xi, 95 p. : ill. A print copy of this thesis is available through the UO Libraries. Search the library catalog for the location and call number. / We investigate some homological properties of graded algebras. If A is an R -algebra, then E (A) := Ext A ( R, R ) is an R-algebra under the cup product and is called the Yoneda algebra. (In most cases, we assume R is a field.) A well-known and widely-studied condition on E(A) is the Koszul property. We study a class of deformations of Koszul algebras that arises from the study of equivariant cohomology and algebraic groups and show that under certain circumstances these deformations are Poincaré-Birkhoff-Witt deformations. Some of our results involve the [Special characters omitted] property, recently introduced by Cassidy and Shelton, which is a generalization of the Koszul property. While a Koszul algebra must be quadratic, a [Special characters omitted] algebra may have its ideal of relations generated in different degrees. We study the structure of the Yoneda algebra corresponding to a monomial [Special characters omitted.] algebra and provide an example of a monomial [Special characters omitted] algebra whose Yoneda algebra is not also [Special characters omitted]. This example illustrates the difficulty of finding a [Special characters omitted] analogue of the classical theory of Koszul duality. It is well-known that Poincaré-Birkhoff-Witt algebras are Koszul. We find a [Special characters omitted] analogue of this theory. If V is a finite-dimensional vector space with an ordered basis, and A := [Special characters omitted] (V)/I is a connected-graded algebra, we can place a filtration F on A as well as E (A). We show there is a bigraded algebra embedding Λ: gr F E (A) [Special characters omitted] E (gr F A ). If I has a Gröbner basis meeting certain conditions and gr F A is [Special characters omitted], then Λ can be used to show that A is also [Special characters omitted]. This dissertation contains both previously published and co-authored materials. / Committee in charge: Brad Shelton, Chairperson, Mathematics; Victor Ostrik, Member, Mathematics; Christopher Phillips, Member, Mathematics; Sergey Yuzvinsky, Member, Mathematics; Van Kolpin, Outside Member, Economics

x, 68 p. : ill. (some col.) / Motivated by the Adams spectral sequence for computing stable homotopy groups, Priddy defined a class of algebras called Koszul algebras with nice homological properties. Many important algebras arising naturally in mathematics are Koszul, and the Koszul property is often tied to important structure in the settings which produced the algebras. However, the strong defining conditions for a Koszul algebra imply that such algebras must be quadratic. A very natural generalization of Koszul algebras called K 2 algebras was recently introduced by Cassidy and Shelton. Unlike other generalizations of the Koszul property, the class of K 2 algebras is closed under many standard operations in ring theory. The class of K 2 algebras includes Artin-Schelter regular algebras of global dimension 4 on three linear generators as well as graded complete intersections. Our work comprises two distinct projects. Each project was motivated by an aspect of the theory of Koszul algebras which we regard as sufficiently powerful or fundamental to warrant an interpretation for K 2 algebras. A very useful theorem due to Backelin and Fröberg states that if A is a Koszul algebra and I is a quadratic ideal of A which is Koszul as a left A -module, then the factor algebra A/I is a Koszul algebra. We prove that if A is Koszul algebra and A I is a K 2 module, then A/I is a K 2 algebra provided A/I acts trivially on Ext A ( A/I,k ). As an application of our theorem, we show that the class of sequentially Cohen-Macaulay Stanley-Reisner rings are K 2 algebras and we give examples that suggest the class of K 2 Stanley-Reisner rings is actually much larger. Another important recent development in ring theory is the use of A ∞ -algebras. One can characterize Koszul algebras as those graded algebras whose Yoneda algebra admits only trivial A ∞ -structure. We show that, in contrast to the situation for Koszul algebras, vanishing of higher A ∞ -structure on the Yoneda algebra of a K 2 algebra need not be determined in any obvious way by the degrees of defining relations. We also demonstrate that obvious patterns of vanishing among higher multiplications cannot detect the K 2 property. This dissertation includes previously unpublished co-authored material. / Committee in charge: Dr. Brad Shelton, Chair; Dr. Victor Ostrik, Member; Dr. Nicholas Proudfoot, Member; Dr. Arkady Vaintrob, Member; Dr. David Boush, Outside Member

En commentant certains résultats des sciences physiques ou mathématiques, plus particulièrement de la seconde moitié du XXe siècle, on cherche à comprendre l'importance philosophique du concept de diagramme, qui est au cœur de la théorie mathématique des catégories, des topoi et des esquisses. Partant du constat que les diagrammes et catégories contraignent à des options ontologiques, on propose pour étudier leur disposition conjointe de suivre quatre concepts fondamentaux qui forment le quadrilatère épistémique (la virtualité, la fonctorialité, l'universalité et la dualité). Le virtuel est nécessaire parce qu'une table n'existe pas de la même manière que le bleu du ciel qui n'a pas de réalité matérielle. La fonctorialité et le lemme de Yoneda imposent de reconsidérer le statut de l'objet. Le théorème de Diaconescu illustre l'idée que la logique immanente d'un lieu est déterminée par le topologique, que la logique n'a pas l'importance qu'on lui accorde parfois. L'universalité et la dualité déplace la notion de vérité qui n'est plus une simple valuation, mais une vérité-foudre, une vérité-événement qui fonctionne par adéquation et résonance de pans entiers de connaissance et non plus par inférence logique. Le diagramme devient le lieu de cette vérité qui passe par le geste. Dès lors, il devient possible de croiser ontologie et topologie en une onto-(po)-logie (ou une ontologie toposique) qui ne soit pas en contraction avec les philosophies de l'immanence. L'univocité de l'Être ne s'oppose pas à l'approche catégorielle. Plus encore : la prégnance des formes duales incite à penser l'hypothèse que l'Un est le dual de l'Être.

Afin d'étudier la concurrence au moyen de techniques issues de la topologie algébrique, on étudie les propriétés de la catégorie des espaces ordonnés. Le foncteur "catégorie fondamentale" associe à chaque tel espace une petite catégorie sans boucle, dont la taille de l'ensemble des objets est trop grand par rapport à l'information qu'elle contient. On définit alors la catégorie des composantes d'une petite catégorie sans boucle et l'on prouve un théorème qui justifie le bien fondé de cette définition ainsi qu'un théorème "à la van Kampen" qui ouvre la voie vers des calculs effectifs. On représente ainsi les programmes écrits en langage PV (on entend ici la version originale de Dijkstra) : plusieurs exemple sont traîtés.

Quasi-hereditary algebras are an important class of algebras with many appli-cations in representation theory, most notably the representation theory of semi-simple complex Lie-algebras. Such algebras sometimes admit an exact Borel sub-algebra, that is a subalgebra satisfying similar formal properties to the Borel sub-algebras from Lie theory. This thesis is divided into two parts. In the first part we classify quasi-hereditary algebras with two simple modules over perfect fields up to Morita equivalence, generalizing a similar result by Membrillo-Hernandez for thealgebraically closed case. In the second part, we take a poset X, a certain set M of constants, and a finite set ρ of paths in the Hasse-diagram of X and construct analgebra A(X, M, ρ) that generalizes the twisted double incidence algebras originally introduced by Deng and Xi. We provide necessary and sufficient conditions for this algebra to be quasi-hereditary when X is a tree, and we show that A(X, M, ρ) admits an exact Borel subalgebra when these conditions are satisfied. Following this, we compute the Ext-algebra of the standard modules of A(X, M, ρ).