Flow Lab

Syllabus / Module 3: First-Order ODEs

Lesson 3.9

Bifurcations

How equilibria appear, vanish and swap stability as a parameter changes: saddle-node, transcritical and pitchfork bifurcations, hysteresis.

Written byClaude Sonnet

🌊
Why this matters for simulation: Flow past a cylinder changes character (steady → vortex shedding) at a critical Reynolds number — a bifurcation.

By the end of this lesson you can…

  • Read a bifurcation diagram: vertical slices are phase lines, solid branches are stable, dashed branches are unstable.
  • Identify and sketch the saddle-node, transcritical and pitchfork bifurcations from their normal forms, and say where each branch exists and is stable.
  • Explain critical slowing down with the relaxation time 1/|f'|, and say honestly what it can and cannot predict.
  • Construct a hysteresis loop from two folds, and distinguish it from the lag of a finite sweep speed.
  • Follow an unstable branch with Newton continuation, and explain why a time-stepper cannot.
  • Recognise when a steady-state stopping rule is fooled by the ghost of a vanished equilibrium, and state why vortex shedding is not a one-dimensional bifurcation.
On this page
·
  1. Parameter-dependent equations and the bifurcation diagram
  2. Saddle-node: equilibria collide and annihilate
  3. Transcritical: two equilibria swap stability
  4. Pitchfork: symmetry and its breaking
  5. Hysteresis
  6. Tipping points
  7. What a solver sees
  8. Flow past a cylinder
  9. Exercise: continuation along stable and unstable branches
  10. Problems
  11. Key Takeaways

Turn one knob of a simulation slowly: a harvest rate, a Reynolds number. For a long time only smooth changes happen, then the answer jumps, and turning the knob back does not undo it. Nothing is broken; the equation has a place where its set of long-run states changes.

3.8 Autonomous Equations, Equilibria & Stability ended on such a place: the harvested logistic equation had two equilibria that merged at h=0.25h = 0.25 and vanished. This lesson studies that event, finds its cousins, and asks what a numerical method sees next to one.

▶ Video Phase lines stacked into a bifurcation diagram, the three local bifurcations, hysteresis loops, and a solver crawling through the ghost of a vanished equilibriumcaptions only
Transcript
  1. Turn a knob slowly and the answer can suddenly jump. Last lesson's harvest model did: as h grows, the graph of f slides down, the equilibria merge at one quarter and vanish. With u equals y minus a half and r equals a quarter minus h, that is u prime equals r minus u squared: a saddle-node.
  2. Stack the phase lines. For r below zero, f is negative everywhere: no equilibria. At zero, one. Above zero, plus and minus root r, with arrows from the sign of f. Let the equilibria trace out branches, solid where stable, dashed where unstable: the bifurcation diagram.
  3. Near the fold the stable slope is minus two root r, so the relaxation time is one over two root r: 2.2361 at h equals 0.2, 5 at 0.24, 15.8114 at 0.249, growing without bound. This critical slowing down warns that a fold is near. It is not a clock: noise can tip the system sooner.
  4. Transcritical: y prime equals r y minus y squared. The equilibria zero and r pass through each other. The slope at zero is r, at r it is minus r, so at r equals zero they swap stability, and the solid and dashed styles switch.
  5. Supercritical pitchfork: y prime equals r y minus y cubed. Zero loses stability at r equals zero, though even there y prime is minus y cubed, still stable. Two new stable branches, plus and minus root r, are born, with slope minus two r.
  6. Subcritical: y prime equals r y plus y cubed minus y to the fifth. Nonzero equilibria have y squared solving z squared minus z minus r equals zero. At r equals minus 0.2 that gives 0.7236 and 0.2764, the harvest equilibria at h equals 0.2: the same quadratic. The large branch is stable, the small one unstable. The fold lies left of the pitchfork.
  7. Start at zero and raise r slowly. It stays at zero through the interval from minus a quarter to zero, because zero is still stable, then at zero jumps up. Lower r: it stays on the large branch until the fold at minus a quarter, then falls. Two stable states coexist over an interval, so the state depends on history: hysteresis.
  8. Another loop: y prime equals r plus y minus y cubed, with folds at plus and minus 0.3849. A real sweep lags. At rate 0.002 the jumps come near plus and minus 0.416, at 0.0005 near 0.397. The lag shrinks as the sweep slows; the loop does not.
  9. A tipping point is a fold whose jump is not undone by returning the parameter. After the jump at 0.3849, bring r back to 0.30 and the state stays up; it falls only at minus 0.3849. These are caricatures; nothing here shows that any real system is near one.
  10. A time-stepper cannot follow unstable branches: at r equals 0.30 the arrows push away from the middle one. Continuation solves f equals zero by Newton's method, stepping r. It follows the middle branch: minus 0.101 at 0.10, minus 0.339 at 0.30, minus 0.429 at 0.35. At 0.40 it fails: the fold is behind us.
  11. The ghost. Past a fold, y prime equals minus epsilon minus y squared has no equilibrium, but y prime is tiny near zero. With epsilon 0.0001, a rule that stops when y prime is under 0.001 fires at t equals 31.2, y equals 0.03, and reports steady. The solution really reaches minus one at 312.16: pi over root epsilon, minus two.
  12. Flow past a 2D circular cylinder is steady below a critical Reynolds number, about 47, from memory, and sheds a periodic vortex street above it. That is a Hopf bifurcation: it cannot occur in one dimension, where solutions never oscillate, so it needs two. A preview of systems, not an example from this lesson.
  13. Recap: a bifurcation is where the phase line changes as a parameter moves. Solid stable, dashed unstable. Slowing down warns of a fold, hysteresis is memory, a ghost fools stopping rules.

