Directed diagrammatic reducibility (Q2215633): Difference between revisions

For context, we first review the main definition of interest in the paper. It is simple to reduce a word in a free group by cancelling adjacent generators with opposite sign. Thinking of the free group as a labeled graph (\(1\)-complex), we see that any trivial word (loop of edges labeled via a map into the free group) must have some edge adjacent to an edge with opposite label. If any word has this property, then actually the graph is a tree, and the group is trivial. The situation is more complicated when we consider the question of when \(\pi_2(K)=0\) for a \(2\)-complex \(K\). A combinatorial condition implying \(\pi_2(K)=0\) is diagrammatic reducibility: the requirement that any combinatorial map of a \(2\)-sphere into \(K\) has some edge at which the two incident \(2\)-cells map to the same \(2\)-cell in \(K\) with opposite orientation (a ``folding edge''). Directed diagrammatic reducibility is a relative version of diagrammatic reducibility, most easily stated when the \(K\) is a presentation complex of a group. If \(S\) is a subset of generators in a presentation \(P\), and \(P_S\) is the sub-presentation, and \(K(P), K(P_S)\) the associated two-complexes, then \(P\) is diagrammatically reducible directed away from \(S\) if every spherical diagram over \(K(P)\) that is not a diagram over \(K(P_S)\) contains a folding edge with a label in the complement of \(S\). In other words, any spherical diagram which maps ``outside'' of \(K(P_S)\) can be reduced. This paper develops the theory of directed diagrammatic reducibility and places it in the context of similar notions such as diagrammatic reducibility of relative presentations. Also, some applications are given, such as to small-cancellation groups. It should be of interest to anyone working in combinatorial group theory or related fields, as the discussion and arguments are accessible and motivated.