The Grassmannian of affine subspaces (Q2031063)

From MaRDI portal
Revision as of 16:47, 30 July 2023 by Importer (talk | contribs) (‎Created a new Item)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
The Grassmannian of affine subspaces
scientific article

    Statements

    The Grassmannian of affine subspaces (English)
    0 references
    0 references
    0 references
    0 references
    8 June 2021
    0 references
    In this article, that follows a previous one by the same authors [SIAM J. Matrix Anal. Appl. 40, No. 2, 371--393 (2019; Zbl 1419.14086)], the Grassmannian \(\textrm{Graff}(k,n)\) of \(k\)-dimensional affine subspaces in \(\mathbb R^n\) is introduced and studied. A general element is of the form \(\mathbb A+b\), where \(\mathbb A\subset\mathbb R^n\) is a \(k\)-dimensional vector space and \(b\) is a vector in \(\mathbb R^n\). The Authors make a summary of the possible descriptions of \(\textrm{Graff}(k,n)\) from several points of view, as an object in algebra, algebraic geometry, differential geometry, or algebraic topology. Each of them leads to introduce and study various objects on it, as for instance Schubert varieties and cycles, homotopy, homology and cohomology, metrics and geodesics. They point out that ``common problems in multivariate data analysis and machine learning are naturally optimization problems over \(\textrm{Graff}(k,n)\)''.\par Just to give an example, a geometric description of \(\textrm{Graff}(k,n)\) is as a Zariski open subset of \(\mathrm{Gr}(k+1,n+1)\), by identifying \(\mathbb A+b\) with the vector subspace of \(\mathbb R^{n+1}\) generated by \(\mathbb A\) and by the vector \(b+e_{n+1}\), where \(e_{n+1}=(0,\ldots,0,1)\). On the other hand, the map \(\tau: \textrm{Graff}(k,n)\to \textrm{Gr}(k,n)\), associating to \(\mathbb A+b\) the vector space \(\mathbb A\), exhibits \(\textrm{Graff}(k,n)\) as the universal quotient bundle on \(\textrm{Gr}(k,n)\).\par In the last part of the article, the Authors introduce three notions of probability densities on \(\textrm{Gr}(k,n)\): uniform, Langevin (or von Mises-Fisher), and Langevin-Gaussian.
    0 references
    affine Grassmannian
    0 references
    affine subspaces
    0 references
    Schubert calculus
    0 references
    homotopy and (co)homology
    0 references
    probability densities
    0 references
    distances and metrics
    0 references
    multivariate data analysis
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references