$$TC(R, x,y): \forall X \{ \text{ X is R-closed} \wedge x \in X \implies y \in X \}$$ where $\text{ X is R-closed }$ is defined by : $$\forall u,v (u \in X \wedge R(u,v) \implies v \in X) $$
It comes from http://www.labri.fr/perso/courcell/Conferences/ExpoLATA-2.pdf (fifth page)
I cannot understand it. From my point of view it doesn't work.
- $\forall X$ means "for every subgraph". I cannot imagine a graph which satisfies a such formula. Just taking not connected vertexes ( it is a subgraph) is contrexample- it doesn't satisfy "X is closed".
So, if we consider $\phi(x,y) = \forall X (X \text { is closed on R } \wedge x \in X) \implies y \in X$
Now, we can say that $(\mathbb{N}, R, y:e, x:0) \models \phi(x, y)$, where $R$ is interpreted: $R(x,y) \iff x = y +2$, is true iff $e$ is even, right?