Flow Lab

Syllabus / Module 7: Systems of ODEs & Dynamical Systems

Lesson 7.9

Limit Cycles & Hopf Bifurcations

Self-sustained oscillations: van der Pol, Poincaré–Bendixson, Hopf bifurcation.

Written byClaude Sonnet

🌊
Why this matters for simulation: The onset of vortex shedding behind a cylinder is a Hopf bifurcation to a limit cycle.

By the end of this lesson you can…

  • Tell a limit cycle (an isolated closed orbit that nearby trajectories approach or leave) from the continuum of closed orbits around a center, and say which one an integrator can keep honest.
  • Convert a planar system to polar form, solve $r' = r(1 - r^2)$ exactly, and read stable and unstable cycles from the sign of $r'$.
  • State the Poincaré–Bendixson theorem (not proved), build a ring-shaped trapping region, and explain why chaos needs three dimensions; state Bendixson's criterion and apply it.
  • Measure the amplitude and period of the van der Pol cycle from zero crossings, and explain amplitude $\approx 2$ and the relaxation period $(3 - 2\ln 2)\mu$, including the fast-slow structure and the stiffness it creates.
  • Derive the Hopf normal form $r' = \mu r - r^3$, $\theta' = 1$, check that the radius is $\sqrt\mu$ and that the eigenvalues $\mu \pm i$ cross the imaginary axis, and distinguish supercritical from subcritical (hysteresis).
  • Quantify how the step size and the method change a measured period, and why a stable cycle absorbs numerical error instead of accumulating it.
On this page
·
  1. Closed orbits: a continuum versus an isolated loop
  2. Polar coordinates: a limit cycle you can solve
  3. Trapping regions and the Poincaré–Bendixson theorem
  4. The van der Pol oscillator
  5. The Hopf bifurcation
  6. Measuring a cycle on a computer
  7. Try it
  8. Computing it
  9. Concept checks
  10. Problems
  11. Key Takeaways

A frictionless pendulum swings forever at whatever amplitude you gave it. That is a center, and the last lesson showed that an integrator drifts off its level curve. A heartbeat, a violin string and the wake behind a cylinder do something else: their oscillation has one size, set by the system, and it recovers if you kick it. That is a limit cycle, and only a nonlinear system can have one. We solve one in polar coordinates, prove when one must exist, measure the van der Pol cycle, and watch a cycle being born as a parameter crosses a critical value: the two-dimensional cousin of 3.9 Bifurcations, and how vortex shedding starts.

▶ Video Trajectories from inside and outside spiral onto one closed orbit; as mu crosses zero the equilibrium releases a cyclecaptions only
Transcript
  1. A pendulum keeps whatever amplitude you gave it: a center, with a continuum of closed orbits. A limit cycle is an isolated closed orbit that nearby trajectories approach. Kick a center and the error stays for good; kick a stable cycle and the flow shrinks it.
  2. Polar coordinates give a cycle you can solve. For this system, r prime equals r times one minus r squared, and theta prime equals one. Every lap takes two pi. The radius grows inside r equals one and shrinks outside, so the unit circle attracts.
  3. The radius solves exactly with u equals r to the minus two, and RK4 agrees: from point two, r of three is 0.97152; from two, 1.000931. Reverse the sign and the circle repels. From 0.9 the radius is 0.00009 at t equals ten; from 1.1 it blows up at 0.8756.
  4. To show a cycle exists, build a trapping region. On r equals one half the radial speed is 0.375, inward; on r equals two it is minus six, also inward. No equilibrium in the ring, so Poincaré–Bendixson, stated, not proved, gives a closed orbit. No chaos in the plane: that needs three dimensions. A divergence of one sign forbids cycles.
  5. Van der Pol has negative damping for x between minus one and one. The origin has trace mu and determinant one: at mu equals one, 0.5 plus or minus 0.866 i; at five, 0.2087 and 4.7913. Everything leaves the origin and large orbits are pulled back, so a cycle exists.
  6. Why amplitude two? Energy obeys E prime equals mu times one minus x squared times y squared. Averaging over a lap gives a squared times one half minus a squared over eight, zero. So a equals two: pumping inside balances loss outside.
  7. Measured amplitudes: 2.0001, 2.0086, 2.0215, 2.0143, all within 1.1 percent of two. Periods: 6.2871, 6.6633, 11.6123, 19.0784, from two pi upward. The asymptote three minus two log two times mu is only a trend: at mu ten the period is 18 percent above it, at mu five, 44 percent.
  8. Large mu is fast and slow. With u equal to x prime over mu minus x plus x cubed over three, x races to the cubic and creeps along it. Creeping from x equals two to the fold takes mu times three halves minus log two. Then x jumps, and two creeps per lap give three minus two log two, times mu.
  9. That fast-slow structure is stiff: the fast eigenvalue is minus 3000 at x equals two when mu is 1000. RK45 takes 17977 steps, Radau 39, and both give x of twenty equal 1.986592. The mean RK45 step is the stability limit, not accuracy.
  10. The Hopf normal form: r prime equals mu r minus r cubed. The eigenvalues mu plus or minus i cross the imaginary axis. For mu positive, a cycle of radius root mu appears: measured 0.2000, 0.5000, 1.0000. Quadrupling mu doubles the radius.
  11. Flip the cubic's sign and add a quintic: subcritical. At mu equals minus 0.2 the origin is stable, with cycles at 0.52573, unstable, and 0.85065, stable. Seed 0.5 decays; 0.55 and 1.2 reach 0.85065. They merge at minus one quarter, radius 0.70711. Raising mu through zero the state jumps to a cycle of radius one: hysteresis.
  12. Why does a stable cycle absorb numerical error? A patch changes area by e to the integral of the divergence per lap. For van der Pol at mu one that integral is minus 7.052, a multiplier of 8.65 times ten to the minus four, matching the measured return map. At mu five it is about e to the minus 85.
  13. Vortex shedding is a Hopf bifurcation, near Reynolds 47, from memory, and numerical damping delta turns mu into mu minus delta. Recap: cycles are isolated and polar form solves them; Poincaré–Bendixson forbids planar chaos; van der Pol is relaxation and stiff; Hopf births cycles. Next: chaos in three dimensions.

