As soon as , the first mode become unstable. The theory of the
**linear stability** had us predicted this behaviour.
But in this domain the solution of the KSE don't keep a weak
magnitude. So the there is a time when the linear stability can not be
amplied.
To conclude, the prediction of the final behaviour is not predicted by the
linear stability.
A study of unlinear stability is really hard to develop. That's why, I
propose a not exhaustive list of the bifurcations that I have meet in the
experimental study.

I have let simulation turn to see if the solution converged or not. I made this experiment for different values of . For , the solution converged toward a function similar to this one which is represented on figure 3.1.

To prouve that the solution is effectively converged, I have plot the norm of the solution. The norm that I have used, is :

The graph obtained is on the figure 3.2