Relative directed homotopy theory of partially ordered spaces (Q2381409): Difference between revisions

The author constructs on the category of pospaces, i.e. of partially ordered topological spaces, under a fixed pospace fibration and cofibration category structures in the sense of [\textit{H. J. Baues}, Algebraic homotopy. Cambridge Studies in Advanced Mathematics, 15. Cambridge etc.: Cambridge University Press (1989; Zbl 0688.55001)]. The homotopy notion is relative directed homotopy in both cases, i.e. relative homotopy preserving partial ordering. In the absolute case, these fibration and cofibration categories become a closed model category such that the homotopy notion is directed homotopy. From the computer scientific point of view, a topological space represents a state space and the partial ordering represents the observable time ordering. As explained in the introduction of the paper, absolute directed homotopy is not convenient for computer science since any execution path, i.e. any nondecreasing continuous map from the segment \([0,1]\) to the space, is directly homotopy equivalent to a constant path. The same problem exists for \textit{P. Bubenik} and \textit{K. Worytkiewicz}'s construction [Homology Homotopy Appl. 8, No. 1, 263--292 (2006; Zbl 1093.55007)] and in \textit{M. Grandis}' approach with his notion of d-homotopy [Cah. Topol. Géom. Différ. Catég. 44, No. 4, 281--316 (2003; Zbl 1059.55009)]. For the author, as for P. Bubenik and K. Worytkiewicz, the solution is to work with a notion of relative directed homotopy. For example, the notion of directed homotopy of execution paths with fixed initial and final states is the main ingredient in the construction of the fundamental category. The base pospace is called by them the context. Unfortunately, for general pospaces, nothing is explained about how one could find a good base pospace. To get an interesting and meaningful notion of relative directed homotopy, one first needs to locate the critical points, for example the branching and merging areas, and the initial and final states. But this is precisely what we are looking for and this is precisely what a homotopical approach of concurrency is supposed to do by reducing the size of the state space. Moreover, working within a category of pospaces under a fixed base pospace does not allow the comparison of pospaces under different base pospaces. The reviewer's approach of directed homotopy, also mentioned in the introduction of the paper, initialized in [\textit{P. Gaucher}, Homology Homotopy Appl. 5, No. 1, 549--599, electronic only (2003; Zbl 1069.55008)] has some similarities with author's approach, and also some differences. In the reviewer's approach, the role of the context is played by the set of objects of a time flow modelled by a small category without identity maps enriched over topological spaces. All possible contexts are then present in the same model category. Moreover, the set of objects of a flow already contains all critical points, by construction. Finally, the weak equivalences of the model category of flows do preserve the context, by construction as well. So the notion of weakly equivalent flows is very similar to relative directed homotopy.