Fig. \ref{fig:type1b_pp2} shows MPPs from the stable fixed point to the saddle (red) and from the stable limit cycle to the saddle (blue). In this case, we have that \(\mu > 1\), where \(\mu\) is the eigenvalue ratio at the saddle. Hence the MPEP is not tangent to the threshold as in the previous example. Like the previous example, MPPs that start at the stable fixed points reach the threshold below the saddle only, while MPPs that start from the limit cycle (light blue) cross the threshold above the saddle only. Notice that there is an effective reflecting barrier near the saddle along \(w = 0\) so that all of the MPPs that reach the threshold from the stable fixed point are very close. We can expect that the exit behavior of the initiation event is nearly one dimensional as \(w\) is approximately fixed along the red curve.

The quasipotential is shown in Fig. \ref{fig:type1b_qp}. The front view (Fig. \ref{fig:type1b_qp}(inset)) shows that the quasipotential has the profile of a double well potential, with a local minimum of the left potential well at the stable fixed point and a local maximum at the saddle. However, the profile appears flat at the bottom of the right potential well, corresponding to the excited state, because of the stable limit cycle. Notice that the left well is very thin (with respect to \(w\)), confirming that \(w\) is approximately constant during the initiation phase.