Closed orbits: a continuum versus an isolated loop

A periodic orbit is a closed loop in the phase plane. In a conservative system (7.8 Conservative Systems & Lyapunov Functions) closed orbits come in a continuum, one per level curve. A limit cycle is an isolated closed orbit: nearby trajectories spiral toward it (stable) or away from it (unstable). For simulation that is the difference between accumulating and correcting errors: on an oscillator an error moves you to a neighbouring level curve for good, while on a stable cycle it is a distance from the loop that the flow shrinks. A consistent scheme ends on a slightly displaced cycle and stays there; we measure this below.

Polar coordinates: a limit cycle you can solve

With x=rcos⁡θ, y=rsin⁡θx = r\cos\theta,\ y = r\sin\theta,

r′=xx′+yy′r,θ′=xy′−yx′r2.r' = \frac{x x' + y y'}{r}, \qquad \theta' = \frac{x y' - y x'}{r^2}.

For x′=x−y−x(x2+y2), y′=x+y−y(x2+y2)x' = x - y - x(x^2 + y^2),\ y' = x + y - y(x^2+y^2) one finds xx′+yy′=r2−r4xx' + yy' = r^2 - r^4 and xy′−yx′=r2xy' - yx' = r^2, so

r′=r(1−r2),θ′=1.r' = r(1 - r^2), \qquad \theta' = 1.

The angle turns at unit speed, so every lap takes 2π2\pi. The radius obeys a one-variable equation (3.8 Autonomous Equations, Equilibria & Stability): r′>0r' \gt 0 inside r=1r = 1 and r′<0r' \lt 0 outside, so r=1r = 1 is stable and the unit circle is an attracting closed orbit. The origin repels. It also solves exactly: u=r−2u = r^{-2} gives u′=−2(u−1)u' = -2(u - 1), hence

r(t)=11+(r0−2−1)e−2t.r(t) = \frac{1}{\sqrt{1 + (r_0^{-2} - 1)e^{-2t}}}.

Lab block A, RK4 on the Cartesian system, h=0.01h = 0.01: from r0=0.2r_0 = 0.2 both RK4 and the formula give r(3)=0.97152r(3) = 0.97152; from r0=2r_0 = 2 both give 1.0009311.000931. Inside and outside converge from opposite sides.

