this is for holding javascript data
Vadim Kosoy added Note_that_the_characterization_of__.tex
about 8 years ago
Commit id: 054c3a2f84aca6865a01aa080c4492c353dbc9de
deletions | additions
diff --git a/Note_that_the_characterization_of__.tex b/Note_that_the_characterization_of__.tex
new file mode 100644
index 0000000..9c902c2
--- /dev/null
+++ b/Note_that_the_characterization_of__.tex
...
Note that the characterization of $P$ depends not only on $f$ but also on $\mu$. That is, the accuracy of an estimator depends on the prior probabilities to encounter different questions. In general, we assume that the possible questions are indexed by elements of $\Words$. Thus we need to consider a probability distribution on $\Words$. However, in the spirit of average-complexity theory we will only require our estimators to be \emph{asymptotically} optimal. Therefore instead of considering a single probability distribution we consider a family of probability distribution indexed by an integer parameter, where the role of the parameter is defining the relevant limit. We thereby arrive at the following:
\begin{definition}
A \emph{word ensemble} $\mu$ is a family of probability distributions $\{\mu^k: \Words \rightarrow [0,1]\}_{k \in \Nats}$.
\end{definition}