Periodic Reporting for period 1 - Modular Functors (Non-semisimple differential graded modular functors)
Période du rapport: 2022-01-01 au 2023-12-31
A modular category is built from elementary building blocks, the so-called simple objects. These simple objects can be added and multiplied (via the tensor product), but they can also be braided, which means that the position of two objects in a tensor product is changed in a certain way. If the modular category has a property that is called semisimplicity, then all objects can be built from elementary building blocks in a rather naive way. If this is not the case, then the modular category is called non-semisimple.
Generally, determining the tensor product of two objects can be a very hard problem. One of the pleasant properties of a semisimple modular category is that this tensor product can be directly computed via the famous Verlinde formula. Superficially, this formula is somewhat surprising, but there is a clear topological picture behind it: The vector space generated by the simple objects can be thought of as the result of applying a certain algebro-topological prescription (a so-called modular functor) to a torus (a space shaped like a doughnut). This has the consequence that the symmetries of the torus act on the vector space spanned by the simple objects of our modular category. There is now a certain symmetry operation (commonly referred to as S-transformation) that transforms the tensor multiplication (the one that we want to compute), into a much simpler multiplication. As a result, the tensor multiplication can be recovered from a very simple multiplication and the S-transformation. This is exactly the origin of the Verlinde formula!
So far, this applies to the semisimple case. In the very interesting non-semisimple case, the vector space of simple objects can be replaced by a chain complex, more specifically the so-called Hochschild (co)chain complex of the modular category. Compared to a vector space, this chain complex now contains not only one layer of information, but infinitely many. Technically speaking, the modular functor must be adjusted and replaced with a chain complex valued version. As we know from previous work of Schweigert and myself, building on works of Lentner-Mierach-Schweigert-Sommerhäuser, the symmetries of the torus still correctly act on this chain complex. Therefore, one could hope that it is still true that the multiplication coming from the tensor product is transformed into a much simpler multiplication, i.e. that there is indeed a differential graded Verlinde formula. Establishing such a formula is the main objective of the project.
The technical obstacles are immediate: For the traditional Verlinde formula, having a second multiplication was key. In the chain complex valued case, it is not clear where this multiplication should even come from. The proposal gives a candidate for such a second multiplicative structure, namely the so-called Deligne structure on the Hochschild cochain complex. The project requires a deep understanding of the Deligne structure for modular categories. The reason for the technical difficulty is that multiplicative structures on chain complexes are very hard to describe since information on infinitely layers must be specified. Even worse, commutativity, which for ordinary products is a property, may become additional structure that one needs to keep track of using a gadget called an operad (for us more specifically the little disks operad).
The insights that were gained cannot only be applied to the main goal of the proposal, but go beyond that. They were also used to endow the Ext algebra of a unimodular pivotal finite tensor category with a framed E_2-structure, thereby improving a result of Menichi.
These results are contained in the article "Homotopy Invariants of Braided Commutative Algebras and the Deligne Conjecture for Finite Tensor Categories" with C. Schweigert (arXiv preprint, submitted).
There are, for the moment, no socio-economic or societal implications.