Q-Math Seminar
Laura González-Bravo (UCM-ICMAT)
A categorical framework for Classical and Quantum Information Geometry: from fields of covariances to the geometry of statistical models
Thursday the 26th of November, 2026, 14:00, Room 2.2.D08
In both classical statistics and quantum information theory, the geometry of a statistical model is described by a Riemannian metric that measures how easily nearby states can be distinguished. Its classical prototype, the Fisher–Rao metric, is built from the covariance of the score. More generally, classical and quantum monotone metrics are closely related to covariances of observables.What if some of the differences between classical and quantum Information Geometry become clearer when we take covariances as our starting point?This change of perspective sheds new light on an old puzzle. In the classical case, Čencov showed that the Fisher–Rao metric is, up to a constant, the only metric compatible with natural probabilistic transformations. In the quantum case, Morozova, Čencov and Petz characterised an infinite family of monotone metrics on faithful states. On pure states, however, unitary invariance singles out the Fubini–Study metric, up to scale. Understanding the structural origin of this contrast is one of the motivations for our approach.Covariances offer a common framework: unlike metrics, they need no manifold, only a state and its observables. Classical and quantum states can both be described as normal states on operator algebras, while Markov maps and quantum channels can be represented by unital completely positive maps. The GNS construction associates a Hilbert space with each state, whose inner product reduces to the statistical covariance on centred random variables in the classical case. Deforming this inner product while preserving compatibility with probabilistic transformations leads to the notion of a field of covariances.In finite dimensions, we classify continuous fields of covariances, revealing how the modular structure of states governs the contrast between classical rigidity and quantum freedom, even on non-commutative algebras. We then develop a complementary construction based on the GNS fibration, which induces geometry on suitable statistical models, including infinite-dimensional ones, without requiring the state space to be a manifold. This approach recovers the Fisher–Rao, Fubini–Study and Bures–Helstrom metrics, together with geometric two-forms that, in the quantum setting, is related to the Berry curvature on pure states and the mean Uhlmann curvature on faithful states.
Laura Saenz Díez (UC3M-ICMAT)
A graph theoretic model for Stone dynamical systems
Thursday the 12th of November, 2026, 14:00, Room 2.2.D08
We present a graph-theoretic model for dynamical systems (X, σ), where X is a second countable Stone space and σ is a k-tuple of (not necessarily commuting) homeomorphisms of X. More precisely, we show that such dynamical systems are in bijective correspondence with a class of coloured Bratteli graphs. We further provide a categorical framework for both classes of objects, describing this correspondence as an equivalence of categories. Finally, we characterize, in terms of the graph, when the k-tuple consists of pairwise commuting homeomorphisms.This is joint work with Joan Bosa (UAB) and Joan Claramunt (UPC).
María Jiménez (UC3M-ICMAT)
Groupoid equivariant neural networks: An approach for steerable CNNs on bounded domains
Thursday the 22nd of October, 2026, 14:00, Room 2.2.D08
Equivariance is a powerful inductive bias in machine learning, allowing neural networks to incorporate known symmetries of the data directly into their architecture. In this talk, we present a groupoid-based framework for constructing equivariant neural networks, in which symmetries can act locally and may have domains of definition with boundaries. We develop the mathematical framework underlying these networks using groupoid representations and introduce the notion of local rigid bisections, which provide coherent transformations between regions of the domain. We then illustrate the framework through an example involving the action groupoid of the Euclidean group on the plane, restricted to a domain with boundary. By explicitly taking the boundary structure into account, the resulting tangent-cone groupoid may have several connected components, leading to additional constraints on the structure of the equivariant channels. Joint work with prof. Alberto Ibort and prof. Juan Manuel Pérez Pardo. arXiv:2609.25987
A. Ibort (UC3M-ICMAT)
Random perturbations of flows and solutions of the Navier-Stokes equation
Thursday the 8th of October, 2026, 14:00, Room 2.2.D08
We show that solutions of the Navier-Stokes equation for an incompressible fluid can be obtained by averaging stochastic perturbations of solutions of Euler's equation. The average of a Brownian perturbation of a steady solution of the Euler equation defines, for small time, a time-dependent diffeomorphism whose generating vector field satisfies a generalized Navier-Stokes equation. The deviation of this generalized equation from the standard Navier-Stokes equation is controlled by a viscous residual whose leading term involves a transport--diffusion term. We identify a natural class of steady, parallel Euler flows for which this residual vanishes and the perturbed flow satisfies the exact Navier-Stokes equation. These results are placed in the context of recent developments including the stochastic Lagrangian representation of Constantin-Iyer, the SALT formulation of Holm, and regularization-by-noise results of Flandoli and collaborators.Link for online session: http://meet.google.com/vpk-ijfj-nov
F. Lledó (UC3M-ICMAT)
Superselection theory of the Quantum Double with boundaries
Thursday the 1st of October, 2026, 14:00, Room 2.2.D08
In this talk I will present the superselection structure of the quantum double model (a natural generalization of Kitaev's celebrated toric code) on the lattice $Z^2$. I will also show how boundaries modify this structure (based on some ongoing work with Joan Claramunt (UPC) and Laura Sáenz (UC3M-ICMAT)).