Merge pull request #67 from samiabobaker/paper-sdt-experiment-description

Clarify how SDT was tested in the paper

Author: samiabobaker

paper/paper.tex

@@ -566,7 +566,7 @@ \subsection{Reproducing known properties}
 
 \begin{itemize}
   \item
-    Five transformation functions: Ellis \cite{Ellis1989}, Ressel \cite{Ressel1996}, Imine \cite{Imine2003}, Suleiman \cite{Suleiman1997} and SDT \cite{Li2008} were all shown not to satisfy TP2 \cite{Oster2005Proving} (and some do not satisfy TP1). We combined each of these functions with the adOPTed algorithm, and all produced convergence violations for $\ge 3$ clients (which is when TP2 is relevant).
+    Five transformation functions: Ellis \cite{Ellis1989}, Ressel \cite{Ressel1996}, Imine \cite{Imine2003}, Suleiman \cite{Suleiman1997} and SDT \cite{Li2008} were all shown not to satisfy TP2 \cite{Oster2005Proving} (and some do not satisfy TP1). We combined the first four with the adOPTed algorithm, and all produced convergence violations for $\ge 3$ clients (which is when TP2 is relevant). We also reproduced convergence violations in SDT using its own control algorithm.
   \item
     The dOPT algorithm is known not to converge \cite{Sun1996}.
   \item