3 citations · 9 across the 10 of their papers we have counts for
12 papers
Reduce, Reuse, Recycle: Categories for Compositional Reinforcement Learning
Georgios Bakirtzis, Michail Savvas, Ruihan Zhao +2
In reinforcement learning, conducting task composition by forming cohesive, executable sequences from multiple tasks remains challenging. However, the ability to (de)compose tasks…
Cosection Localization for D-Manifolds and -Shifted Symplectic Derived Schemes, Revisited
Michail Savvas
This is a continuation of prior work of the author on cosection localization for d-manifolds. We construct reduced virtual fundamental classes for derived manifolds with surjective…
Good Moduli Spaces in Derived Algebraic Geometry
Eric Ahlqvist, Jeroen Hekking, Michele Pernice +1
We develop a theory of good moduli spaces for derived Artin stacks, which naturally generalizes the classical theory of good moduli spaces introduced by Alper. As such, many of the…
Stabilizer Reduction for Derived Stacks and Applications to Sheaf-Theoretic Invariants
Jeroen Hekking, David Rydh, Michail Savvas
We construct a canonical stabilizer reduction for any derived -algebraic stack over as a sequence of derived Kirwan blow-ups, under mild natural…
Categorical semantics of compositional reinforcement learning
Georgios Bakirtzis, Michail Savvas, Ufuk Topcu
Compositional knowledge representations in reinforcement learning (RL) facilitate modular, interpretable, and safe task specifications. However, generating compositional models req…
Virtual Riemann-Roch Theorems for Almost Perfect Obstruction Theories
Michail Savvas
This is the third in a series of works devoted to constructing virtual structure sheaves and -theoretic invariants in moduli theory. The central objects of study are almost perf…
The d-critical structure on the Quot scheme of points of a Calabi-Yau 3-fold
Andrea T. Ricolfi, Michail Savvas
The Artin stack of -dimensional sheaves of length on carries two natural d-critical structures in the sense of Joyce. One comes from its descrip…
Localizing Virtual Structure Sheaves for Almost Perfect Obstruction Theories
Young-Hoon Kiem, Michail Savvas
Almost perfect obstruction theories were introduced in an earlier paper by the authors as the appropriate notion in order to define virtual structure sheaves and -theoretic inva…
Cosection Localization and Vanishing for Virtual Fundamental Classes of D-Manifolds
Michail Savvas
We establish cosection localization and vanishing results for virtual fundamental classes of derived manifolds, combining the theory of derived differential geometry by Joyce with…
Intrinsic Stabilizer Reduction and Generalized Donaldson-Thomas Invariants
Michail Savvas
Let be a stability condition on the bounded derived category of a Calabi-Yau threefold and a moduli stack parametrizing $…
K-Theoretic Generalized Donaldson-Thomas Invariants
Young-Hoon Kiem, Michail Savvas
We introduce the notion of almost perfect obstruction theory on a Deligne-Mumford stack and show that stacks with almost perfect obstruction theories have virtual structure sheaves…
Generalized Donaldson-Thomas Invariants via Kirwan Blowups
Young-Hoon Kiem, Jun Li, Michail Savvas
We develop a virtual cycle approach towards generalized Donaldson-Thomas theory of Calabi-Yau threefolds. Let be the moduli stack of Gieseker semistable sheaves of fi…