next up previous
Next: Conclusions Up: Proof for The Indirect Previous: Notations

   
Bounding the Number of Calls to PS(M)

We've seen:
For every $l_I>\frac{\ln{\varepsilon(1-\gamma)}}{\ln\gamma}$,

\begin{displaymath}\Pr( \left\Vert\widehat{v}_{l_I}-v^*\right\Vert \geq \varepsi...
...
\frac{\gamma^{l_I}}{1-\gamma})}{\gamma(1-\gamma^{l_I})}}^2}
\end{displaymath} (22)

Let's find mI and lI such that

\begin{displaymath}\Pr( \left\Vert\widehat{v}_{l_I}-v^*\right\Vert \geq \varepsilon) \leq \delta
\end{displaymath} (23)

By the previous claims, it suffices to find mI and lI s.t.:
1.
$l_I > \frac{\ln{(\varepsilon(1-\gamma))}}{\ln\gamma}$
2.
$\delta = l_I\cdot\ \vert S\vert\cdot \vert A\vert\cdot 2e^{-2\cdot m_I\cdot
(...
...repsilon - \frac{\gamma^{l_I}}{1-\gamma}) }
{ \gamma (1-\gamma^{l_I}) } })^2}$
Note that $log_{\gamma}{2} > 0$ (since $0<\gamma<1$). Then we can set (taking upper value if necessary):

  \begin{eqnarray*}
l_I &=& \frac{\ln{(\varepsilon(1-\gamma))}}{\ln\gamma} +
...
...
log_{\gamma}{2}\\
&=& \log_\gamma(2\varepsilon(1-\gamma))
\end{eqnarray*}


The number of calls to PS(M) is mI:


Here, again, mI and lI must be natural numbers, but taking their upper value won't change the asymptotic complexity, so, the number of calls to PS(M) by the indirect algorithm is:
$O(\frac{1}{\varepsilon^2} \cdot \ln{(\frac{\vert S\vert\cdot
\vert A\vert}{\delta})} + \frac{1}{\varepsilon^2} \cdot
\ln\ln\frac{1}{\varepsilon} )$


Yishay Mansour
2000-05-30