Parameter-dependent equations and the bifurcation diagram

Let the right side depend on a parameter rr:

y′=f(y,r).y' = f(y, r).

For each fixed rr this is an autonomous equation with its own phase line. A bifurcation is a value r∗r^* at which the phase line changes qualitatively: equilibria appear, vanish, or swap stability.

The bifurcation diagram is the whole family at once: rr horizontal, equilibria y∗y^* vertical, each branch y∗(r)y^*(r) solid where stable and dashed where unstable. A vertical slice at one rr is that rr‘s phase line, with arrows read off the sign of ff. The sim below draws it beside the graph of ff and the solutions.

Where? If f(y∗,r∗)=0f(y^*, r^*) = 0 and fy(y∗,r∗)≠0f_y(y^*, r^*) \ne 0, the implicit function theorem gives one smooth equilibrium y∗(r)y^*(r) through that point, and fyf_y keeps its sign nearby, so nothing appears, vanishes or swaps. A bifurcation needs fy=0f_y = 0: the linear test of lesson 3.8 goes silent exactly at the interesting points. That is necessary, not sufficient: for y′=r−y3y' = r - y^3, fy=0f_y = 0 at (0,0)(0, 0), yet the single branch y∗=r1/3y^* = r^{1/3} is smooth and stable for every rr.

Saddle-node: equilibria collide and annihilate

The normal form is

y′=r−y2.y' = r - y^2 .

For r>0r \gt 0 the equilibria are y∗=±ry^* = \pm\sqrt r with fy=−2yf_y = -2y: +r+\sqrt r is stable (fy=−2rf_y = -2\sqrt r), −r-\sqrt r unstable (+2r+2\sqrt r). At r=0r = 0 they merge at y=0y = 0, where f=−y2≤0f = -y^2 \le 0: semi-stable, fy=0f_y = 0. For r<0r \lt 0, f<0f \lt 0 everywhere and every solution decreases for ever. The diagram is a sideways parabola, upper half solid, lower half dashed.

This is the harvest model in other coordinates. Put u=y−12u = y - \tfrac12 in y′=y(1−y)−hy' = y(1-y) - h: then y(1−y)−h=14−h−u2y(1-y) - h = \tfrac14 - h - u^2, so with r=14−hr = \tfrac14 - h it is u′=r−u2u' = r - u^2. At h=0.2h = 0.2: r=0.05r = 0.05, u=±0.2236u = \pm 0.2236, y=0.7236y = 0.7236 and 0.27640.2764, the equilibria of lesson 3.8. Raising hh is lowering rr.

Why this is the generic fold. If f=fy=0f = f_y = 0 at (y∗,r∗)(y^*, r^*) with fyy≠0f_{yy} \ne 0 and fr≠0f_r \ne 0, then f≈fr (r−r∗)+12fyy(y−y∗)2f \approx f_r\,(r - r^*) + \tfrac12 f_{yy}(y - y^*)^2: after rescaling, r−y2r - y^2 or its mirror image.

