Delta edge-homotopy invariants of spatial graphs via disk-summing the constituent knots (Q733338)

The author works in the piecewise linear category. Let \(G\) be a finite graph. An embedding of \(G\) into the 3-sphere is called a spatial embedding of \(G\) or a spatial graph. A graph \(G\) is said to be a planar if there exists an embedding of \(G\) into the 2-sphere, and a spatial embedding of a planar graph \(G\) is said trivial if it is ambient isotopic to an embedding of \(G\) into the 2-sphere in the 3-sphere. A trivial spatial embedding of a planar graph is unique up to ambient isotopy [cf. \textit{W. K. Mason}, Trans. Am. Math. Soc. 142, 269--290 (1969; Zbl 0183.51403)]. A delta move is a local deformation on a spatial graph which is known to be an unknotting operation [cf. \textit{S. V. Matveev}, Mat. Zametki 42, No. 2, 268--278 (1987; Zbl 0634.57006) and \textit{H. Murakami} and \textit{Y. Nakanishi}, Math. Ann. 284, No. 1, 75--89 (1989; Zbl 0646.57005)]. A delta move is called a self delta move if all three strings in the move belong to the same spatial edge. Two spatial embeddings of a graph are said to be delta edge-homotopic if they can be transformed into each other by self delta moves and ambient isotopies [cf. the author, Rev. Mat. Complut. 15, No. 2, 543--570 (2002; Zbl 1024.57005)]. If the graph is homeomorphic to the disjoint union of 1-spheres, then this equivalence relation coincides with self \(\Delta\)-equivalence [cf. \textit{T. Shibuya}, Osaka J. Math. 37, No. 1, 37--55 (2000; Zbl 0961.57002)] or delta link homotopy [\textit{Y. Nakanishi}, Toplogy Appl. 121, No. 1--2, 169--182 (2002; Zbl 1001.57013)] on oriented links. For self \(\Delta\)-equivalence on oriented links, \textit{T. Shibuya} proposed the following conjectures in [Osaka J. Math. 33, No. 3, 751--760 (1996; Zbl 0940.57009)] and [loc. cit.]. Conjecture 1.1. Two cobordant oriented links are self \(\Delta\)-equivalent. Conjecture 1.2. Any boundary link is self \(\Delta\)-equivalent to the trivial link. To the Conjecture 1.1, \textit{Y. Nakanishi} and \textit{T. Shibuya} gave a negative answer [J. Knot Theory Ramifications 9, No. 5, 683--691 (2000; Zbl 0998.57022)], and Conjecture 1.2 was solved affirmatively by \textit{T. Shibuya} and \textit{A. Yasuhara} [Math. Proc. Camb. Philos. Soc. 143, No. 2, 449--458 (2007; Zbl 1130.57009)]. On the outcome of these results, in the present paper the author investigates a more general case. A spatial embedding of a planar graph is said to be slice if it is cobordant to the trivial spatial embedding (the definition of spatial graph-cobordism is specified by \textit{K. Taniyama} in [Topology 33, No. 3, 509--523 (1994; Zbl 0823.57006)]). A spatial embedding of a graph is called a \(\partial\)-spatial embedding if all knots in the embedding bound Seifert surfaces simultaneously such that the interiors of the surfaces are mutually disjoint and disjoint from the image of the embedding [cf. the author and \textit{R. Shinjo}, Q. J. Math. 56, No. 2, 239--249 (2005; Zbl 1079.05028)]. If the graph is homeomorphic to the disjoint union of 1-spheres, then this definition coincides with the definition of a boundary link. Any nonplanar graph does not have a \(\partial\)-spatial embedding [cf. the author and R. Shinjo, op. cit.]. Then the author asks the following questions. Question 1.3. (1) Is any slice spatial embedding of a planar graph delta edge-homotopic to the trivial spatial embedding? (2) Is any \(\partial\)-spatial embedding of a graph delta edge-homotopic to the trivial spatial embedding? In fact, for spatial theta curves, the affirmative answers to Question 1.3 (1) and (2) have already been given by the author in [Math. Proc. Camb. Philos. Soc. 138, No. 3, 401--420 (2005; Zbl 1072.57003)]. But the author's purpose in the present paper is to give negative answers to the Questions 1. 3 (1) and (2), as follows. Theorem 1.4. (1) There exist infinitely many slice spatial embeddings of a graph up to delta edge-homotopy. (2) There exist infinitely many \(\partial\)-spatial embeddings of a graph up to delta edge-homotopy. To accomplish this, the author constructs some invariants of spatial graphs by considering a disk-summing operation among the constituent knots in a spatial graph and shows the delta edge-homotopy invariance of them. In the last section of the paper the author gives some remarkable examples which imply Theorem 1.4.