Steven Strogatz · Book 7: Scientist Gallery

Part I · The Example Underneath the Helix

Two lines on page 199

Nonlinear Dynamics and Chaos reaches limit cycles in Chapter 7, defines one on page 198 as an isolated closed trajectory, and then gives the first example anyone learns:

EXAMPLE 7.1.1: A SIMPLE LIMIT CYCLE (Strogatz, 2nd ed., p. 199) ṙ = r(1 − r²), θ̇ = 1, r ≥ 0 r* = 0 unstable, r* = 1 stable; every other trajectory spirals onto the unit circle, and on it x(t) = cos(t + θ₀).

Now the flow at the centre of Book 6's WP-22, the contact-geometric toy model on the solid cylinder:

ṙ = f(r)(1 − e^{−z}), θ̇ = 1, ż = 1 closure conditions: f(1) = 0, f′(1) = −2, f(r)(r−1) < 0 canonical closure: f_cub(r) = r − r³

Substituting the canonical closure and deleting the third coordinate returns Example 7.1.1 exactly. The helix $\Gamma = \{r = 1\}$ is that example's unit circle carried along $z$ at unit rate; $T^{*} = 2\pi$ is its $\dot\theta = 1$; and the transverse eigenvalue $-2$ is $\frac{d}{dr}[r - r^{3}]$ at $r = 1$, which is $1 - 3 = -2$. ch-strogatz-verify.py block [1] integrates the example from six starting radii and lands on $r = 1$ to fourteen places.

What the closure conditions are

The three conditions on $f$ are not findings about the flow; they are the definition of an admissible closure, and $f'(1) = -2$ is imposed on every closure that qualifies. Block [2] confirms that both canonical closures — the cubic and the saturating $f_{\mathrm{sat}}(r) = -2(r-1)/(1+(r-1)^{2})$ — satisfy all three, and that both have $f'(1) = -2$ because they were required to. A statement that concludes $\mu = -2$ from that hypothesis has returned the hypothesis. The theorem in WP-22 does more than that, and Part II is what it does.

The ledger

quantity	whose	where it comes from
Γ = {r=1} exists and attracts	textbook	Example 7.1.1, p. 199
T* = 2π	textbook	θ̇ = 1; Rule of Thumb 2, p. 254
μ_max = −2	textbook	f′(1), imposed; = d/dr[r−r³] at 1
multiplier e^−4π	textbook	consequence of the two above
μ survives the e^−z modulation	corpus	WP-22 Thm 2.1 — integrability
neutral line z = 0, time reversal	corpus	WP-22 Thm 3.1
closure-dependent finite-time escape	corpus	WP-22 §4, cubic vs. bounded
base-point drift of the multiplier	corpus	ch-grothendieck
contact form α = dz − r²dθ, Reeb field, no-go	corpus	outside Strogatz entirely

Four of the headline numbers the series quotes for Γ are the example's. Five of the nine rows are the corpus's own, and every one of them involves $z$ — which is the coordinate Strogatz does not have.

Part II · What the Third Coordinate Buys

The exponent survives, and that is the theorem

Linearising about $\Gamma$ with $\varepsilon := r - 1$ gives a transverse equation that is independent of the closure, and WP-22 solves it in closed form:

ε̇ = 2ε(e^{−z} − 1), z = z₀ + t ε(t) = ε₀ · exp( −2t + 2e^{−z₀}(1 − e^{−t}) )

Block [3] checks that closed form against RK4 from four base points and two horizons and finds agreement to $10^{-12}$ relative. The interesting quantity is the second term: it is bounded by $2e^{-z_{0}}$ for every $t$, so $|\mu + 2| \le 2e^{-z_{0}}/t$ and the exponent returns to $-2$ no matter where the trajectory started. The modulation is integrable along the $z$-flow and therefore contributes nothing to the rate. That is a real result about a real perturbation, and it is not in the textbook, because the textbook has no swept modulation to be perturbed by. The value $-2$ was never the content; its invariance is.

Where the constant stops being constant

Over one full period from a finite base point the transverse exponent is not $-4\pi$:

∫₀^{2π} 2(e^{−(z₀+t)} − 1) dt = −4π + 2e^{−z₀}(1 − e^{−2π}) z₀ = 0 exponent −10.570105 multiplier 2.567211×10⁻⁵ z₀ = 2 exponent −12.296206 multiplier 4.569049×10⁻⁶ z₀ = 8 exponent −12.565701 multiplier 3.489679×10⁻⁶ z₀ → ∞ exponent −4π multiplier 3.487342×10⁻⁶

Block [4] computes that column both in closed form and by quadrature and finds them equal to $10^{-8}$. The saturated value $e^{-4\pi} = 3.487342\times10^{-6}$ is the one ch-feynman uses for its periodic-orbit weight, and the drift below it is what ch-grothendieck found when it asked whether the multiplier could serve as an analytic index. It cannot: an index does not move with where you start. Both readings are consistent; this chapter only supplies the reason the number moves, which is the $z$-dependence of the contraction rate, which is the corpus's own addition to the example.

