diff --git a/appunti/appunti-luca-borghi/appunti-luca-borghi.pdf b/appunti/appunti-luca-borghi/appunti-luca-borghi.pdf
index cb33eac..bcc3881 100644
Binary files a/appunti/appunti-luca-borghi/appunti-luca-borghi.pdf and b/appunti/appunti-luca-borghi/appunti-luca-borghi.pdf differ
diff --git a/appunti/appunti-luca-borghi/appunti-luca-borghi.tex b/appunti/appunti-luca-borghi/appunti-luca-borghi.tex
index c6dfa7b..db43217 100644
--- a/appunti/appunti-luca-borghi/appunti-luca-borghi.tex
+++ b/appunti/appunti-luca-borghi/appunti-luca-borghi.tex
@@ -389,7 +389,7 @@ \subsubsection{Bounds consistency}
 first program, we specify (or better, we suggest) to use domain propagation,
 namely we keep $GAC$. Results are these:
 
-\includegraphics[scale=0.2]{sudoku_res.png}
+\includegraphics[scale=0.2]{sudoku-res.png}
 
 Just 3 seconds, not bad! But when we try to use bounds propagation (namely $BC$)
 or even not to suggest the propagation way (the third case), the results is that
@@ -972,7 +972,7 @@ \subsubsection{Population-based methods}
     is parametric) and then choose the value $v_i \in D(X_i)$ according to the
     probability:
 
-\includegraphics[scale=0.2]{prob_ants.png}    
+\includegraphics[scale=0.2]{prob-ants.png}    
 
     where $\alpha$ and $\beta$ are parameters used to balance the pheromone and
     heuristic factors.
@@ -1012,7 +1012,7 @@ \subsubsection{Population-based methods}
     \item fix part of the variables value;
     \item relax the remaining variables.
 \end{itemize}
-\includegraphics[scale=0.25]{fix_relax.png}
+\includegraphics[scale=0.25]{fix-relax.png}
 
 The good fact about this neighbourhood structure is that it works with all
 problems. Now, two issues are: which percentage of variables to fix? Which