Critical slowing down. At the stable equilibrium fy=−2rf_y = -2\sqrt r, so the relaxation time is

τ=1∣fy∣=12r,r>0,\tau = \frac{1}{|f_y|} = \frac{1}{2\sqrt r}, \qquad r \gt 0,

which diverges as r→0+r \to 0^+. For the harvest model r=14−hr = \tfrac14 - h gives τ=1/1−4h\tau = 1/\sqrt{1-4h}, the figures quoted in lesson 3.8 and two more:

hhrrτ=1/(2r)\tau = 1/(2\sqrt r)
0.2000.2000.05000.05002.2362.236
0.2400.2400.01000.01005.0005.000
0.2490.2490.00100.001015.8115.81

Kick the system and time its recovery: a recovery time that grows as a parameter drifts is a known, measurable early-warning signal of a fold.

What it does not do. It says a bifurcation is approaching, not when the system tips: noise can push the state across the unstable equilibrium before rr reaches 00, since the gap shrinks like r\sqrt r. Rising τ\tau is a reason to look harder, not a clock.

◆ Interactive Bifurcation explorer: drag r, click to pick y(0), and press Sweep r. Start on the saddle-node, then try the others.

Loading simulation…

  • Saddle-node. Drag rr through 00: two equilibria, one (“inconclusive”), none. Sweep r: the state follows the stable branch, crawls past r=0r = 0, escapes, and does not return.
  • Transcritical, supercritical pitchfork. The sweep legs nearly retrace each other; the small gap is the lag of a finite sweep speed.
  • Subcritical pitchfork, cusp. A loop, below.

Transcritical: two equilibria swap stability

y′=ry−y2=y (r−y).y' = r y - y^2 = y\,(r - y).

The equilibria are y∗=0y^* = 0 and y∗=ry^* = r for every rr, with fy=r−2yf_y = r - 2y: fy(0)=rf_y(0) = r, fy(r)=−rf_y(r) = -r. For r<0r \lt 0, 00 is stable and rr unstable; for r>0r \gt 0 the roles swap; at r=0r = 0 they coincide and f=−y2f = -y^2 is semi-stable. Nothing is destroyed: the equilibria pass through each other and exchange stability.

The logistic connection. Write logistic growth as y′=ry−c y2y' = r y - c\,y^2, carrying capacity K=r/cK = r/c. With crowding c=1c = 1 fixed and the growth rate rr as parameter, K=rK = r and this is the transcritical form. As rr falls through 00 the equilibrium 00 goes from unstable to stable: the line between persistence and extinction. The branch y=r<0y = r \lt 0 is a negative “population”, meaningless in ecology but visible in the mathematics.

Pitchfork: symmetry and its breaking

Supercritical. y′=ry−y3=y (r−y2)y' = r y - y^3 = y\,(r - y^2) is unchanged by y→−yy \to -y. For r≤0r \le 0 the only equilibrium is 00. For r>0r \gt 0 there are three, 00 and ±r\pm\sqrt r, with fy=r−3y2f_y = r - 3y^2:

fy(0)=r>0  (unstable),fy(±r)=r−3r=−2r<0  (stable).f_y(0) = r \gt 0 \;\text{(unstable)}, \qquad f_y(\pm\sqrt r) = r - 3r = -2r \lt 0 \;\text{(stable)}.

At r=0r = 0 the equation is y′=−y3y' = -y^3: stable by the sign of ff although fy=0f_y = 0. One stable state splits into an unstable one and two stable ones, and an arbitrarily small asymmetry in the start decides which: symmetry breaking. The relaxation times 1/(2r)1/(2r) at ±r\pm\sqrt r and 1/r1/r at 00 diverge as r→0+r \to 0^+. (Checked on 20002000 values of rr in (0,5](0, 5].)

Subcritical. y′=ry+y3−y5=y (r+y2−y4)y' = r y + y^3 - y^5 = y\,(r + y^2 - y^4): the cubic now pushes outward and the quintic catches the solution later. Away from 00, with z=y2z = y^2, the equilibria solve z2−z−r=0z^2 - z - r = 0:

y2=1±1+4r2,real only for r≥−14.y^2 = \frac{1 \pm \sqrt{1 + 4r}}{2}, \qquad \text{real only for } r \ge -\tfrac14 .

