Color-preserving automorphisms in Fraisse classes (new version)

Again, a question closely related to a characterization of standard atoms (not to be confused with standard alphabets:). The question is a refinement of a question posted in Automorphisms vs color-preserving automorphisms in Fraisse classes, which has been answered negatively (see the recent post The colored open question closed).

Continue reading

A stronger Kleene theorem

This post continues the discussion on the Kleene theorem from here. The previous post gave a notion of regular expressions that had the same expressive power as nondeterministic automata. The idea was that a star corresponds to a graph where the set of vertices had one orbit, and where the edges where labelled by simpler regular expressions. In the case without atoms, such a graph is necessarily a self loop, which corresponds to the standard Kleene star. This post is about a stronger requirement: instead of saying that the graph has one orbit of vertices, we say that it has one orbit of edges.

Continue reading

Normalization of pp-formulas

Classicaly,  primitive positive formulas form the smallest fragment of first-order logic which is closed under conjunction and existential quantification. Pp-formulas can be turned in to a normal form \exists^\ast \bigwedge_i \phi_i – this is due to Skolemization, i.e. the fact that \bigwedge_i \exists x \phi_i can be turned into \exists x_1,\ldots,x_n \bigwedge \phi_i. In this case, Skolemization is a consequence of the axiom of finite choice.

We show that in the world with atoms,  not every orbit-finite pp-formula can be converted into an equivalent one in prenex form. This is another demonstration of the failure of the axiom of choice.

This raises the question: what is the appropriate notion of a pp-formula, which does have normal forms?

Continue reading