fixed mistake

src/algorithm.tex

@@ -282,7 +282,7 @@ This results in an overall worst case performance of $2\cdot O(t^2)+t+1 = O(t^2)
  \Begin{
  	futureUses := empty map\tcp*{1}
  	replacements := empty map\tcp*{1}
- 	theoryGraph := sort(theoryGraph)\tcp*{$t^2$}
+ 	theoryGraph := sort(theoryGraph)\tcp*{$t^3$}
  	theoryGraphRev := reverse(theoryGraph)\tcp*{$t$}
  	\For{theory $\leftarrow$ theoryGraph\tcp*{$\times t$}}{
  		add theory $\rightarrow$ usedTheories(theory) $\cup$ futureUses(theory) to futureUses\tcp*{$t$}
@@ -298,7 +298,7 @@ This results in an overall worst case performance of $2\cdot O(t^2)+t+1 = O(t^2)
 
 The inner loop runs up to $t$ times over up to $t$.\\
 The outer loop runs up to $t$ times over up to $t+O(t^2+)+t^2$.\\
-This results in an overall worst case performance of $t^2+t+t\cdot (t+O(t^2)+t^2) +3= O(t^3)$.
+This results in an overall worst case performance of $t^3+t+t\cdot (t+O(t^2)+t^2) +3= O(t^3)$.
 
 Since the total number of theories can be quite large a cubic runtime is hardly ideal. However it should be noted that the worst case requires the average theory to include most of the other theories.