Poincaré/Koszul duality (Q1729877)

It is very well known that Poincaré and Koszul dualities play central roles in algebraic topology and Lie algebras. The present paper revolves around a natural question: what is Poincaré duality in the setting of factorization homology? A factorization algebra is a homology theory for \(n\)-manifolds. These theories are natural with respect to embeddings of manifolds and satisfy a multiplicative generalization of the Eilenberg-Steenrod axioms for ordinary homology. The origins of factorization homology go back to the configuration spaces of Salvatore and Segal or the algebraic approaches to conformal field theories of Beilinson and Drinfeld, and their applications are remarkable not only in topological field theories, where they appear in the works of Costello on renormalization of field theories, but also in algebraic geometry, where they were used by Lurie and Gaitsgory in the proof of the Weil conjecture on Tamagawa numbers for algebraic groups. Other interesting examples can be found when doing research on the connections between gauge theory, quantum groups or knot and \(3\)-manifolds invariants. In this case, the authors focus on two aspects of factorization homologies: on the one hand, factorization homology theories are characterized by a monoidal generalization of the Eilenberg-Steenrod axioms for usual homology, so that factorization homology specializes to ordinary homology in the case the target symmetric monoidal category is that of chain complexes with direct sum. Therefore, it makes sense to ask that different values of factorization homology theories, valued in a general symmetric monoidal category, likewise enjoy a relationship specializing to that of Poincaré duality. On the other hand factorization homology theories are algebraic models for physical field theories. The two sources of motivation offer two pointers on where to look for Poincaré duality in factorization homology. The first says that the factorization homology with coefficients in an algebra \(A\) should be equivalent to some other construction, not factorization homology, but some cohomological variant. The TQFT motivation suggests that the choice of coefficients for this factorizable generalization of cohomology should be related to the Koszul dual of \(A\). Assembling these hints, the authors conclude that such a Poincaré duality should relate the factorization homology with coefficients in \(A\) to a not necessarily perturbative form of factorization homology with coefficients related to the Koszul dual of \(A\). The paper is divided into four parts: in the first one the authors arrive at a geometric presentation of an \(n\)-disk coalgebra structure on the \(n\)-fold iterated bar construction of an augmented \(n\)-disk algebra. This fact is applied to construct the Poincaré/Koszul duality map, which goes from factorization homology with coefficients in an \(n\)-disk algebra to factorization cohomology with coefficients in the Koszul dual \(n\)-disk coalgebra. Section 2 introduces two (co)filtrations of factorization homology and cohomology coming from cardinality filtrations. This section includes the proof that the Poincaré/Koszul duality map is an equivalence when the algebra \(A\) is connected. Section 3 introduces factorization homology with coefficients in a formal moduli problem and proves that the Poincaré/Koszul duality map is an equivalence in the case of a \((-n)\)-coconnective \(n\)-disk algebra over a field. The paper concludes by specializing these results to the case of associative algebras and Lie algebras in section 4.