Higher topological complexity and its symmetrization

From MaRDI portal
(Redirected from Publication:740534)



Abstract: We develop the properties of the n-th sequential topological complexity TCn, a homotopy invariant introduced by the third author as an extension of Farber's topological model for studying the complexity of motion planning algorithms in robotics. We exhibit close connections of TCn(X) to the Lusternik-Schnirelmann category of cartesian powers of X, to the cup-length of the diagonal embedding XhookrightarrowXn, and to the ratio between homotopy dimension and connectivity of X. We fully compute the numerical value of TCn for products of spheres, closed 1-connected symplectic manifolds, and quaternionic projective spaces. Our study includes two symmetrized versions of TCn(X). The first one, unlike Farber-Grant's symmetric topological complexity, turns out to be a homotopy invariant of X; the second one is closely tied to the homotopical properties of the configuration space of cardinality-n subsets of X. Special attention is given to the case of spheres.


The `topological complexity' \(\mathrm{TC}(X)\) of a space \(X\) defined by Farber is essentially the Švarc genus of the fibration equivalent to the diagonal \(\Delta: X \to X\times X\). Rudyak extended this to the iterated diagonal \(\Delta_n : X \to X^n\), leading to the notion of `higher topological complexity' \(\mathrm{TC}_n(X)\). In this paper, the authors compute \(\mathrm{TC}_n(X)\) for important examples (\S 3). Next they introduce two versions of `symmetric topological complexity': \(\mathrm{TC}_n^S(X)\) and \(\mathrm{TC}_n^\Sigma(X)\), the first one extending a notion of Farber and Grant, the second one being a homotopy invariant (\S 4). Finally they discuss some considerations about the case of spheres (\S 5). The authors prove that \(\mathrm{TC}_n(S^{k_1} \times S^{k_2} \times \dots \times S^{k_m}) = m(n-1)+l\) where \(l\) is the number of even dimensional spheres, extending the well-known fact that \(\mathrm{TC}(S^k) =\) 1 if \(k\) is odd and 2 if \(k\) is even. They also prove that for every closed simply connected symplectic manifold \(M^{2m}\) we have \(\mathrm{TC}_n(M) = nm\). They define \(\mathrm{TC}_n^\Sigma(X)\) as the least integer \(k\) which satisfies \(X^n = A_0 \cup \dots \cup A_k\), where each \(A_i\) is open, symmmetric and admits a continuous equivariant section \(s_i: A_i \to X^{J_n}\) for the map \(e_n: X^{J_n} \to X^n\) which is the fibration equivalent to the iterated diagonal \(\Delta_n : X \to X^n\). The fibration \(e_n\) induces a fibration at the level of orbit spaces of the action of the symmetric group \(\Sigma_n\) on the configuration spaces: \(\epsilon_n: Y_n(X) = e_n^{-1} (\mathrm{Conf}_n (X))/\Sigma_n \to \mathrm{Braid}_n(X) = \mathrm{Conf}_n(X) / \Sigma_n\). The authors prove that if \(X\) is a Euclidean neighborhood retract, then \(\mathrm{genus}(\epsilon_n) \leq \mathrm{TC}_n^\Sigma(X) \leq \mathrm{genus}(\epsilon_n) + \dots + \mathrm{genus}(\epsilon_2) + n-1\). This suggests the definition (for \(n \geq 2\)): \(\mathrm{TC}_n^S(X) = \mathrm{genus}(\epsilon_n) + \dots + \mathrm{genus}(\epsilon_2) + n-1\). It is shown that if \(k\) is a positive odd integer, for \(X = S^k\) and \(i \geq 2\), \(\mathrm{genus}(\epsilon_i) \geq 1\); further, \(\mathrm{genus}(\epsilon_i) = 1\) provided \(k = 1\). Finally the authors show that \(\mathrm{TC}_n^S(S^1) = 2(n-1)\), extending the known equality \(\mathrm{TC}_2^S(S^k) = 2\) (valid for any \(k > 0\)).




Cited in
(70)








This page was built for publication: Higher topological complexity and its symmetrization

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q740534)