
★ A coincidence, but the same order of magnitude.
02Five reasons the order leaves the integers; three ways the axis ends
| Reason | Episode | Formula |
|---|---|---|
| ① superposition of many time constants | Ep. 2 | \(g(\tau)\propto1/\tau\) |
| ② dimension | Ep. 30 | \(\alpha=(d-3)/2\) |
| ③ continuous variation at criticality | Ep. 13 | \(s=\tfrac12\pm\sqrt{\tfrac14-\lambda}\) |
| ④ the bulk mass | Ep. 39 | \(\alpha=2\sqrt{\tfrac{d^2}4+m^2L^2}\) |
| ⑤ the order is not a single number | Ep. 42 | width \(=\log_2\frac{1-p}{p}\) |
| How the axis ends | Episode | What is happening |
|---|---|---|
| ① a quantity that does not fit | Ep. 41 | integers (winding, Chern) |
| ② the axis is not single | Ep. 42 | the order becomes a distribution |
| ③ no power law at all | Ep. 48 | divergence, logarithmic approach, finite size |
03★ Recomputing every number
Rather than trust the published numbers, we recompute them by another route and compare (extending what Episode 43 did, now up to Episode 49).
| Episode | Item | Recomputed | As published |
|---|---|---|---|
| Ep. 1 | \(\text{dB/oct}=6\alpha\); \(\text{phase}=90\alpha\) | 3 / 45 | same |
| Ep. 33 | Koide's \(K\) | 0.666661 | 2/3 |
| Ep. 37 | 3D return probability | 0.338718 | 0.340537 |
| Ep. 38 | Planck length [\(10^{-35}\) m] | 1.616255 | 1.6162 |
| Ep. 39 | ratio in \(z\), \(e^{\pi/|\nu|}\) | 535.491656 | 535.4917 |
| Ep. 42 | dB/oct difference from K41 | 0.081164 | 0.081164 |
| Ep. 44 | the f-sum rule | 1.570771 | \(\pi/2=1.570796\) |
| Ep. 45 | \(\sum ne^{-\delta n}-1/\delta^2\) | −0.083333 | \(-1/12\) |
| Ep. 45 | Casimir pressure at 10 nm [\(10^5\) Pa] | 1.300126 | 1.3001 |
| Ep. 46 | solar-mass \(S/k_B\) [\(10^{77}\)] | 1.049430 | 1.0494 |
| Ep. 46 | time to leave the measurable band [ms] | 1.467402 | 1.4662 |
| Ep. 47 | \(n_s\) in dB/oct | −0.105300 | −0.10530 |
| Ep. 48 | marginal coupling's order at \(10^{32}\) | −0.012984 | −0.012984 |
| Ep. 49 | Bode phase [deg] | −44.996840 | −45 |
The point of §03
28 items; zero FAILs.
★ But what this guarantees reaches only as far as "today's self does not contradict yesterday's" (Episode 43's exercise). What guards against contradicting nature is the list of failed predictions.
★ Only the Watson integral is off, by 0.5 % — because the singularity at the origin is being sampled on a \(200^3\) grid.
04Counting the series itself
| Item | Count |
|---|---|
| episodes published (Japanese) | 49 |
| body files (Japanese / English) | 49 / 49 |
| ★ verification scripts | 48 |
| ★ lines of that code | 13,966 |
| bytes of that code | 8.18e+5 |
| characters of Japanese body text (approx.) | 6.95e+5 |
| ★ dependencies | ★ zero |
The point of §04
★ Everything runs on bare Node as it stands. The FFT, the Jacobi eigensolver, the complex gamma function — each written where it was needed.
★★ This is policy, not taste — "the calculation can be reproduced" is very nearly the only guarantee this series has.
★ The figures in this section were obtained by counting files as this article was being written — adding Episode 50 increases each by one.
05★ What has not been touched
| Left undone | Declared in | What it is |
|---|---|---|
| handling noisy real data | Ep. 49 | decade estimates acquire statistical error |
| quantum many-body localization | Ep. 40 | the heart of time crystals; only classical here |
| designing real materials and devices | Ep. 41, 44 | only minimal models here |
| the cosmological constant problem | Ep. 45 | ★ only differences; absolute values are impossible |
| resolving the information paradox | Ep. 46 | ★ the gap between "unmeasurable" and "absent" |
| whether inflation happened | Ep. 47 | only calculations inside the model |
| ★ a replacement for the axis | Ep. 48 | ★ looked for; not found |
| general non-equilibrium statistics | untouched | fluctuation–dissipation was used; the framework was not |
| power laws outside physics | untouched | ★ this series stayed within physics |
| proposing experiments | partly | what to measure came out; no hardware discussion |
The point of §05
★ Writing down what remains is the best handover to whoever walks on.
★★ Of this table, the first line looks closest to reach — a procedure for saying, statistically, "a power law over how many decades" from noisy real data. Episode 49's handbook was only tested on noiseless functions.
★ The furthest is the cosmological constant problem — these tools handle only differences, so it is out of reach in principle.
06The one line of fifty episodes
Episode 1's question
"Differentiate a sound and it goes high, integrate it and it goes low.
So what does the space in between — a fractional order — mean?"
★★★ Fifty episodes later
The order \(\alpha\) turned out to be the dB/oct, the phase, the dimension,
the Hurst exponent, the AdS mass and the structure-function exponent —
one and the same number.
And there is exactly one condition for using that axis —
scale invariance holding over some number of decades.
★★★ No formula unknown to humanity came out of it.
What came out is "another reading of the same formula" and "what to measure."