Part III · One Word, Two Books

“Degenerate Hopf” does not mean the same thing in both

WP-22's fourth result is a degenerate Hopf bifurcation at the axis: the first Lyapunov coefficient $a(z) = 1 - e^{-z}$ coincides with the linear coefficient $\lambda(z)$, so the emergent cycle radius $\sqrt{\lambda/a}$ is pinned at $1$ instead of opening as $\sqrt{\lambda}$. Measured against Strogatz's Rule of Thumb 1 (p. 254) — the size of the limit cycle grows continuously from zero, proportional to $\sqrt{\mu - \mu_{c}}$ — the degeneracy claim is correct, and block [5] prints the two columns side by side: the generic radius falls to $10^{-3}$ as $z \to 0$ while the model's stays at exactly $1$.

The collision is in the term. Strogatz reserves degenerate Hopf bifurcation, on p. 256, for a different phenomenon:

“At $\mu = 0$ we do not have a true Hopf bifurcation because there are no limit cycles on either side of the bifurcation. Instead, at $\mu = 0$ we have a continuous band of closed orbits surrounding the origin. These are not limit cycles!”

— Strogatz, Nonlinear Dynamics and Chaos, 2nd ed., p. 256, on $\ddot x + \mu\dot x + \sin x = 0$

The helix has an isolated attracting cycle on both sides and no band of closed orbits anywhere. Its degeneracy is a vanishing first Lyapunov coefficient — Guckenheimer and Holmes' usage, and their book is the second entry in the same bibliography. Two books, one term, two objects. Nothing is wrong; one clause is missing.

The correction this chapter asked for — carried into the source 2026-09-15

Name the sense. “Degenerate” is meant in the vanishing-first-Lyapunov-coefficient sense of Guckenheimer & Holmes, not in the nonlinear-centre sense of Strogatz p. 256 — a reader arriving from the textbook in the bibliography would look for a band of closed orbits and find an isolated attracting cycle instead. WP-22's abstract and §5 now say so.

And say where $z$ is frozen. The $2\times2$ Jacobian in that theorem is the Jacobian of the $(x,y)$ subsystem at a held-fixed $z$. In the flow itself $\dot z = 1$ everywhere, so the three-dimensional field has no zero at all — block [6] scans $[0,4]\times[-4,4]$ and finds $\min|(\dot r, \dot\theta,\dot z)| = \sqrt2$ — and Strogatz §8.2 opens “Suppose a two-dimensional system has a stable fixed point.” The caveat was in WP-22's exercises — one says “freeze $z$”, and solution 4 disqualifies the codimension-one normal forms outright because the whole field vanishes at $z = 0$ — and not beside the theorem. It is beside the theorem now, which is retitled Degenerate Hopf at $r=0$, frozen $z$.

Two more went in with them. $\Gamma$ is a helix and not a periodic orbit of the flow, so $T^{*} = 2\pi$ is the period of the $(r,\theta)$ projection — the count is in WP-120; and Example 7.1.1 is now cited where its data are used. All four are listed, dated, in that paper's own Corrections section. No result was withdrawn, and the four figures the source directory was missing were regenerated from its companion script, so the document rebuilds from its own sources again.

A bibliography entry with no citation

Block [7] reads WP-22's own source. When this chapter was written, \bibitem{Strogatz} appeared once, \cite{Strogatz} zero times, and the string 7.1.1 zero times; across the whole geometry repository the name occurred in that one bibliography line and nowhere else. Example 7.1.1 is the standard first example of a limit cycle and belongs to nobody — nothing was taken improperly. What was missing was the pointer: a reader was never told that the flow they were looking at is the one on page 199, and so was never in a position to ask which of its properties arrived with it.

The pointer went in on 2026-09-15 — five citations, three mentions of the example, and a paragraph in §2 naming what the closure conditions are the local data of. Block [7] now checks the repaired state. Same instrument, same file, and the numbers it prints have changed because the file has.