The "++" root is positive for every r>−14r \gt -\tfrac14 (the large branch); the "−-" root is positive only for −14<r<0-\tfrac14 \lt r \lt 0 (the small branch). At r=−0.2r = -0.2: y2=0.7236y^2 = 0.7236 and 0.27640.2764, so y=±0.8507y = \pm 0.8507 and ±0.5257\pm 0.5257. These are the harvest equilibria at h=0.2h = 0.2: the same quadratic, y2−y+h=0y^2 - y + h = 0 with h=−rh = -r. The fold at r=−14r = -\tfrac14 is the harvest bifurcation, at y2=12y^2 = \tfrac12.

Stability, exactly. On an equilibrium z2=z+rz^2 = z + r, so fy=r+3z−5z2=−2z−4rf_y = r + 3z - 5z^2 = -2z - 4r. With s=1+4rs = \sqrt{1+4r} and z=(1±s)/2z = (1 \pm s)/2 this is fy=−s (s±1)f_y = -s\,(s \pm 1):

  • large branch (upper sign): fy=−s(s+1)<0f_y = -s(s+1) \lt 0, stable for every r>−14r \gt -\tfrac14;
  • small branch (lower sign): fy=s(1−s)>0f_y = s(1 - s) \gt 0 for 0<s<10 \lt s \lt 1, that is for −14<r<0-\tfrac14 \lt r \lt 0: unstable there;
  • the origin: fy(0)=rf_y(0) = r, stable for r<0r \lt 0, unstable for r>0r \gt 0. At r=0r = 0, f=y3−y5>0f = y^3 - y^5 \gt 0 just above 00: unstable, although fy=0f_y = 0.

Check at r=−0.2r = -0.2: fy=−0.647f_y = -0.647 and +0.247+0.247. So: below r=−14r = -\tfrac14 only 00 (stable); at −14-\tfrac14 a saddle-node creates the large and small pair; for −14<r<0-\tfrac14 \lt r \lt 0 there are three stable states (00 and ±\pm large) separated by unstable small branches; at r=0r = 0 the small branch collapses onto the origin, which turns unstable. The fold sits to the left of the pitchfork, and that ordering makes hysteresis.

✅ Concept check

For $y' = r - y^3$, $f_y = 0$ at $(y, r) = (0, 0)$. Is $r = 0$ a bifurcation point?

✅ Concept check

For $y' = r y + y^3 - y^5$ at $r = -0.1$, how many equilibria are stable?

Hysteresis

Definition. A system shows hysteresis when its long-run state depends on its history, because two stable states coexist over an interval of rr.

Subcritical pitchfork. Start at r=−0.5r = -0.5 on y=0y = 0 and raise rr very slowly. The state stays at 00 through the whole bistable interval (−14,0)(-\tfrac14, 0), because 00 is still stable. At r=0r = 0 it turns unstable and the state jumps to the large branch, y=±1y = \pm 1. Now lower rr: the state stays on the large branch down to r=−14r = -\tfrac14, where it folds away, and falls to 00. In the slow limit the jumps are at r=0r = 0 (up) and r=−14r = -\tfrac14 (down): a loop of width 14\tfrac14.

The cusp model. y′=r+y−y3y' = r + y - y^3 has equilibria on r=y3−yr = y^3 - y. Folds are where fy=1−3y2=0f_y = 1 - 3y^2 = 0, so y=±1/3y = \pm 1/\sqrt3, at r=∓(133−13)=±233=±0.3849r = \mp(\tfrac{1}{3\sqrt3} - \tfrac{1}{\sqrt3}) = \pm\tfrac{2}{3\sqrt3} = \pm 0.3849. For ∣r∣<0.3849|r| \lt 0.3849 there are three equilibria, two stable and one unstable; at r=0r = 0: 00 (unstable, fy=1f_y = 1) and ±1\pm 1 (stable, fy=−2f_y = -2). Sweeping up from r=−0.8r = -0.8 (where y=−1.2756y = -1.2756) the state runs to the fold at (0.3849,−0.5774)(0.3849, -0.5774) and jumps to y=1.1547y = 1.1547; sweeping down it runs along the upper branch to (−0.3849,+0.5774)(-0.3849, +0.5774) and jumps to −1.1547-1.1547. Each jump has length 3\sqrt3.