An unstable cycle. Reverse the sign: r′=r(r2−1)r' = r(r^2 - 1) (the sim’s “unstable ring”). Now r=1r = 1 repels and the origin attracts: from r0=0.9r_0 = 0.9 the radius is 0.000090.00009 at t=10t = 10; from r0=1.1r_0 = 1.1 it blows up at t∗=12ln⁡11−r0−2=0.8756t^* = \tfrac12\ln\frac{1}{1 - r_0^{-2}} = 0.8756, and the sim stops with “BLEW UP” at t=0.890t = 0.890, a little late because RK4 only notices a singularity within a step of it. An unstable cycle is a basin boundary, and a forward run never lands on it.

Trapping regions and the Poincaré–Bendixson theorem

How do you know a cycle exists when you cannot solve? Build a trapping region. For the ring 0.5≤r≤20.5 \le r \le 2 above, the radial speed on the inner circle is r′=0.5(1−0.25)=0.375>0r' = 0.5(1 - 0.25) = 0.375 \gt 0 and on the outer circle r′=2(1−4)=−6<0r' = 2(1 - 4) = -6 \lt 0 (lab A; the sim’s bottom panel plots r′(r)r'(r)). The flow enters the ring on both sides, and the only equilibrium, the origin, is outside it.

So the ring must contain a cycle. The idea: a closed curve in the plane has an inside and an outside and orbits cannot cross (3.7 Existence & Uniqueness), so a trapped orbit can only wind onto a loop. Consequence: no chaos in the phase plane of an autonomous system. Chaos needs bounded trajectories that never settle, which needs room to cross over and under: three variables, as in the Lorenz equations of 7.10 Chaos & the Lorenz System. A forced 2D oscillator counts as three, since time is a variable.

No cycle: Bendixson’s criterion (from memory, one-line proof). If ∇⋅f\nabla\cdot\mathbf f has one strict sign on a simply connected region DD, no closed orbit lies in DD. On a closed orbit f\mathbf f is tangent, so the flux ∮f⋅n ds=0\oint\mathbf f\cdot\mathbf n\,ds = 0; Green’s theorem (2.10 Green, Stokes & the Divergence Theorem) says it equals ∬∇⋅f dA≠0\iint\nabla\cdot\mathbf f\,dA \ne 0. Example: x′=y, y′=−x−x3−yx' = y,\ y' = -x - x^3 - y has divergence −1-1, so no periodic orbit. The test is silent when the divergence changes sign (van der Pol: μ(1−x2)\mu(1 - x^2)) or vanishes (Hamiltonian flows).

The van der Pol oscillator

x′′−μ(1−x2)x′+x=0x'' - \mu(1 - x^2)x' + x = 0 with y=x′y = x' reads

x′=y,y′=μ(1−x2)y−x.x' = y, \qquad y' = \mu(1 - x^2)y - x.

The damping term has negative damping (energy input) for ∣x∣<1|x| \lt 1 and ordinary damping for ∣x∣>1|x| \gt 1. The Jacobian at the origin is (01−1μ)\begin{pmatrix}0&1\\-1&\mu\end{pmatrix}, trace μ\mu, determinant 11: eigenvalues 12(μ±μ2−4)\tfrac12(\mu \pm \sqrt{\mu^2 - 4}). That is an unstable spiral for μ<2\mu \lt 2 (μ=1\mu = 1: 0.5±0.866i0.5 \pm 0.866i; sim readout) and an unstable node for μ>2\mu \gt 2 (μ=5\mu = 5: 0.20870.2087 and 4.79134.7913). Everything leaves the origin and large trajectories are pulled back, so Poincaré–Bendixson gives a cycle (unique and stable by Liénard’s theorem; from memory).

Why the amplitude is about 2. For small μ\mu the motion is nearly x=acos⁡tx = a\cos t, and E=12(x2+y2)E = \tfrac12(x^2 + y^2) obeys E′=μ(1−x2)y2E' = \mu(1 - x^2)y^2. Over a lap the net change must vanish. Averaging (1−a2cos⁡2t) a2sin⁡2t(1 - a^2\cos^2t)\,a^2\sin^2t with ⟨sin⁡2⟩=12\langle\sin^2\rangle = \tfrac12 and ⟨cos⁡2sin⁡2⟩=18\langle\cos^2\sin^2\rangle = \tfrac18 gives a2(12−a28)=0a^2(\tfrac12 - \tfrac{a^2}8) = 0, so a=2a = 2: pumping inside ∣x∣<1|x| \lt 1 balances loss outside.