07What is worth keeping anyway
| Worth keeping | Where it was learned |
|---|---|
| ★ Never delete a calculation you lost | Since Ep. 15. All 10 failed predictions are still in the text |
| ★ Never call organizing a discovery | Ep. 43. Episodes 38 and 39 say "translation" in their own text |
| Always count how many decades it lasts | Ep. 27, 48, 49 |
| Always record which moment you measured with | Ep. 42, 47 |
| Compute "difference to test ÷ measurement error" first | Ep. 42, 47, 49 |
| What has not been checked does not become a claim | Ep. 29, 30, 38 |
The point of §07
★ The most productive episode was the one that was most wrong. Episode 32 predicted the field would settle in the trough; it did not, and out of that came "symmetry breaking requires dissipation," which paid off directly in Episode 40 (time crystals) and Episode 44 (metamaterials).
★★ The failed predictions carried us further than the successful ones — that was the biggest surprise of fifty episodes.
Exercises
- §06 says no new formula came out. Were the fifty episodes wasted, then?
Show answer
Not wasted — but not a "discovery" either. As Episode 43 put it, re-shelving does three things — ① you notice when the same thing appears twice ② you notice what is missing ③ you learn which quantities are sensitive.
★ ③ is the most practical thing this series produced — Episode 42's "intermittency moves the dB/oct by only 0.081," Episode 47's "\(\phi^2\) passes on \(n_s\) but is excluded by \(r\)." Both let you say, before measuring, whether the measurement is worth doing.
★★ But that is organizing existing knowledge, not new physics. ★ Continuing to say so is the thing this series guarded most carefully. - §04 stresses "zero dependencies." Why is that a policy?
Show answer
Because "the calculation can be reproduced" is very nearly this series' only guarantee. No experiments were performed here; where literature values were used, they are labelled as such.
★ What remains is only "do the stated numbers really follow from the stated formulas?" — and that had to be checkable by anyone.
★★ Depending on a library means depending on its version — a calculation that will not run in ten years guarantees nothing. ★ That is why the FFT, the Jacobi method and the complex gamma function were each written where needed (13,966 lines in total). - Of §05's unfinished list, which should be taken up next, and why?
Show answer
The first line: handling noisy real data. Three reasons —
★ ① The tools already exist: Episode 49's handbook works on noiseless functions; only the statistics need adding.
★ ② The payoff is large: put error bars on "a power law over how many decades," and every claim since Episode 27 becomes quantitative.
★ ③ It is falsifiable: applied to real data, the handbook's failures will show up immediately. Which fits the rule set in Episode 15 — having declared the hypotheses falsifiable, go looking for falsifications.
★★ The furthest, conversely, is the cosmological constant problem: these tools handle only differences, so it is out of reach in principle (Episode 45 §08).
Summary — the map
Episode 1's \(\text{dB/oct}=6\alpha\) and \(\text{phase}=90\alpha\) were still the axis 49 episodes later. Aliases kept arriving — dimension \(d=2\alpha+3\), Hurst \(H=\alpha-\frac12\), AdS's \(\nu=\alpha/2\), spectral index \(\beta=2\alpha\) — which is what made "turbulence is 5/6-order" sayable.
The order leaves the integers for five reasons (superposition, dimension, criticality, bulk mass, distribution), and the axis ends in three ways (integers, distributions, no power law).
All 28 numbers were recomputed and checked (zero FAILs) — though that only guarantees internal consistency; what guards against contradicting nature is the list of failed predictions.
Counting the series itself — 49 episodes, 49 body files in each language, 48 verification scripts, 13,966 lines, zero dependencies. FFT, Jacobi, complex gamma: each written where needed, because reproducibility is very nearly the only guarantee here.
★★★ Fifty episodes later —
the order \(\alpha\) turned out to be the dB/oct, the phase, the dimension,
the Hurst exponent, the AdS mass and the structure-function exponent.
One and the same number.
One condition for using it — scale invariance holding over some number of decades.
★ And honestly — no formula unknown to humanity came out of it. What came out is "another reading of the same formula" and "what to measure."
★★ The most productive episode was the one that was most wrong. Episode 32's prediction failed, and the "symmetry breaking requires dissipation" that came out of it paid off directly in Episodes 40 and 44. ★ The failed predictions carried us further than the successful ones — the biggest surprise of fifty episodes.
★ Next time: "how many decades of power law," from noisy real data — error bars for Episode 49's handbook.
★★ How far wrong does a slope estimated from a short time series go? And what is the shortest record that earns the phrase "a power law"? ★ Having spoken in decades since Episode 27, it is only right to put an error bar on the decades themselves.
This document is Episode 50 of the "Thinking in Waves" series. Computations are in kenshou/chizu.js. What this episode is: it introduces no new physics; it organizes Episodes 1–49 onto one page and recomputes, independently, the numbers this series published. There is therefore no discovery in this episode. The checks in §03 confirm that the series' own statements do not contradict each other; they do not confirm that the formulas or their premises are correct (the same caveat as Episode 43). The Watson integral's 0.5 % discrepancy is numerical error from a \(200^3\) grid; a finer grid approaches the literature value of 0.340537. The statistics in §04 were obtained by counting files while this article was being written, and increase when Episode 50 is added. In §01's table, "dimension \(d=2\alpha+3\)" simply inverts Episode 30's \(\alpha=(d-3)/2\); it is not a claim that non-integer \(d\) has physical meaning. The "what is left" list and the choice of "what to take up next" in §05 are this series' assessment of itself, not a third party's. — To print, use your browser's Print and "Save as PDF."