the Hopf bifurcation
/ HOPF (rhymes with 'top') /
All the bifurcations so far rearrange resting states. The Hopf bifurcation does something new: it gives birth to an oscillation. Picture a quiet steady state — a heart at rest, a chemical mixture sitting at fixed concentrations, an electronic circuit holding a constant voltage. As a parameter crosses a threshold the steady state goes unstable not by drifting off, but by starting to ring: a small, self-sustained oscillation springs up where there was none. A limit cycle is born.
The Hopf bifurcation lives in two or more dimensions, because oscillation needs room to circle. The tell-tale sign is in the linearization at the equilibrium: a complex-conjugate pair of eigenvalues alpha(r) +/- i beta(r) crosses the imaginary axis as the parameter r varies — the real part alpha changes sign from negative to positive while the imaginary part beta stays nonzero. Negative real part means inward spiral (decaying oscillation, stable); positive real part means outward spiral (growing oscillation, unstable); and right at alpha = 0 the spiral neither grows nor decays. Nonlinear terms then catch the growing spiral and bend it onto a closed loop of fixed amplitude — a limit cycle of frequency roughly beta. Supercritical Hopf gives a stable small cycle that grows like sqrt(r - r_c), a smooth onset of oscillation; subcritical Hopf is abrupt and can jump to a large-amplitude cycle with hysteresis.
This is the standard route by which a system that was steady starts to oscillate on its own: the onset of a heartbeat rhythm, the flutter of an aircraft wing, the spontaneous ticking of a chemical clock, the firing of a neuron. Because the linear analysis only signals the eigenvalues crossing, the deeper Hopf theorem is what guarantees an actual periodic orbit appears and tells you, via the sign of a 'first Lyapunov coefficient', whether the onset is the safe supercritical kind or the dangerous subcritical kind.
The Van der Pol oscillator x'' - mu (1 - x^2) x' + x = 0 has the origin as equilibrium. For mu < 0 it is a stable spiral (oscillations die out); as mu increases through 0 the eigenvalue pair crosses the imaginary axis and the origin becomes an unstable spiral, while a stable limit cycle grows around it. At mu = 0 the system undergoes a Hopf bifurcation and begins to oscillate on its own.
A stable spiral turns unstable and spawns a surrounding limit cycle — the standard birth of self-sustained oscillation.
Crossing eigenvalues are necessary but not sufficient to conclude a limit cycle: the Hopf theorem also needs the crossing to be transversal (nonzero speed) and the nonlinear terms to be generic. And the eigenvalue test tells you nothing about whether the cycle is stable — that requires the higher-order Lyapunov coefficient.