I am reading the following paper: http://ieeexplore.ieee.org/document/7798534/

For the equation (5):
$$\dot{a}=(1-a_i^2)a_i-\beta\sum_{k-i}(a_k\cos(\theta_k-\theta_i)-a_i)$$ $$\dot{\theta}_i=\beta\sum_{k-i}\frac{a_k}{a_i}\sin(\theta_i-\theta_k)$$

  1. Both are dynamics for amplitude $a$ and phase $\theta$.
  2. $k-i$ means there is an edge linking the node $k$ and $i$.
  3. $\beta$ is just a coefficient

My question is the following:

enter image description here

  1. Why the amplitude dynamics are invariant?

and the following:

enter image description here

  1. Why it is "global phase symmetry"
  2. Why the equilibria are semi-stable with respect to rotation?


  1. If you add an $i$-independent time-dependent term $\delta\theta=\omega t+\gamma$ to $\theta_i$ then the difference $\theta_k-\theta_i$ that appears in the amplitude equation is unchanged. The phase equation is changed by a term $d\delta\theta/dt=\omega$.
  2. Global phase symmetry means that if $\psi_n=a_ne^{i\theta_n}$ is a solution, then also $\psi_n e^{i\phi t}$ with a constant phase $\phi$ is a solution. A local phase symmetry would have a $\phi_n$ that depends on $n$, which does not appply here.
  3. The limit cycle is semi-stable, rather than stable, because only amplitude perturbations decay in time, phase perturbations do not.
| cite | improve this answer | |
  • $\begingroup$ To 1., can I use the same argument to say that the phase equation is unchanged and so phase dynamic is invariant? I am confused about 3., how to view it? Thanks! $\endgroup$ – sleeve chen Nov 3 '17 at 21:10
  • $\begingroup$ I clarified 1; concerning 3, the typical picture is the mexican hat --- amplitude perturbations are radial and damped, while phase fluctuations are azimuthal and undamped. $\endgroup$ – Carlo Beenakker Nov 3 '17 at 22:07
  • $\begingroup$ To 1., I am still confused that for amplitude eq it is $\theta_k-\theta_i$ and for the phase eq it is $\theta_i-\theta_k$. Why one is unchanged and the other one is changed by $\omega$? To 3., how to obtain the picture you mentioned? (By Taylor expansion which gives the linear equation for perturbation and then plot it?) Sorry for many questions. Thanks! $\endgroup$ – sleeve chen Nov 4 '17 at 2:07
  • $\begingroup$ it has nothing to do with the order $\theta_k-\theta_i$ or $\theta_i-\theta_k$; it just that $\delta\theta=\omega t+\gamma$ depends on time, so $(d/dt)(\theta_i+\delta\theta)=\dot{\theta}_i+\omega$; so the phase equation gets an offset $\omega$ which the amplitude equation does not. $\endgroup$ – Carlo Beenakker Nov 4 '17 at 10:02
  • $\begingroup$ But based on your argument, in the amplitude equation the value of the term $\cos(\theta_k-\theta_i-\delta \theta)$ will also be changed due to $\delta \theta$. $\endgroup$ – sleeve chen Nov 4 '17 at 10:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.