127d0966584eb372.tex
1: \begin{proof} 
2: Let us show the three points required by \Cref{def:approx_seq} as follows.
3: 
4: i) Given an arbitrary $\varepsilon>0$, there is a common modulus of uniform continuity $\eta$ such that \eqref{eq:character_continu} holds for all the intervals $(\sigma_k,\sigma_{k+1})$. For $\nu$ large enough, one has, for each $\i \in\I\esnu$, 
5:  $\mu_\i\esnu< \eta$ so that for all $\th \in \Theta_\i\esnu$, $|\bar{\cutp}_{i}\esnu-\th|< \eta$; hence $d_H\big( \X_\th, \X_\i\esnu\big)=d_H( \X_\th, \X_{ \bar{\cutp}_{i}\esnu}) < \varepsilon$. 
6: 
7: ii) According to the continuity property, for all $(\xx,\yyag)\in \M^2$:
8: \begin{equation*}
9: \norm{  \deriv_1 f\esnu_\i\left(\txt {\mu_i\esnu } \xx, \yyag \right) \ -
10: \  \deriv_1 f_\th(\xx,\yyag) } = \norm{  \deriv_1 f_{ \bar{\cutp}_{i}\esnu}\Big(\xx , \yyag \Big) \ - \ \deriv_1 f_\th(\xx,\yyag)  }< \varepsilon . 
11: \end{equation*}
12: 
13: To be rigorous, one would  we need to ensure$\spa \X_\th$ to be the same for all $\th\in \Theta\esnu_i$: if not, one can further divide $\Theta\esnu_i$ into a finite number groups so that  players in each group have the same $\spa \X_\th$. This is possible because $\{\spa \X_\th\}_\th$ is finite. %
14: 
15: 
16: 
17: 
18: 
19: iii) By definition, $D\esnu=0$.
20: \end{proof}
21: