On sequential versions of distributional topological complexity

From MaRDI portal





This paper studies the sequential (or iterative) version of the concept of \textit{distributional topological complexity}. This concept extends Farber's \textit{topological complexity} (\(TC\)) and the \textit{distributional} variant previously introduced by \textit{A. Dranishnikov} and \textit{E. Jauhari} [EMS Ser. Ind. Appl. Math. 4, 363--385 (2024; Zbl 1557.55001)] for the case \(m=2\). The aim is to measure the minimal number of continuous motion-planning rules, possibly chosen in a probabilistic manner, connecting multiple points of a topological space under fixed timing, while allowing the system to split into several components and reassemble during motion. The new invariant family \((dTC_m)_{m\ge 2}\) parallels the classical sequence \((TC_m)\), and its structural properties -- such as homotopy invariance, upper and lower bounds, and product behavior -- are investigated.\N\NThe formal definition of \(dTC_m(X)\) is based on the path space \(P(X)\) with the compact-open topology and the space \(B_n(P(X))\) of probability measures of finite support (at most \(n\) points) on \(P(X)\). A \textit{distributional \(m\)-motion planner} with at most \(n+1\) ``pieces is a continuous map from \(X^m\) to \(B_{n+1}(P(X))\) that sends the \(m\)-tuple of configurations to probabilistic choices of paths under the fixed timing \(t_i = \frac{i-1}{m-1}\). The smallest such \(n\) is \(dTC_m(X)\). This invariant is always homotopy-invariant and relates to similar invariants such as \(TC_m(X)\) and \(dcat(X^k)\), typically satisfying\N\[\Ndcat(X^{m-1}) \ \le \ dTC_m(X) \ \le \ TC_m(X),\N\quad dTC_m(X) \le dcat(X^m).\N\]\NIt increases (or stays the same) with \(m\), satisfies a max-type lower bound for products\N\[\N\max\{dTC_m(X), dTC_m(Y)\} \le dTC_m(X\times Y),\N\]\Nand admits an upper bound\N\[\NdTC_m(X) \le (dTC(X)+1)^{m-1} - 1.\N\]\N\NLower bounds are obtained using Alexander-Spanier cohomology and symmetric product spaces \(SP^k(X)\). Constructing cohomology classes with nontrivial product that vanish upon restriction to the diagonal yields bounds of the form \(dTC_m(X) \ge n\). On the other hand, using the Ganea-Schwarz viewpoint, \(dTC_m(X)\) equals the \textit{distributional sectional category} of a certain fibration, which immediately gives another proof of homotopy invariance.\N\NFor \(H\)-spaces there is an exact formula:\N\[\NdTC_{m+1}(X) = dcat(X^m),\N\]\Nleading to examples such as \(S^k\) for \(k=1,3,7\) with \(dTC_m = m-1\), and the \(n\)-torus \(T^n\) with \(dTC_m = n(m-1)\). Explicit computations are given for various standard spaces: for odd spheres \(dTC_m(S^{2k-1}) = m-1\), for even spheres \(dTC_m(S^{2k}) = m\), for closed surfaces of genus \(g\ge 2\) the value is \(2m\), for the torus \(T^2\) the value is \(2(m-1)\), for products of \(n\) spheres it is \(n(m-1) + \ell_n\) where \(\ell_n\) is the number of even-dimensional factors, and for products of a surface with spheres it is \(m(n+2) - n + \ell_n\).\N\NFinally, the author compares \(dTC_m\) with related invariants \(ATC_m\) and \(acat\) [\textit{B. Knudsen} and \textit{S. Weinberger}, SIAM J. Appl. Algebra Geom. 8, No. 3, 713--732 (2024; Zbl 1552.55001)], which differ in the topology placed on the measure space, and shows that in many computed examples they agree, though differences also occur. Overall, the article offers both theoretical foundations and explicit calculations for sequential distributional topological complexity across important classes of spaces, highlighting its connections and distinctions from classical motion-planning invariants.\N\N\textbf{Table of \(dTC_m\) Values for Key Spaces:}\N\begin{center}\N\begin{tabular}{|l|c|l|}\N\hline\N\textbf{Space Type} & \textbf{\(dTC_m\) Formula} & \textbf{Notes} \\ \hline\NOdd sphere \(S^{2k-1}\) & \(m-1\) & For \(k=1,3,7\) (\(H\)-space) and all odd spheres \\ \hline\NEven sphere \(S^{2k}\) & \(m\) & \\ \hline\NTorus \(T^n\) & \(n(m-1)\) & \(n =\) dimension of the torus \\ \hline\NClosed surface \(\Sigma_g\) (\(g \ge 2\)) & \(2m\) & Higher genus surface \\ \hline\NSphere surface \(\Sigma_0 = S^2\) & \(m\) & Genus 0 surface \\ \hline\NTorus surface \(\Sigma_1 = T^2\) & \(2(m-1)\) & Genus 1 surface \\ \hline\NProduct \(\prod_{j=1}^n S^{k_j}\) & \(n(m-1) + \ell_n\) & \(\ell_n =\) number of even-dimensional factors \\ \hline\N\(\Sigma_g \times \prod S^{k_j}\) & \(m(n+2) - n + \ell_n\) & Genus \(g\) surface with \(n\) spheres, \(\ell_n\) even-dimensional \\ \hline\N\end{tabular}\N\end{center}











This page was built for publication: On sequential versions of distributional topological complexity

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7018165)