Measured (RK4, h=0.01h = 0.01, seed (0.1,0)(0.1, 0), last lap; lab block B and the sim readout):

μ\muamplitudeperiod2π2\pi(3−2ln⁡2)μ(3 - 2\ln 2)\mu
0.12.00016.28716.28320.1614
12.00866.66336.28321.6137
52.021511.61236.28328.0685
102.014319.07846.283216.1371

The amplitude stays within 1.1%1.1\% of 22 over two decades of μ\mu. The period is 2π2\pi at small μ\mu (the linear frequency Im λ≈1\mathrm{Im}\,\lambda \approx 1) and grows with μ\mu, but the large-μ\mu formula is an asymptotic trend: at μ=10\mu = 10 the measured period is 18%18\% above it (gap 2.9412.941) and at μ=5\mu = 5 it is 44%44\% above (gap 3.5443.544). The gap shrinks slowly (like μ−1/3\mu^{-1/3}; from memory), so quote the formula as a trend and measure when it matters.

Relaxation oscillations, and the origin of (3−2ln⁡2)μ(3 - 2\ln2)\mu. Put u=x′/μ−x+x3/3u = x'/\mu - x + x^3/3. The equation gives u′=−x/μu' = -x/\mu and x′=μ(u+x−x3/3)x' = \mu(u + x - x^3/3): for large μ\mu, xx is fast and uu is slow. So xx races to the curve u=x3/3−xu = x^3/3 - x and creeps along it, with (x2−1)x′=−x/μ(x^2 - 1)x' = -x/\mu, i.e. dt=−μx2−1xdxdt = -\mu\frac{x^2 - 1}{x}dx. Creeping from x=2x = 2 to the fold at x=1x = 1 takes

μ∫12x2−1x dx=μ[x22−ln⁡x]12=μ(32−ln⁡2),\mu\int_1^2\frac{x^2 - 1}{x}\,dx = \mu\Big[\tfrac{x^2}2 - \ln x\Big]_1^2 = \mu\big(\tfrac32 - \ln 2\big),

then xx jumps (time of order 1/μ1/\mu) to x=−2x = -2 (the line u=−23u = -\tfrac23 meets the curve again there) and creeps back. Two creeps per lap give (3−2ln⁡2)μ(3 - 2\ln 2)\mu, and the jump explains the amplitude 2.01432.0143 at μ=10\mu = 10.

Stiffness. On the slow curve the fast eigenvalue is μ(1−x2)\mu(1 - x^2), which is −3000-3000 at x=2x = 2 for μ=1000\mu = 1000: a quickly decaying mode with a slow solution riding on it, the definition of stiffness in 4.5 Stability Regions & Stiffness. Lab block G, μ=1000\mu = 1000 on [0,20][0, 20], rtol=10−6\texttt{rtol} = 10^{-6}: RK45 takes 1797717977 steps (125822125822 evaluations), Radau 3939 steps (284284 evaluations), and both give x(20)=1.986592x(20) = 1.986592. The mean RK45 step 1.11×10−31.11\times10^{-3} is essentially the stability limit 3.3066/3000=1.10×10−33.3066/3000 = 1.10\times10^{-3} of that lesson: accuracy did not set it (4.4 Adaptive Step Size & RK45).

The Hopf bifurcation

In 3.9 Bifurcations equilibria appeared, vanished or swapped stability, but nothing could oscillate. In two dimensions an equilibrium can lose stability to an oscillation. The normal form is

x′=μx−y−x(x2+y2),y′=x+μy−y(x2+y2),x' = \mu x - y - x(x^2 + y^2), \qquad y' = x + \mu y - y(x^2+y^2),

and the polar computation above gives r′=μr−r3, θ′=1r' = \mu r - r^3,\ \theta' = 1. The Jacobian at the origin, (μ−11μ)\begin{pmatrix}\mu&-1\\1&\mu\end{pmatrix}, has eigenvalues μ±i\mu \pm i. For μ<0\mu \lt 0 that is a stable spiral; at μ=0\mu = 0 the pair sits on the imaginary axis with imaginary part ±1\pm1, not 00 (that is what separates Hopf from the real crossing of a saddle-node or pitchfork); for μ>0\mu \gt 0 it is unstable, and r′=r(μ−r2)r' = r(\mu - r^2) has a stable root: a limit cycle of radius μ\sqrt\mu and period 2π2\pi. The radial equation is exactly the supercritical pitchfork of the bifurcations lesson (for r≥0r \ge 0), and the eigenvalue crosses at unit speed, d Re λ/dμ=1d\,\mathrm{Re}\,\lambda/d\mu = 1. At μ=0\mu = 0 this is the borderline case a=−1a = -1 of the last lesson: r′=−r3r' = -r^3 decays only like r0/1+2r02tr_0/\sqrt{1 + 2r_0^2t}, so transients near onset are very long.

