deletions | additions       diff --git a/Deterministic and stochastic Arnold tongue .tex b/Deterministic and stochastic Arnold tongue .tex  index c71e3c1..40864dd 100644  --- a/Deterministic and stochastic Arnold tongue .tex  +++ b/Deterministic and stochastic Arnold tongue .tex 
...
\subsection{Deterministic and 2:1 mode locking}  To investigate whether our system can adopt different mode locked state, we drew the Arnold tongues of the deterministic model. The presence of synchronized regions (black, Figure \ref{fig:arnold}A) centered around each rational number $T_2/T_1 = p/q$ shows that this is the case, also reflecting the pulsed (i.e. localized in the phase plane) nature of the inferred coupling.  To test if cells could adopt different mode locked states, we estimated the $2:1$ attractor from the deterministic model (blue, Figure \ref{fig:arnold}B) and computed for each trace the average distance to the attractor. We found that a fraction of the cell cells  closely follow the $2:1$ attractor, as illustrated in Figure \ref{fig:arnold}B-D. In total about $7\%$ of the cell traces were closer to the $2:1$ attractor than to the $1:1$ (Section \ref{sec:distanceToAttractor}). This indicates that cells with long cell cycle durations can adopt a $2:1$ mode locked state where the cell divides every two circadian peaks. However, the length of our recordings and the stochastic nature of the processes make difficult to distinguish between more complicated scenario (e.g. $3:2$ vs $1:1$).