Delta link-homotopy on spatial graphs (Q1870108)

\textit{K. Taniyama} [Topology 33, 509-523 (1994; Zbl 0823.57006)] introduced eight different equivalence relations (1) ambient isotopy, (2) cobordism, (3) isotopy, (4) \(I\)-equivalence, (5) edge-homotopy, (6) vertex-homotopy, (7) homology, (8) \(\mathbb{Z}_2\)-homology on spatial graphs and found all implications between them that hold in the general case. Note that edge-homotopy and vertex-homotopy were introduced as natural generalizations of \textit{link homotopy} on links. In the given paper, the author defines two new equivalence relations on spatial graphs. Two spatial embeddings \(f,g\colon G\to S^3\) are (DEH) \textit{delta edge-homotopic} if \(f\) and \(g\) can be transformed into one another by self-delta moves and ambient isotopies, where a self-delta move is a delta move on an edge, and (DVH) \textit{delta vertex-homotopic} if \(f\) and \(g\) can be transformed into one another by quasi-adjacent delta moves and ambient isotopies, where a quasi-adjacent delta move is a delta move on exactly two adjacent edges. It is known that the delta move on knots is a local move which is an unknotting operation [\textit{H. Murakami} and \textit{Y. Nakanishi}, Math. Ann. 284, 75-89 (1989; Zbl 0646.57005)]. Delta edge-homotopy and delta vertex-homotopy on spatial graphs are considered by the author as natural generalizations of delta link-homotopy on links. The main result of this paper is Theorem 1.1 in which all the relations between (1), (2),\dots,(8) and (DEH), (DVH) that hold in general case are stated. Moreover these ten equivalence relations on spatial graphs are different. This provides a combinatorial criterion for any two spatial embeddings \(f,g\colon G\to S^3\) of a graph \(G\) to be delta vertex-homotopic. Let \(\Gamma (G)\) denote the set of all cycles of \(G\). For a given weight \(\omega\colon\Gamma (G)\to {\mathbb{Z}}_r\), one assigns to each spatial embedding \(f\colon G\to S^3\) the value \(\tilde{\alpha}_{\omega}(f)\) by the formula \(\tilde{\alpha}_{\omega}(f)\equiv \sum _{\gamma\in \Gamma (G)}\omega(\gamma)a_2(f(\gamma)) (\text{mod} r)\) where \(a_2(J)\) denotes the second coefficient of the Conway polynomial of a knot \(J\). Theorem 1.3 provides a sufficient condition for \(\tilde{\alpha}_{\omega}\) to be a delta edge-homotopy (delta vertex-homotopy) invariant of a spatial embedding of \(G\). By using the delta edge-homotopy invariant \(\tilde{\alpha}_{{\omega}_s}\), defined for a specific weight map \(\omega _s\colon\Gamma (G)\to {\mathbb Z}_{d_s}\), the author shows that there exists a non-trivial \(\theta\)-curve up to delta edge-homotopy. Using the weights \(\omega\colon\Gamma (G)\to {\mathbb Z}_r\), balanced on the edges (pairs of adjacent edges) of the graph \(G\), and the third derivative at \(1\) of the Jones polynomial of knots, the author defines in the similar manner the integral-valued delta edge-homotopy (delta vertex-homotopy) invariants \(n_{\omega}\) and studies them.