Measured radius (lab C, sim; RK4, h=0.01h = 0.01): μ=0.04,0.25,1\mu = 0.04, 0.25, 1 give 0.2000,0.5000,1.00000.2000, 0.5000, 1.0000 to four decimals. The square-root law: quadrupling μ\mu from 0.250.25 to 11 multiplies the radius by 2.00002.0000, and a factor 6.256.25 (from 0.040.04 to 0.250.25) multiplies it by 2.5000=6.252.5000 = \sqrt{6.25}. The sim’s bottom panel sweeps μ\mu and plots measured radii on the parabola.

Supercritical versus subcritical. Flip the cubic’s sign and add a quintic to keep it bounded: r′=μr+r3−r5r' = \mu r + r^3 - r^5 (the sim’s “subcritical”). Cycles satisfy r2=12(1±1+4μ)r^2 = \tfrac12\big(1 \pm \sqrt{1 + 4\mu}\big). For μ>0\mu \gt 0 there is one stable cycle. For −14<μ<0-\tfrac14 \lt \mu \lt 0 there are two and the origin is also stable (eigenvalues −0.2±i-0.2 \pm i at μ=−0.2\mu = -0.2): the radii are 0.525730.52573 (unstable) and 0.850650.85065 (stable). In lab D (to t=1500t = 1500) the seed r0=0.5r_0 = 0.5 ends at 0.000000.00000, while 0.550.55 and 1.21.2 both end at 0.850650.85065. At μ=−14\mu = -\tfrac14 the cycles merge at r=0.70711r = 0.70711 and vanish. The consequence is hysteresis and a jump: as μ\mu rises through 00 the origin loses stability and the state leaps to a cycle that already has radius 11 at μ=0\mu = 0; lowering μ\mu again it stays on the large cycle until μ=−0.25\mu = -0.25. The sim’s sweep (small seed green, large seed blue) draws the loop. The supercritical form is gentle and reversible, the subcritical one abrupt (the hysteresis of the bifurcations lesson). That every Hopf bifurcation reduces to one of these forms, the sign of one coefficient deciding, is from memory.

Measuring a cycle on a computer

An upward zero crossing is a pair of samples with xi<0≤xi+1x_i \lt 0 \le x_{i+1}; linear interpolation puts it at ti+h xi/(xi−xi+1)t_i + h\,x_i/(x_i - x_{i+1}), with error of order h2h^2. The period is the gap between successive crossings, the amplitude the largest xx between them. The sim reads the last two laps and prints “not settled” unless the lap before agrees to 10−410^{-4}. For μ=0.1\mu = 0.1 this matters: Re λ=0.05\mathrm{Re}\,\lambda = 0.05, so a seed at 0.10.1 needs about ln⁡20/0.05=60\ln 20/0.05 = 60 time units merely to reach the cycle (the sim runs to T=300T = 300).

Step size and method (lab E, μ=1\mu = 1, T=100T = 100; reference RK4 with h=0.001h = 0.001: amplitude 2.008622.00862, period 6.663296.66329):

method, hhamplitudeperiod error
RK4, 0.010.012.00862+0.00000+0.00000
Euler, 0.010.012.02779+0.04980+0.04980
Euler, 0.0010.0012.01052+0.00488+0.00488

RK4 at h=0.01h = 0.01 matches the reference to five decimals. Euler is first order and shows it: a tenfold smaller step cuts the error by 10.210.2. The step size bites harder for large μ\mu (sim, μ=10\mu = 10, seed 1 = (0.1,0)(0.1, 0); the outside seed (3,0)(3, 0) is already unstable at h=0.07h = 0.07, where ∣λ∣≈80|\lambda| \approx 80): h=0.07h = 0.07 gives period 19.32019.320 (+1.3%+1.3\% over 19.07819.078), h=0.08h = 0.08 no longer settles, and h=0.085h = 0.085 blows up at t=48.9t = 48.9. The RK4 limit 2.7853/30=0.0932.7853/30 = 0.093 of the stiffness lesson (fast eigenvalue about −30-30) is the right range; the observed threshold is a little lower.

