On uniqueness of end sums and 1-handles at infinity (Q2311439)

The paper under review presents a detailed study of uniqueness of the operation of \textit{end summing}. Many examples are given in all categories (TOP, PL and DIFF) where uniqueness fails for different reasons. On the other hand, general hypotheses are given in order to assure the operation is unique and well-defined in all categories. The classical operations of connected sum (for closed manifolds) and boundary sum (of compact manifolds with boundary) are old and relatively well understood. In contrast, the so-called \textit{end sum}, namely the boundary sum analogue for open manifolds, is a simple and useful operation that is not yet fully understood, since it can be not uniquely determined. In simple words, to obtain an end sum of two open manifolds, one attaches a 1-handle \textit{at infinity}, by following a properly embedded ray in each summand (in such a way a 1-handle at infinity is a sort of a piece of tape joining the two manifolds). End sum uniqueness can fail for several reasons, for instance when a summand has disconnected boundary, or for orientation problems, or in dimension 3 when the rays may be knotted. In the first sections, the authors present several interesting examples in higher dimensions (at least 4) and in all possible categories (homotopy, TOP, PL and DIFF) where the end sum is actually not uniquely determined, by proving also that this fact cannot be detected in weaker categories. Actually, these failures of uniqueness come from the presence of a complicated fundamental group behavior at the relevant ends (for example when rays determining a given end are not properly homotopic). In Section 4, the authors prove that in dimension at least 4, the \textit{semistability} property of the ends (also called \textit{Mittag-Leffler} condition) implies uniqueness of end sums and 1-handle attaching. Recall that an inverse sequence of groups \((H_n)\) is said to be semistable if the images of the bonding homomorphisms \(f_m: H_m\to H_n\) are the same for all but finitely many values of \(m > n\). Now, consider an exhausting increasing sequence of compact subsets \(K_n\) of a manifold \(X\), together with the inverse sequence of fundamental groups \((\pi_1 (X-K_n))\), with base points in a geodesic ray. This sequence of groups is semistable if and only if any two proper rays in \(X\) are properly homotopic. Note also that simply connected ends and topologically collared ends are both semistable. In the final sections uniqueness of adding collections of 0- and 1-handles at infinity is addressed. As corollaries of their results in this setting the authors presents new proofs of the ``hyperplane unknotting theorem'' (each proper embedding of \(\mathbb R^{n-1}\) in \(\mathbb R^{n}\) for \(n\geq4\) is unknotted) and of the the ``TOP Schoenflies theorem''.