Poset functor cocalculus and applications to topological data analysis
Bjørnar Gullikstad Hem · 2025
Read Poset functor cocalculus and applications to topological data analysis by Bjørnar Gullikstad Hem (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-05996.txt.
Introduction
Bjørnar Gullikstad Hem introduces a new approximation tool for functors in a 2025 mathematics paper. The tool, called poset cocalculus, adapts a known calculus idea to functors defined on distributive lattices. It builds a tower of approximations graded by codegree.
Hem applies the tool to problems in topological data analysis. The graded approximation of a multipersistence module stays stable under a suitable distance between modules. One classic filtration of geometric data turns out to be exactly the codegree two approximation of another. The codegree one approximation of maps between complexes relates, in a precise sense, to continuous maps between their geometric forms. The paper links abstract homotopy theory with concrete data tools.
Catalog record
| CREATOR | Bjørnar Gullikstad Hem |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 160 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-05996
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Poset functor cocalculus and applications to topological data analysis Bjørnar Gullikstad Hem Abstract . We introduce a new flavor of functor cocalculus, named poset cocalculus , as a tool for studying approximations in topological data analysis. Given a functor from a distributive lattice to a model category, poset cocalculus produces a Taylor telescope of degree n n approximations of the functor, where a degree n n functor takes strongly bicartesian ( n + 1 ) (n+1) –cubes to homotopy cocartesian ( n + 1 ) (n+1) –cubes. We give several applications of this new functor cocalculus. We prove that the degree n n approximation of a multipersistence module is stable under an appropriate notion of interleaving distance. We show that the Vietoris-Rips filtration is precisely the degree 2 approximation of the Čech filtration, and we draw connections between poset cocalculus and discrete Morse theory. We demonstrate that the degree 1 approximation of the space of simplicial maps between two simplicial complexes is in some sense the space of continuous maps between their realizations, and that this statement can be made precise. 1. Introduction 1.1. Background 1.1.1. Topological data analysis Topological data analysis (TDA) can be described as the general approach of applying methods from topology to data analysis. One of the most common methods in TDA is persistent homology . Given a point cloud, persistent homology provides successive approximations to the homology groups of the underlying space, by constructing an \mathbb{R}_{\geq 0} –parameterized filtration of simplicial complexes from the point cloud, and then post-composing with the functor H_{i}(-;\mathbb{Fb}) , i.e., homology in some degree i over some field \mathbb{Fb} . In total, this gives a functor F\colon\mathbb{R}_{\geq 0}\to\mathsf{Vec}_{\mathbb{Fb}} from the poset \mathbb{R}_{\geq 0} (i.e., nonnegative real numbers) to \mathsf{Vec}_{\mathbb{Fb}} (vector spaces over \mathbb{Fb} ), which we call a persistence module . There are several methods of constructing an \mathbb{R}_{\geq 0} -parametrized filtration from a point cloud. Two of the most common ones are the Čech filtration and its less computationally expensive approximation, the Vietoris-Rips filtration . Given a point cloud V\subset\mathbb{R}^{d} , the Čech filtration at t\in\mathbb{R}_{\geq 0} is the simplicial complex \{\sigma\subset V:f_{\mathbf{C}}(\sigma)\leq t\} , where f_{\mathbf{C}} is the function that sends a simplex to its radius (i.e., the radius of the smallest ball containing the simplex). The Vietoris-Rips complex is defined similarly, except that f_{\mathbf{C}} is replaced by f_{\mathbf{VR}} , which sends a simplex to its diameter (i.e., the