The cycle does not accumulate error. Euler at h=0.01h = 0.01 moved the cycle (amplitude 2.0282.028, period +0.050+0.050) but lap 33 and the last lap both take 6.71316.7131: nothing compounds. The reason is contraction. By Liouville’s formula (the linear case was det⁡eAt=eTt\det e^{At} = e^{Tt} in 7.5 The Matrix Exponential & Fundamental Matrices) a patch around the orbit changes area by exp⁡∮∇⋅f dt\exp\oint\nabla\cdot\mathbf f\,dt per lap; one multiplier is 11 (a shift along the orbit only moves the phase; from memory), so the other is e∮∇⋅f dte^{\oint\nabla\cdot\mathbf f\,dt}. For van der Pol μ=1\mu = 1 the divergence is 1−x21 - x^2, the lap integral is −7.052-7.052 and the multiplier 8.65×10−48.65\times10^{-4} (lab F and the sim, seed 1). Check: nudge the orbit by 0.010.01 on the line x=0x = 0 and run one lap; the return map has slope 8.73×10−48.73\times10^{-4}, within 1%1\%. At μ=5\mu = 5 the sim reports about e−85e^{-85}: one lap erases a disturbance to rounding error. For the Hopf form the lap integral is −4πμ-4\pi\mu: the sim reads −3.145-3.145 at μ=0.25\mu = 0.25 against −3.1416-3.1416 (the lap is rounded to the grid), a factor 0.04320.0432 per lap.

Try it

◆ Interactive Seeds converge to the cycle: van der Pol, Hopf, subcritical Hopf and the polar rings, with mu and h sliders and measured amplitude and period

Loading simulation…

What to try. (1) Van der Pol: μ=0.1,1,5,10\mu = 0.1, 1, 5, 10, reading the period against the dashed asymptotes in the bottom panel. (2) Hopf: μ=0.25\mu = 0.25 (radius 0.50.5), μ=1\mu = 1, μ=0.04\mu = 0.04, then μ<0\mu \lt 0 and watch every seed decay. (3) Subcritical at μ=−0.20\mu = -0.20: click inside and outside the unstable cycle. (4) Unstable ring: seeds 0.90.9 and 1.11.1. (5) Van der Pol μ=10\mu = 10: raise hh to 0.070.07, 0.080.08, 0.0850.085.

Computing it

🐍 PythonLab: polar solution, van der Pol and Hopf measurements, hysteresis, step size and method, lap multiplier, stiffness (blocks A to G)
⌘/Ctrl + Enter

Now build the measuring tool yourself: an integrator, a crossing detector with interpolation, and the cycle’s amplitude and period.

🧪 Coding exerciseExercise: measure the van der Pol limit cycle (amplitude and period) from zero crossings
⌘/Ctrl + Enter

Concept checks

✅ Concept check

For $r' = r(1 - r^2)$, what is $r'$ on the circle $r = 0.5$?

✅ Concept check

A planar autonomous system has a ring-shaped region $R$ with the flow pointing into $R$ on both boundary circles and no equilibrium in $R$. What follows?

✅ Concept check

In the Hopf normal form with $\mu \gt 0$, by what factor does the cycle radius change when $\mu$ is multiplied by $4$ (lab block C: $\mu = 0.25 \to 1$)?

✅ Concept check

Van der Pol at $\mu = 10$ has measured period $19.08$, while $(3 - 2\ln2)\mu = 16.14$. What is the honest reading?

Problems

✏️ Pen & paperSolve a radial equation●○○

For r′=r(1−r2)r' = r(1 - r^2) with r0=2r_0 = 2, use u=r−2u = r^{-2} to find r(t)r(t) and evaluate r(3)r(3).

Show worked solution

u′=−2r−3r′=−2(r−2−1)=−2(u−1)u' = -2r^{-3}r' = -2(r^{-2} - 1) = -2(u - 1), so u=1+(u0−1)e−2tu = 1 + (u_0 - 1)e^{-2t} with u0=14u_0 = \tfrac14, and r(3)=(1−0.75e−6)−1/2=1.000931r(3) = (1 - 0.75e^{-6})^{-1/2} = 1.000931. This matches RK4 in lab block A. The solution decreases to 11 from above without crossing it, as the sign of r′r' predicts.

