The efficiency of cre-loxp system is extremely vital to our design; it may fluctuate a lot in that the DNA are reversible.
The existing period of arabinose and the concentration of arabinose are two main variables, which we can easily control in our wetlab experiments. We find by modeling prediction that the concentration of arabinose doesn't matter much while the time variable is of importance.
Based on our model, we designed some experiments to test the inversion efficiency. This is the result we got with flow cytometor:

In terms of the overall tendency and some critical points, the result is well in accordance with the modeling prediction: the concentration of arabinose doesn't matter while the time variable is important. The result we got by further detailed analysis, to some extent, verify our model. Then to simplify the experiment procedure with a similar effect we add a feed forward control to the design.