Cross-references · where the mathematics is

The chapters carrying these objects

page	what it carries
Book 6 · WP-22	The contact-geometric toy model on the solid cylinder: the six results, the closure conditions, the degenerate-Hopf theorem, the no-go. The primary object of this chapter. Index entry at Book 6.
vol2-toymodel	Volume II's explicit instantiation, $\dot r = r(1-r^{2}) + 2(r-1)e^{-z}$ — the same example with an additive contact coupling instead of a multiplicative one. Every operator and boundary instantiated and checked by direct computation.
vol2-contact	The contact realization: the manifold, the form, and the $\mu_{\max} < 0 \Leftrightarrow c > 0 \Leftrightarrow \tau > 0$ chain, whose full proof is the open Floquet + SDE item.
Sessão S2	Teorema 2.1 and the asymmetric basin, in Portuguese, and the one place in the corpus that already says it outright: “a forma normal de Hopf sozinha entrega a taxa” — the Hopf normal form alone delivers the rate. The radial eigenvalue is worked as $-2 + \varepsilon e^{-z}$ there.
ch-grothendieck	Measures the transverse multiplier as an index candidate and finds it base-point dependent. Part II above supplies the mechanism.
ch-feynman	Uses $T^{*} = 2\pi$, $\mu_{\max} = -2$, $e^{-4\pi} = 3.487342\times10^{-6}$ and the periodic-orbit weight $1/(2\sinh 2\pi)$. The saturated values, correctly used as such.
WP-58	Derives $r - r^{3}$ from a flat-rotation-curve effective force rather than assuming it as a normal form, and keeps $\partial_{r}[r(1-r^{2})]\|_{1} = -2$ explicitly. The one place the corpus argues for the radial term instead of importing it.
ch-cosmic-no-go	The contact-Hamiltonian obstruction between transverse attractors and de Sitter expansion — WP-22's sixth result, and wholly outside the textbook.
Book 3 · π φ μ η Δ Σ Ω	Where the transverse system is stated for Book 3's chapter cluster, in the additive form.
Ch 15 · The Complex Turn	Vol IV's treatment of the transverse Lyapunov exponent.

Place in the Series

The register for a Strogatz citation here is foundational, not evidentiary. He is not evidence for dm³ and has never worked on it; what his book supplies is the object the corpus's central model is a modification of, and the vocabulary the corpus has been using without opening it. The gallery has chapters on people who built what a statement needed — Grothendieck — and on people who changed the language a statement was made in — Tao. This one is different in kind. It is the chapter for the book already on the shelf, whose Chapter 7 supplies the corpus's attractor, whose Chapter 8 supplies its bifurcation vocabulary, and whose page 256 is why one of those words needs a qualifier.

The finding is small and it is worth printing plainly: building on the standard first example of a limit cycle is the right thing to have done, and four of the numbers that follow are the example's rather than the model's. The results that are the model's own are the five that involve $z$ — the invariance of the exponent under an integrable modulation, the neutral line, the closure-dependent escape, the drift of the multiplier, and the contact no-go. Those are the ones to press on.

References

Strogatz 2015	S. H. Strogatz, Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering, 2nd ed., Westview Press 2015; CRC Press reissue 2018. ISBN 978-0-8133-4910-7. §7.0 p. 198 (a limit cycle is an isolated closed trajectory) · Example 7.1.1 p. 199 · §8.2 pp. 251–256 (supercritical normal form p. 253; Rules of Thumb p. 254; degenerate Hopf p. 256).
Guckenheimer & Holmes	J. Guckenheimer and P. Holmes, Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields, Springer 1983 — the sense in which WP-22's degeneracy is meant, and the second entry in its bibliography.
WP-22	A Contact-Geometric Toy Model on the Solid Cylinder: Transverse Stability, Closure-Dependent Escape, Degenerate Hopf Structure, and a Cosmological No-Go. Source at `book6/differential-equations/helix-toy-model/helix_toy_model.tex`.
Citations checked	`book7/strogatz-citations-verify.py` — 2026-09-17, against the 2018 printing held at the desk (532 pdf pages, 481 printed pages located under pypdf 6.18 — the count depends on the extractor, so the run prints its version beside it). Twenty page-and-example citations across this chapter, WP-122, ch-van-der-pol, ch-smale, ch-euler and ch-conley, each asserted as specific strings on a specific printed page, one of them a verbatim quotation; exit 0, all twenty hold. The script assumes no fixed offset between pdf index and printed page, because there is none in this printing — the run reports three distinct values, 4 to 18 — and reads the number off a running head that is set tight against it. Until this run the page numbers were remembered, not checked.
Verification	`book7/ch-strogatz-verify.py` — 8 blocks, standard library only. Example 7.1.1 integrated from six radii; both closures checked against all three closure conditions; the closed form against RK4 to $10^{-12}$; the per-period exponent in closed form and by quadrature; the generic and pinned radii tabulated to $z = 10^{-6}$; a grid scan showing the field has no zero; and a read of WP-22's own source for the bibitem-to-citation count. Its `[HONESTY]` block states that no theorem checked here is false and that the finding is about provenance and one word.
Internal	Book 6 · vol2-toymodel · vol2-contact · Sessão S2 · ch-grothendieck · ch-feynman · WP-58 · ch-cosmic-no-go