1 Let us first have a discussion on the stop criterion of the cited algorithm.
2 We claim that even if the variation of the dual function
3 (recalled in equation (\ref{eq:dualFunction}))
5 threshold, this does not ensure that the lifetime has been maximized.
6 Minimizing a function on a multiple domain (as the dual function)
7 may indeed easily fall into a local trap because some of introduced
8 variables may lead to uniformity of the output.
12 \includegraphics[scale=0.5]{amplrate.png}
14 \caption{Relations between dual function variation and convergence of all the $q_i$}
15 \label{fig:convergence:scatterplot}
18 To explain this, we introduce the maximum amplitude rate $\zeta$
19 of the sequence of $q$ which is defined as
21 \dfrac{\max_{i \in N} \{q_i\}}
22 {\min_{i \in N} \{q_i\}}-1.
24 The Figure~\ref{fig:convergence:scatterplot} presents
25 a scatter plot between $\zeta$, which
26 is represented in $y$-coordinate
28 value of the threshold for dual function that is represented in
32 Experiments shown that even if the dual
33 function seems to be constant
34 (variations between two evaluations of this one is less than $10^{-5}$)
35 not all the $q_i$ have the same value, \textit{i. e.}, $\zeta$ is still large.
36 For instance, even with a threshold set to $10^{-5}$ there still can be more than
37 45\% of differences between two $q_i$.
38 To summarize, a very small threshold is a necessary condition, but not
39 a sufficient criteria to observe convergence of $q_i$.
40 In the following, we consider the system are $\epsilon$-stable if both
41 maximum amplitude rate and the dual function are less than a threshold