What a real sweep does. The above is the infinitely slow limit. The sim and the playground cell integrate with RK4 (step 0.050.05) while rr moves at speed dr/dtdr/dt. At dr/dt=0.002dr/dt = 0.002 the cusp state moves fastest at r=+0.416r = +0.416 going up and −0.416-0.416 going down; at 0.00050.0005, ±0.397\pm 0.397; static, 0.38490.3849. For the subcritical pitchfork at 0.0020.002 (with a noise floor ∣y∣≥0.05|y| \ge 0.05, since the exact equilibrium 00 would hold the state for ever) it is r=+0.081r = +0.081 up (static 00) and −0.289-0.289 down (static −0.25-0.25). A finite sweep lags: the jump comes after the bifurcation. The lag shrinks as the sweep slows; the loop does not, converging to the width of the bistable interval. That is the test: a lag closes as the sweep slows, hysteresis does not.

◆ Interactive Subcritical pitchfork: press Sweep r and watch the two legs jump at different r. Then choose the cusp from the menu.

Loading simulation…

In the bistable interval click on both sides of the dashed small branch: the solution goes to 00 or to the large branch. The unstable branch is the boundary between the basins, invisible in a time-stepper’s long-run output, and it merges with a stable branch at the fold.

Tipping points

A tipping point is a fold where a system on a vanishing branch must jump to a distant state, and returning the parameter to its old value does not undo the jump. In the cusp model the lower state disappears at r=0.3849r = 0.3849 and the state jumps up; at r=0.30r = 0.30 it is still on the upper branch, and to get back you must go to r=−0.3849r = -0.3849. That is hysteresis. The harvest model is the collapse version: past h=0.25h = 0.25 the population goes extinct, and lowering hh does not revive it because y=0y = 0 is absorbing.

What this does not say. These are caricatures, showing a mechanism: a fold plus a distant stable state to land in. Whether a particular lake, fishery or ice sheet has such a fold, and how close it is, is an empirical question. Nothing here shows that any real system is near a tipping point.

What a solver sees

The ghost of a vanished equilibrium. Take the normal form just past the fold, y′=r−y2y' = r - y^2 with r=−εr = -\varepsilon, ε>0\varepsilon \gt 0:

y′=−(ε+y2).y' = -(\varepsilon + y^2).

There is no equilibrium, but ∣y′∣|y'| is tiny near y=0y = 0: a bottleneck. Separating variables (lesson 3.3), the time to pass it is

T=∫−∞∞dyε+y2=1ε[arctan⁡yε]−∞∞=πε.T = \int_{-\infty}^{\infty} \frac{dy}{\varepsilon + y^2} = \frac{1}{\sqrt\varepsilon}\Big[\arctan\frac{y}{\sqrt\varepsilon}\Big]_{-\infty}^{\infty} = \frac{\pi}{\sqrt\varepsilon}.

Over y∈[−1,1]y \in [-1, 1] the exact time is 2εarctan⁡1ε=π/ε−2+O(ε)\frac{2}{\sqrt\varepsilon}\arctan\frac{1}{\sqrt\varepsilon} = \pi/\sqrt\varepsilon - 2 + O(\varepsilon), and RK4 reproduces it:

ε\varepsilonRK4, y=1→−1y = 1 \to -1exactπ/ε\pi/\sqrt\varepsilon
10−210^{-2}29.42229.42229.42329.42331.41631.416
10−310^{-3}97.34697.34697.34797.34799.34699.346
10−410^{-4}312.159312.159312.159312.159314.159314.159

The passage time grows like 1/ε1/\sqrt\varepsilon, the mirror of critical slowing down, τ=1/(2r)\tau = 1/(2\sqrt r).