✏️ Pen & paperA Hopf cycle at mu = 0.09●●○

For the supercritical normal form with μ=0.09\mu = 0.09 give the eigenvalues at the origin, the cycle’s radius and period, and the factor by which a small distance from the cycle shrinks per lap.

Show worked solution

Eigenvalues 0.09±i0.09 \pm i: an unstable spiral. r′=r(0.09−r2)r' = r(0.09 - r^2) has the stable root r=0.3r = 0.3, and the period is 2π2\pi. The derivative of μr−r3\mu r - r^3 at r=0.3r = 0.3 is μ−3μ=−0.18\mu - 3\mu = -0.18, so a lap multiplies a small distance by e−0.18⋅2π=0.3227e^{-0.18\cdot2\pi} = 0.3227 (by hand; the sim’s prediction line prints the same formula).

✏️ Pen & paperBirth of two cycles●●●

For r′=μr+r3−r5r' = \mu r + r^3 - r^5 find when cycles exist, their radii at μ=−0.2\mu = -0.2, and their stability.

Show worked solution

Cycles satisfy μ+r2−r4=0\mu + r^2 - r^4 = 0, i.e. s=r2=12(1±1+4μ)s = r^2 = \tfrac12(1 \pm \sqrt{1 + 4\mu}): two positive roots for −14<μ<0-\tfrac14 \lt \mu \lt 0, one for μ>0\mu \gt 0, merging at μ=−14\mu = -\tfrac14 (r=0.70711r = 0.70711). At μ=−0.2\mu = -0.2: r2=0.27639r^2 = 0.27639 and 0.723610.72361, so r=0.52573r = 0.52573 and 0.850650.85065 (lab D). With g′=μ+3r2−5r4g' = \mu + 3r^2 - 5r^4: at 0.525730.52573, g′=0.2472>0g' = 0.2472 \gt 0, unstable; at 0.850650.85065, g′=−0.6472<0g' = -0.6472 \lt 0, stable. The origin (g′(0)=μ=−0.2g'(0) = \mu = -0.2) is stable, so the inner cycle separates two basins: lab seeds 0.50.5 and 0.550.55 end on opposite sides. (The g′g' values are by hand.)

Key Takeaways

  • A limit cycle is an isolated closed orbit; a center has a continuum. A stable cycle corrects errors, so a consistent scheme shifts it but does not accumulate error (Euler at h=0.01h = 0.01: period error +0.050+0.050, identical on lap 33 and the last lap).
  • Polar form turns x′=x−y−xr2x' = x - y - xr^2, y′=x+y−yr2y' = x + y - yr^2 into r′=r(1−r2)r' = r(1 - r^2), θ′=1\theta' = 1: unit circle, period 2π2\pi, exact solution via u=r−2u = r^{-2} (r(3)=1.000931r(3) = 1.000931 from r0=2r_0 = 2, RK4 and formula). The sign of r′r' decides attraction; r′=r(r2−1)r' = r(r^2 - 1) is an unstable cycle.
  • Poincaré–Bendixson (stated): trapped in a bounded region with no equilibrium means approaching a closed orbit, hence no chaos in the plane. Bendixson (stated, via Green): a divergence of one sign forbids cycles.
  • Van der Pol: amplitude about 22 (measured 2.0001,2.0086,2.0215,2.01432.0001, 2.0086, 2.0215, 2.0143 at μ=0.1,1,5,10\mu = 0.1, 1, 5, 10); period 19.0819.08 at μ=10\mu = 10 against the trend 16.1416.14. Large μ\mu is fast-slow and stiff: RK45 1797717977 steps, Radau 3939 at μ=1000\mu = 1000.
  • Hopf: eigenvalues μ±i\mu \pm i cross the imaginary axis; the supercritical cycle has radius μ\sqrt\mu (measured 0.2,0.5,1.00.2, 0.5, 1.0; quadrupling μ\mu doubles it). Subcritical at μ=−0.2\mu = -0.2: stable origin and stable cycle (0.850650.85065) split by an unstable one (0.525730.52573), hence hysteresis.
  • Vortex shedding onset (about Re=47\mathrm{Re} = 47, from memory) is a Hopf bifurcation; uniform numerical damping δ\delta shifts μ\mu to μ−δ\mu - \delta. Next, three dimensions and chaos: 7.10 Chaos & the Lorenz System.
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: