scientific article; zbMATH DE number 2222243

From MaRDI portal
Publication:5701015





In directed algebraic topology, paths and homotopies are in general irreversible, and instead of a fundamental groupoid, one associates a fundamental category \(\uparrow\Pi_1(X)\). In the non-directed case, the equivalences of fundamental groupoids induced by homotopy equivalence are equivalences of categories, and one may take the categorical skeleton as a minimal model, choosing one point in each path component. The fundamental categories should be studied up to directed homotopy equivalence. However, a homotopy will only induce a natural transformation of fundamental categories, and it is in general not invertible. Hence it does not induce an equivalence of fundamental categories, and minimal models should not be expected to be skeleta. In the present paper, two dual notions, \textit{future} and \textit{past} equivalence, are introduced. Minimal models of a category up to these equivalences are introduced as the ``shape of the category. A future equivalence \((f,g;\varphi,\psi)\) between categories \(X,Y\) is a pair of functors \(f:X\rightleftarrows Y:g\) and natural transformations \(\varphi:1_X\to gf\) and \(\psi: 1_Y\to fg\). These are required to satisfy the coherence conditions \(f\varphi=\psi f:f\to fgf\) and \(\varphi g=g\psi:g\to gfg\). This gives an equivalence relation on categories. (And past equivalence is defined similarly.) A category is future contractible if and only if it has a terminal object. The analogue of components in a category \(X\) is defined as follows: A morphism \(a':x\to x'\) is \textit{future regular} if given \(a':x\to x\) there is a commutative diagram \(ha=ka'\) and given two maps \(a_1,a_2:x'\to x\) with \(a_1a=a_2a\), there is an \(h\) such that \(ha_1=ha_2\). The future regularity equivalence relation om objects is generated by \(x\sim^+ x'\) if there is an \(a\) as above. The objects \(sp^+(X)\) of a future spectrum of \(X\) is a subset of the objects, which contains precisely one object \(sp^+(x)\) in each future regularity class \([x]\), and there is precisely one morphism \(\eta x: x\to sp^+(x)\) (\(sp^+(x)\) is a terminal object in the equivalence class considered as a full subcategory). Moreover, every morphism \(a:x\to sp^+(x')\) factors as \(a=h.\eta x\) for a unique \(h:sp^+(x)\to sp^+(x')\). The future spectrum \(Sp^+(X)\) is the full subcategory on this set of objects. Clearly not all categories have a future spectrum -- future regularity classes may not have a terminal object, but if \(X\) has a future spectrum, then it is unique. If for \(i:F\to X\) and \(j:G\to X\), \(i(F)\) and \(j(G)\) are future spectra, then there is a unique invertible natural transformation \(F\to G\). Similarly, a past spectrum is defined. Whenever \(X\) has both a future and a past spectrum, a \(pf\) spectrum, which is a diagram containing both the future and the past spectrum, is defined. Such a \(pf\)-spectrum gives a minimal injective and an associated minimal projective model.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5701015)