This breaks “run until steady”. A solver that stops when ∣y′∣<tol|y'| \lt \text{tol} meets ∣y′∣=ε+y2|y'| = \varepsilon + y^2, which dips below tol exactly when y2<tol−εy^2 \lt \text{tol} - \varepsilon: possible only if ε<tol\varepsilon \lt \text{tol}. With ε=10−4\varepsilon = 10^{-4} and tol =10−3= 10^{-3}, from y=1y = 1, the rule fires at t=31.20t = 31.20 with y=0.0300y = 0.0300 and the program reports a steady state. There is none: the true solution goes through 00 and reaches y=−1y = -1 at t=312.16t = 312.16. Residual tests that work far from a bifurcation fail near one. A defence: evaluate the linearisation where the rule fires. Here ∣fy∣=2∣y∣=0.06|f_y| = 2|y| = 0.06, tiny, which signals a bottleneck, and 1/∣fy∣1/|f_y| says how long the run must be.

Euler near a fold. The limit h<2/∣fy∣h \lt 2/|f_y| of lesson 3.2 is set by the largest ∣fy∣|f_y| visited (22 at y=±1y = \pm 1, so h<1h \lt 1), not by the bottleneck where fy→0f_y \to 0. Accuracy is the issue: for ε=10−4\varepsilon = 10^{-4}, h=0.5h = 0.5 takes 623623 steps and gets 311.800311.800 against 312.159312.159; h=0.01h = 0.01 gets 312.159312.159.

🐍 PythonPlayground: slow sweeps and hysteresis, Newton continuation, and the crawl through a bottleneck
⌘/Ctrl + Enter

Flow past a cylinder

✅ Concept check

Why can no equation $y' = f(y, r)$ show vortex shedding (a steady state giving way to a periodic orbit)?

Exercise: continuation along stable and unstable branches

Implement natural-parameter continuation. It must stop when Newton fails (a fold) or the solution jumps to another branch.

🧪 Coding exerciseExercise: Newton continuation of the subcritical pitchfork
⌘/Ctrl + Enter

Problems

✏️ Pen & paperA pitchfork hiding in a sine●●○

Consider y′=ry−sin⁡yy' = r y - \sin y. (a) Find the bifurcation of y=0y = 0 and decide whether it is super- or subcritical. (b) Estimate the nonzero equilibria near it and their stability. (c) Where do the nonzero equilibria fold?

Show worked solution

(a) fy(0)=r−cos⁡0=r−1f_y(0) = r - \cos 0 = r - 1, so 00 is stable for r<1r \lt 1 and unstable for r>1r \gt 1. Expanding, f=(r−1)y+y3/6−y5/120+…f = (r-1)y + y^3/6 - y^5/120 + \dots; the cubic term at r=1r = 1 is +y3/6+y^3/6, pushing outward: subcritical.

(b) (r−1)+y2/6≈0(r-1) + y^2/6 \approx 0 gives y2≈6(1−r)y^2 \approx 6(1-r), for r<1r \lt 1, with fy=r−cos⁡y>0f_y = r - \cos y \gt 0: unstable. At r=0.9r = 0.9 solving sin⁡y=0.9 y\sin y = 0.9\,y gives y=0.7867y = 0.7867, y2=0.619y^2 = 0.619 (estimate 0.6000.600), fy=+0.194f_y = +0.194.

(c) Equilibria satisfy r=sin⁡y/yr = \sin y / y, which decreases until tan⁡y=y\tan y = y at y=4.4934y = 4.4934, where r=−0.2172r = -0.2172 and fy=(sin⁡y−ycos⁡y)/y=0f_y = (\sin y - y\cos y)/y = 0: a fold. Beyond it the branch is stable (y=5y = 5: r=−0.1918r = -0.1918, fy=−0.475f_y = -0.475). The same structure as the lesson’s subcritical pitchfork, with the fold to the left.

✏️ Pen & paperHow long can a stopping rule be fooled?●●○

A solver integrates y′=−(ε+y2)y' = -(\varepsilon + y^2) from y(0)=1y(0) = 1 and stops when ∣y′∣<tol=10−6|y'| \lt \text{tol} = 10^{-6}. (a) For which ε\varepsilon does the rule fire? (b) For ε=10−8\varepsilon = 10^{-8}, at which yy does it fire, and how long would the true solution need to reach y=−1y = -1 from there? Compare with π/ε\pi/\sqrt\varepsilon.

Show worked solution

(a) ∣y′∣=ε+y2|y'| = \varepsilon + y^2 has minimum ε\varepsilon at y=0y = 0, so the rule fires iff ε<10−6\varepsilon \lt 10^{-6}.

