Twenty rods, no water. 0.3 of a swing for noise to reach the top.
Turbulence is what water and air do when they get wild. A 2026 preprint gets its main signs from a hanging chain of 20 rods, each half the size of the one before.
Everything in it is a computer run. One swing here means one turn of the biggest rod.
Next question: how can a chain of rods act like a storm?
In water, every ingredient of turbulence arrives at once.
Stir a cup of tea and you get swirls inside swirls. Big swirls hand their energy to smaller ones, until it ends up as heat (p.1).
But water also has pressure, spinning vortices and space. So nobody can say which ingredient causes which sign of turbulence (p.1).
Physicists built toy models to pull the ideas apart. Those models are designed by hand, so they cannot be built and measured (p.2).
Next question: could a real object do it with almost nothing in it?
Source: p.1, p.2
What is inside each system
- Pressure
- Vortices
- Space
- Many sizes
- Many sizes
Water's list: p.1. The chain's list: the paper's design, p.2.
Build a chain where every rod is half the size and half the weight of the last.
Twenty rods hang from a pivot. No gravity and no springs (p.2, p.13, p.16).
Rod 1 has length 2 and mass 1, in the paper's own units. Rod 20 is 524,288 times shorter and lighter (computed: 2 to the power 19).
Each rod is tied only to its neighbors. Yet each one carries the weight of everything beyond it, so every rod feels the whole chain (p.3-4).
The authors note that a chain like this is a lumped version of a whip (p.4).
Next question: shake the big end. Where does the energy go?
Source: p.2 Eq.1, p.3-4, p.16. Lengths computed from the halving rule.Rod length by rod number: halving, twenty times
Computed: length of rod k = 2 x 0.5 to the power (k minus 1). Units are the paper's own (p.16).
Shake the big end, and energy runs down the chain at a steady rate.
A motor pushes the two biggest rods. Friction acts at the joints (p.4-5).
In the middle links, the same amount of energy passes through every link, every second. Physicists call this a cascade (p.6).
Weaken the friction and the steady stretch grows. It ends near link 9 when friction is 1e-9, and near link 14 when it is 1e-14. That is 1.0 links per tenfold drop, against 1.1 predicted (p.6).
Next question: what if friction goes all the way to zero? Does the heat go too?
Source: p.4-6, Fig.2b. The 1.1 is computed from Eq.21 on p.6.
Turn the friction almost off, and the heat does not go away.
Friction is what turns motion into heat. You would expect less friction to mean less heat (p.10).
The authors tried friction from 1e-4 down to 1e-14, ten decades (p.17). With a fixed push, the heat made per second settles at about 0.34 and stops falling (p.10).
The rods swing ever more wildly at ever smaller sizes. That makes up for the weak friction (p.10).
Contested in real 3D fluids: whether this is the strong or the weak form is still debated (p.10, ref 12). The chain shows the strong form within the range tested.
Next question: distant rods tug on each other all the time. Why does the flow stay orderly?
Source: p.10, Fig.6a, p.16, Table I (p.17)Heat made per second, weak-friction end
Mean energy dissipation rate, fixed push (amplitude 0.1, frequency 0.01). The plateau is a flat stretch in the data, not a proof of a limit (p.10, p.16).
Distant rods tug at each other all the time, but the tugs cancel out.
Every rod feels every other rod at every instant. Even several links away, the tug is about as big as the average flow (p.7).
Forward tugs and backward tugs cancel. What is left, the net, halves with each step of distance (p.7).
Seven steps away, the net is 1/128 of the instant tug (computed from 2 to the power minus 7). The paper says about two orders of magnitude (p.7).
Why it cancels: the paper ties it to the smaller rods being lighter. That is the authors' reading (p.7).
Next question: so the flow is local on average. How fast can a tiny wobble spread?
Source: p.7, Fig.3a. Bar lengths are computed from the 2 to the power minus n law.Net pull from a rod n steps away
Soft bar: an order-of-size estimate from the text, not a measured value (p.7). Other bars: computed, 2 to the power minus n, fraction of the instant tug.
A speck of noise climbs to the biggest rod in about 0.3 of one swing.
Every real chain jiggles from heat. The authors add that tiny noise at the smallest rods and run 8,192 simulations: 64 groups of 128 that start from the exact same state (computed: 64 x 128; p.11, p.15).
The small rods wobble first. The wobble then climbs, one rod size at a time. By 0.3 of one big-rod swing, the big rods already differ from run to run (p.11).
Why so fast? Small rods swing fast. If each step up takes one local swing, the steps halve each time, and the last step takes as long as the 19 steps before it put together (computed from the idea on p.11).
Next question: if a chain can do that, what could we do with it?
Source: p.11, Figs.6b-c, p.15. Bar heights computed under the paper's one-swing-per-step idea.The climb, step by step, from the smallest rod to the biggest
Rod 1's turnover time = 1. The paper's own number: the uncertainty reaches the large scales by about 0.3 of one big-rod period (p.11).
You may need less than you think to get turbulence. And you might tune it.
The authors' reading: three ingredients are enough. A set of sizes from big to small, a coupling between them, and a push at one end with drag at the other (p.12).
They also predict that how rough the flow is depends on how fast mass shrinks compared with length (p.6, Eq.23). Same rate gives this chain. Mass that stays the same gives the 1/3 rule of real fluids.
That is a prediction. The paper simulates only one setting (p.6), and reports no built chain (p.15-16).
What you can do: the authors link a live chain you can shake in a browser (see Sources), and say their data comes out when the paper is published (p.14).
Next question: what could go wrong with all this?
Source: p.6 Eq.23, p.12, p.14, p.15-16, p.22A prediction, not a result: the roughness dial
| Mass shrinks | Dial | Reads as |
|---|---|---|
| Not at all (alpha = 0) | 1/3 | The rule for real fluids (p.6) |
| As fast as length (alpha = 1) | 0 | The chain in this paper (p.6) |
| Faster than length (alpha above 1) | below 0 | Spikier than the fluid rule allows (p.6, p.22) |
Dial = (1 minus alpha) / 3 (Eq.23). Not tested in the simulations. The authors call it a candidate, not a theorem (p.22).
What the paper shows, and what it only hopes for.
Everything here is a computer run of 20 rods with a size ratio of 2 (p.16). As far as the paper reports, no one has built the chain or the circuit version (p.15-16).
Not tested yet: other size ratios (p.6), a chain with endless links (p.13), and real fluid data. The rule linking ratios to roughness is a guess from units, not a proven theorem (p.22).
The data comes only at publication (p.14). This is a preprint, one day old, and I found no replication.
Next question: will the data hold up? That is the next story.
Source: p.6, p.13, p.14, p.15-16, p.16, p.22Shown in simulation, or only claimed
- Steady energy flow (p.6-7)
- Local net from tangled tugs (p.7)
- Heat plateau, 0.34 (p.10)
- Noise climbing to big rods (p.11)
- Other size ratios (p.6)
- A built chain (p.15-16)
- Endless links (p.13)
- Real fluid test
Real fluid test: none found in the 25 pages. Absence by my read.
Six numbers from the story
rods, each half the size of the one before.
p.2, p.16times lighter: rod 20 vs rod 1. Computed, 2 to the power 19.
p.16links gained per tenfold drop in friction: seen vs predicted.
p.6, Eq.21heat made per second, flat as friction falls.
p.10of one big-rod swing for noise to reach the big rods.
p.11noisy runs: 64 x 128. Computed.
p.11, p.15