Pinned equivalence relations (Q2428814)

In this very elegant paper the author answers a question of Kechris about pinned equivalence relations by showing that \(F_2\) is not the smallest unpinned equivalence relation. Recall that a Borel equivalence relation \(E\) on a Polish space \(X\) is \textit{unpinned} if there is a forcing notion \(P\) and a \(P\)-name \(\dot x\) for an element of \(X\) such that \(P\times P\) forces the left and the right evaluations of \(\dot x\) to be \(E\)-equivalent and \(P\) forces that \(\dot x\) is not \(E\)-equivalent to any ground model element of \(X\). An equivalence relation is \textit{pinned} if it is not unpinned. Recall also that \(F_2\) is the equivalence on \((2^\omega)^\omega\) with \((x_n:n<\omega)F_2(y_n:n<\omega)\) if \(\{x_n:n<\omega\}=\{y_n:n<\omega\}\). It is well known that \(F_2\) is unpinned and the class of unpinned equivalence relations is closed upwards in the Borel reducibility order. Kechris asked whether unpinned equivalence relations are exactly those which are above \(F_2\) in the Borel reducibility order. The author answers this question in the negative by showing that there is a Borel set \(Y\subseteq(2^\omega)^\omega\) such that \(F_2\restriction Y\) is unpinned and strictly below \(F_2\) in the Borel reducibility order. The set \(Y\) is defined as the set of those sequences in \((2^\omega)^\omega\) which form a clique in a certain Borel (\(F_\sigma\)) graph \(G\subseteq 2^\omega\). The graph \(G\) has the property that there is an uncountable clique in \(G\) but there is no clique of cardinality \(\aleph_2\) (no matter what the continuum is). Such a graph was constructed in ZFC by \textit{W. Kubiś} and \textit{S. Shelah} [``Analytic colorings'', Ann. Pure Appl. Logic 121, No. 2--3, 145--161 (2003; Zbl 1022.03026)] and a particularly simple construction was given by \textit{W. Kubiś} and \textit{B. Vejnar} [``Covering an uncountable square by countably many continuous functions'', Proc. Am. Math. Soc. 140, No. 12, 4359--4368 (2012)]. The point is that such a graph cannot have a perfect clique. The author proves that if \(G\) is a Borel graph on \(2^\omega\), \(Y\) is the set of those sequences in \((2^\omega)^\omega\) which form a clique in \(G\) and \(E_G=F_2\restriction Y\), then (i) the equivalence relation \(E_G\) is unpinned if and only if \(G\) has an uncountable clique (Claim 3.1), (ii) \(F_2\) is Borel-reducible to \(E_G\) if and only if \(G\) has a perfect clique (Claim 3.2).