math.RT

arXiv:2501.01136v1 Announce Type: new Abstract: Multi-agent reinforcement learning has emerged as a powerful framework for enabling agents to learn complex, coordinated behaviors but faces persistent challenges regarding its generalization, scalability and sample efficiency. Recent advancements have sought to alleviate those issues by embedding intrinsic symmetries of the systems in the policy. Yet, most dynamical systems exhibit little to no symmetries to exploit. This paper presents a novel framework for embedding extrinsic symmetries in multi-agent system dynamics that enables the use of symmetry-enhanced methods to address systems with insufficient intrinsic symmetries, expanding the scope of equivariant learning to a wide variety of MARL problems. Central to our framework is the Group Equivariant Graphormer, a group-modular architecture specifically designed for distributed swarming tasks. Extensive experiments on a swarm of symmetry-breaking quadrotors validate the effectiveness of our approach, showcasing its potential for improved generalization and zero-shot scalability. Our method achieves significant reductions in collision rates and enhances task success rates across a diverse range of scenarios and varying swarm sizes.

arXiv:2412.17099v1 Announce Type: cross Abstract: A finite group with an integral representation has two induced canonical actions, one on polynomials and one on Laurent polynomials. Knowledge about the invariants is in either case applied in many computations by means of symmetry reduction techniques, for example in algebraic systems solving or optimization. In this article, we realize the two actions as the additive action on the symmetric algebra and the multiplicative action on the group algebra of a lattice with Weyl group symmetry. By constructing explicit equivariant isomorphisms, we draw algorithmic relations between the two, which allow the transfer and preservation of representation- and invariant-theoretic properties. Our focus lies on the multiplicative coinvariant space, which is identified with the regular representation and harmonic polynomials.

arXiv:2408.00949v2 Announce Type: replace Abstract: Equivariant neural networks are neural networks with symmetry. Motivated by the theory of group representations, we decompose the layers of an equivariant neural network into simple representations. The nonlinear activation functions lead to interesting nonlinear equivariant maps between simple representations. For example, the rectified linear unit (ReLU) gives rise to piecewise linear maps. We show that these considerations lead to a filtration of equivariant neural networks, generalizing Fourier series. This observation might provide a useful tool for interpreting equivariant neural networks.