Combinatoire et algorithmique des factorisations tangentes à l'identité

Abstract : Combinatorics has solved many problems in Mathematics, Physics and Computer Science, in return these domains inspire new questions to combinatorics. This memoir entitled "Combinatorics and algorithmics of factorization tangent to indentity includes several works on the combinatorial deformations of the shuffle product. The aim of this thesis is to write factorizations wich principal term is the identity through the use of tools relating mainly to combinatorics on the words (orderings, grading etc). In the classical case, let F be the free algebra. Due to the fact that F is an enveloping algebra, one has an exact factorization of the identity of End(F) = F⨶F as an infinite product of exponentials (End(F) being endowed with the shuffle product on the left and the concatenation on the right, a faithful representation of the convolution product) as follows : first on begins with a PBW basis, second one computes the family of coordinate forms and then non-trivial (combinatorial) properties of theses families in duality gives the factorization. Starting from the other side and writing the same product does give exactly identity only under very restrictive conditions that we clarify here. In many other (deformed) cases, the explicit construction of pairs of bases in duality requires combinatorial and algorithmic studies that we provide in this memoir.
Liste complète des métadonnées

Cited literature [2 references]  Display  Hide  Download
Contributor : Abes Star <>
Submitted on : Tuesday, June 13, 2017 - 5:15:06 PM
Last modification on : Wednesday, February 6, 2019 - 1:25:19 AM
Document(s) archivé(s) le : Tuesday, December 12, 2017 - 3:46:50 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01538597, version 1



Ladji Kane. Combinatoire et algorithmique des factorisations tangentes à l'identité. Algorithme et structure de données [cs.DS]. Université Paris-Nord - Paris XIII, 2014. Français. ⟨NNT : 2014PA132059⟩. ⟨tel-01538597⟩



Record views


Files downloads