maximum distance between two points in the simplex). One then constructs a barcode diagram from the persistence module, a visual representation from which one can deduce the homology of the underlying manifold, given a sufficiently good point sample. Construction of the barcode diagram is possible due to the structure theorem [ 8 ] , which says that a persistence module can be decomposed into a direct sum of simple components known as interval modules . The other important theorem that makes persistent homology interesting is the stability theorem, which guarantees the stability of the barcode diagram under small perturbations of the initial data. Despite the stability theorem, persistent homology suffers from robustness issues. For example, persistent homology is highly unstable to outliers in the point data [ 5 ] . A possible solution to this is multipersistent homology [ 13 , 7 ] , where instead of constructing an \mathbb{R}_{\geq 0} –parameterized filtration from the point data, one constructs an (\mathbb{R}_{\geq 0})^{k} –parameterized filtration for some integer k . This would allow one to, for example, take into account the certainty of point data and in this way avoid the outlier problem. Analogously to the single-parameter case, multipersistent homology gives a multipersistence module , i.e., a functor F\colon(\mathbb{R}_{\geq 0})^{k}\to\mathsf{Vec}_{\mathbb{Fb}}. However, an analogue of the structure theorem fails to hold for such functors when k\geq 2 . Thus, you cannot construct a barcode diagram for multipersistent homology, which is a well-known problem in TDA. Instead of fixing the homology degree i when constructing a (multi-)persistence module, we could instead consider the functor H_{\bullet}(-;\mathbb{Fb}) that sends a simplicial complex X to the chain complex with H_{i}(X;\mathbb{Fb}) in degree i and only zero differentials. This would give a functor F\colon(\mathbb{R}_{\geq 0})^{k}\to\mathrm{Ch}_{\mathbb{Fb}}, where \mathrm{Ch}_{\mathbb{Fb}} is the category of chain complexes over \mathsf{Vec}_{\mathbb{Fb}} . As we care about such functors up to quasi-isomorphism, we can study them from the point of view of model categories, using the injective model structure on \mathrm{Ch}_{\mathbb{Fb}} [ 20 , Theorem 2.3.13] . 1.1.2. Functor calculus Functor calculus is a general framework for studying functors, and is inspired by calculus of functions, and in particular the Taylor expansion of a function. The general idea of functor calculus is to introduce the notion of a degree n functor , analogous to a degree n polynomial, and then study the degree n approximations of a functor. Typically, one gets a tower of approximations of the following form, which under reasonable conditions converges in a well-defined sense to the functor studied. \displaystyle\hbox to97.93pt{\vbox to109.27pt{\pgfpicture\makeatletter\hbox{\hskip 48.96605pt\lower-54.63412pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke} \lxSVG@begingroup@{fill} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin} {}{}{{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin} {{}{}{{{}}{{}}{{}}{{}}{{}}{{}}}{{{\lx@inpgf@ignorespaces}}}{{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-45.63304pt}{-46.89172pt}\lxSVG@begingroup@{transform} \pgfsys@hbox{60}\lxSVG@closescope }}}{{{\lx@inpgf@ignorespaces{}}}{{}}{{}}{{}}{{}}{{}}{{}}}} \lxSVG@closescope }}} \par{ {}{}{}}{}{ {}{}{}}{}{ {}{}{}}{}{ {}{}{}}{}{ {}{}{}}{}\lx@inpgf@ignorespaces\hbox{\hbox{\hbox{\hbox{\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin} { {}{}{}}{}{ {}{}{}} {{{{{}}{ {}{}}{}{}{{}{}}}}}{}{{{{{}}{ {}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{{ { \lxSVG@setdash{0.0pt}\lx@inpgf@ignorespaces\lxSVG@roundcap\lx@inpgf@ignorespaces\lxSVG@roundjoin\lx@inpgf@ignorespaces{} {\lx@inpgf@ignorespaces}{}{} {}{}{} \lxSVG@stroke}} }{}{}{{}}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 37.7 -10.1 L 37.7 -50.51}{fill:none} {{}{{}}{}{}{{}}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin} \lxSVG@transformcm{0.0}{-1.0}{1.0}{0.0}{27.24275pt}{-36.70041pt}\lxSVG@begingroup@{transform} \lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke-dasharray,stroke-dashoffset} \lxSVG@begingroup@{stroke-linecap} \lxSVG@begingroup@{stroke-linejoin} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}}\lx@inpgf@ignorespaces \lxSVG@closescope }}}\hbox{{\lxSVG@begingroup@{_scopebegin} { {}{}{}}{}{ {}{}{}} {{{{{}}{ {}{}}{}{}{{}{}}}}}{}{{{{{}}{ {}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{}{}{{}}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 37.7 51.06 L 37.7 10.65}{fill:none} {{}{{}}{}{}{{}}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin} \lxSVG@transformcm{0.0}{-1.0}{1.0}{0.0}{27.24275pt}{7.50035pt}\lxSVG@begingroup@{transform} \lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke-dasharray,stroke-dashoffset} \lxSVG@begingroup@{stroke-linecap} \lxSVG@begingroup@{stroke-linejoin} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}}\lx@inpgf@ignorespaces \lxSVG@closescope }}}\hbox{{\lxSVG@begingroup@{_scopebegin} { {}{}{}}{}{ {}{}{}} {{{{{}}{ {}{}}{}{}{{}{}}}}}{}{{{{{}}{ {}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{}{}{{}}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -39.27 -51.06 L 30.46 50.6}{fill:none} {{}{{}}{}{}{{}}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin} \lxSVG@transformcm{0.56564}{0.82465}{-0.82465}{0.56564}{22.12479pt}{36.73547pt}\lxSVG@begingroup@{transform} \lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke-dasharray,stroke-dashoffset} \lxSVG@begingroup@{stroke-linecap} \lxSVG@begingroup@{stroke-linejoin} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}}\lx@inpgf@ignorespaces \lxSVG@closescope }}}\hbox{{\lxSVG@begingroup@{_scopebegin} { {}{}{}}{}{ {}{}{}} {{{{{}}{ {}{}}{}{}{{}{}}}}}{}{{{{{}}{ {}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{}{}{{}}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -32.34 -51.06 L 23.4 -10.43}{fill:none} {{}{{}}{}{}{{}}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin} \lxSVG@transformcm{0.8081}{0.58904}{-0.58904}{0.8081}{17.07144pt}{-7.41815pt}\lxSVG@begingroup@{transform} \lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke-dasharray,stroke-dashoffset} \lxSVG@begingroup@{stroke-linecap} \lxSVG@begingroup@{stroke-linejoin} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}}\lx@inpgf@ignorespaces \lxSVG@closescope }}}\hbox{{\lxSVG@begingroup@{_scopebegin} { {}{}{}}{}{ {}{}{}} {{{{{}}{ {}{}}{}{}{{}{}}}}}{}{{{{{}}{ {}{}}{}{}{{}{}}}}}{{}}{}{}{}{}{}{}{}{{}}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -28.99 -61.16 L 11.42 -61.16}{fill:none} {{}{{}}{}{}{{}}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{8.45245pt}{-44.20076pt}\lxSVG@begingroup@{transform} \lxSVG@begingroup@{_scopebegin} \lxSVG@begingroup@{stroke-dasharray,stroke-dashoffset} \lxSVG@begingroup@{stroke-linecap} \lxSVG@begingroup@{stroke-linejoin} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}}\lx@inpgf@ignorespaces \lxSVG@closescope }}} \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}} (1.1.1) Among the flavors of functor calculus are Goodwillie calculus [ 14 , 15 , 16 ] , abelian functor calculus [ 21 ] and manifold calculus [ 17 ] . Manifold calculus is particularly relevant to our project, as it is used to study a specific class of functors from a poset to spaces. Manifold calculus has been used successfully to study the homotopy type of components of the space of embeddings \textrm{emb}(M,N) for manifolds M,N . For example, manifold calculus has been used to study long knots, which are embeddings from \mathbb{R} into \mathbb{R}^{n} that are linear outside a compact set. Using manifold calculus to study these knots is particularly effective when n>3 , as in this case the tower of approximations for \textrm{emb}(M,N) converges [ 17 ] . In [ 26 ] , it is explained how, when n>3 , manifold calculus gives rise to a