(b) It fires when y2<tol−εy^2 \lt \text{tol} - \varepsilon, at y1=9.95×10−4y_1 = 9.95 \times 10^{-4}. The remaining time is 1ε(arctan⁡y1ε+arctan⁡1ε)=104(1.4706+1.5707)=30 413\frac{1}{\sqrt\varepsilon}\big(\arctan\frac{y_1}{\sqrt\varepsilon} + \arctan\frac{1}{\sqrt\varepsilon}\big) = 10^4(1.4706 + 1.5707) = 30\,413, confirmed by adaptive integration (30 413.2530\,413.25); π/ε=31 416\pi/\sqrt\varepsilon = 31\,416. The “steady state” is about 3×1043 \times 10^4 time units from the end of the passage.

✏️ Pen & paperThe lag of a slow sweep●●●

Let r(t)=−1+ϵtr(t) = -1 + \epsilon t in y′=ry−y3y' = r y - y^3, ϵ=0.01\epsilon = 0.01, y(0)=10−6y(0) = 10^{-6}. (a) For small ∣y∣|y|, solve the linearised equation and find the rr at which ∣y∣|y| returns to 10−610^{-6}. (b) At which rr does ∣y∣|y| reach 0.10.1? (c) Repeat (b) for y(0)=10−3y(0) = 10^{-3} and say why the sim uses a noise floor.

Show worked solution

(a) ln⁡(y/y0)=∫r dt=(r2−1)/(2ϵ)\ln(y/y_0) = \int r\,dt = (r^2 - 1)/(2\epsilon), which is 00 at r=+1r = +1, however slow the sweep: the state decayed while r<0r \lt 0 and must climb back.

(b) (r2−1)/(2ϵ)=ln⁡(0.1/10−6)=11.51(r^2 - 1)/(2\epsilon) = \ln(0.1/10^{-6}) = 11.51 gives r=1+0.02×11.51=1.1092r = \sqrt{1 + 0.02 \times 11.51} = 1.1092; integrating the full equation gives 1.10921.1092.

(c) ln⁡100=4.605\ln 100 = 4.605 gives r=1.0450r = 1.0450 (integration: 1.04511.0451). The instability starts at r=0r = 0 but is invisible until r≈1r \approx 1: a delayed bifurcation. A noise-free sweep from deep in the stable region would stay at 00 far past the critical value, so the sim holds the state off 00; real systems have noise.

Key Takeaways

  • A bifurcation is an rr where the phase line of y′=f(y,r)y' = f(y, r) changes qualitatively; it needs fy=0f_y = 0, which is not sufficient (y′=r−y3y' = r - y^3). The bifurcation diagram stacks the phase lines: stable branches solid, unstable dashed.
  • Saddle-node y′=r−y2y' = r - y^2: ±r\pm\sqrt r for r≥0r \ge 0 merge and vanish; the harvest model is this with r=14−hr = \tfrac14 - h. Critical slowing down: τ=1/(2r)\tau = 1/(2\sqrt r), a warning that a fold is near, not a clock.
  • Transcritical ry−y2ry - y^2: 00 and rr swap stability. Pitchforks: supercritical ry−y3ry - y^3 has ±r\pm\sqrt r stable; subcritical ry+y3−y5ry + y^3 - y^5 has a stable large branch for r>−14r \gt -\tfrac14, left of the pitchfork.
  • Hysteresis: two stable states coexist over an interval of rr (jumps at 00 and −14-\tfrac14; cusp model at ±0.3849\pm 0.3849). A finite sweep lags; the lag closes as the sweep slows, the loop does not.
  • A tipping point is a fold whose jump is not undone by returning rr; these models are caricatures.
  • Time-stepping cannot follow unstable branches; continuation can. Near a fold the passage takes π/ε\pi/\sqrt\varepsilon and a ∣y′∣<tol|y'| \lt \text{tol} stopping rule can stop on the ghost.
  • Vortex shedding behind a cylinder (critical Reynolds number about 4747 in 2D) is a Hopf bifurcation: it needs two dimensions, see 7.9 Limit Cycles & Hopf Bifurcations.
Finished this lesson?

Mark it complete to track your progress (saved in this browser). Reading every section does this for you.

Where this leads

These later lessons build directly on this one: