Limit cycle

A limit cycle is an isolated periodic orbit of a dynamical system. It is attracting if nearby orbits approach it in forward time. A continuous family of closed orbits around a center equilibrium does not consist of limit cycles, because its members are not isolated.

Near a hyperbolic attracting limit cycle, a weak perturbation can be projected onto the neutral phase direction while transverse perturbations decay. Solving the periodic adjoint variational equation gives a phase response function. Averaging its inner product with weak forcing or coupling yields the first-order phase dynamics.

Linearizing weakly coupled locked phases gives $0=\xi -\Omega 1-kL\theta$, where $L$ is the directed graph Laplacian. Multiplying by a normalized left null vector ${\pi }^{\mathsf{T}}$ eliminates coupling and gives the displayed weighted frequency. In a balanced network, including symmetric reciprocal coupling, $\pi =1/N$ and the frequency is the arithmetic mean. A zero diagonal in the coupling matrix alone does not imply balance. For two directed weights $p,q>0$, the weights are $(q,p)/(p+q)$.

A periodic phase response function describes the first-order phase shift produced by a perturbation along a limit cycle. With period-one orbit $U$, choose the periodic adjoint solution normalized by ${\int }_0^{1}Y\cdot U^{\prime }ds=1$. The Fredholm solvability condition then projects forcing $p$ to ${\int }_0^{1}Y\cdot pds$. For voltage-difference coupling, the averaged interaction is $h(\varphi )={\int }_0^1{Y}^1(s)U^1(s+\varphi )ds$.

For an autonomous ordinary differential equation, the tangent to a periodic orbit gives a unit Floquet multiplier. A hyperbolic attracting cycle has that multiplier simple and all transverse multipliers strictly inside the unit disk. Thus the periodic variational operator has a one-dimensional null space spanned by the tangent. Nonlinear asymptotic attraction without hyperbolicity does not suffice: $\dot{r}=-(r-1{)}^3$, $\dot{\phi }=2\pi$ attracts to a unit-period circle but its radial variational equation is neutral.

The planar system $\dot{x}=y+(1-x^2-y^2)x$, $\dot{y}=-x+(1-x^2-y^2)y$ has polar coordinates $\dot{r}=r(1-r^2)$, $\dot{\theta }=-1$. Every nonzero orbit approaches the clockwise unit-circle limit cycle, of period $2\pi$. For $r(t_0)=r_0>0$,Interior orbits approach the origin backward in time, but exterior orbits have backward finite-time blow-up of an ordinary differential equation at $t=t_0+\frac{1}{2}\log (1-r_0^{-2})$. Thus forward convergence to a limit cycle does not imply existence for all negative times.

On the punctured plane, the vector field $\dot{x}=(1/r-1)x-y$, $\dot{y}=(1/r-1)y+x$ has the displayed polar equations. Its exact trajectories satisfy $r(t)=1+(r(0)-1)e^{-t}$ and $\theta (t)=\theta (0)+t$. Every forward trajectory approaches the unit circle, rotating counterclockwise, from inside or outside according to its initial radius. The circle is an attracting limit cycle with least period $2\pi$. The origin is not an equilibrium: the original vector field is undefined there. The singularity prevents an application of a whole-plane Bendixson-Dulac criterion to this periodic orbit.