Francesco Ticozzi

Tommaso Grigoletto, Alain Sarlette, Francesco Ticozzi et al.

We present an algebraic framework for approximate model reduction of Markovian open quantum dynamics that guarantees complete positivity and trace preservation by construction. First, we show that projecting a Lindblad generator on its center manifold -- the space spanned by eigenoperators with purely imaginary eigenvalue -- yields an asymptotically exact reduced quantum dynamical semigroup whose dynamics is unitary, with exponentially decaying transient error controlled by the generator's spectral gap. Second, for analytic perturbations of a Lindblad generator with a tractable center manifold, we propose a perturbative reduction that keeps the reduced space fixed at the unperturbed center manifold. The resulting generator is shown to remain a valid Lindbladian for arbitrary perturbation strengths, and explicit finite-time error bounds, that quantify leakage from the unperturbed center sector, are provided. We further clarify the connection to adiabatic elimination methods, by both showing how the algebraic reduction can be directly related to a first-order adiabatic-elimination and by providing sufficient conditions under which the latter method can be applied while preserving complete positivity. We showcase the usefulness of our techniques in dissipative many-body quantum systems exhibiting non-stationary long-time dynamics.

Tommaso Grigoletto, Francesco Ticozzi

The problem of reducing a Hidden Markov Model (HMM) to one of smaller dimension that exactly reproduces the same marginals is tackled by using a system-theoretic approach. Realization theory tools are extended to HMMs by leveraging suitable algebraic representations of probability spaces. We propose two algorithms that return coarse-grained equivalent HMMs obtained by stochastic projection operators: the first returns models that exactly reproduce the single-time distribution of a given output process, while in the second the full (multi-time) distribution is preserved. The reduction method exploits not only the structure of the observed output, but also its initial condition, whenever the latter is known or belongs to a given subclass. Optimal algorithms are derived for a class of HMM, namely observable ones.

Riccardo Zattra, Giacomo Baggio, Umberto Casti et al.

Transformers, powered by the attention mechanism, are the backbone of most foundation models, yet they suffer from quadratic complexity and difficulties in dealing with long-range dependencies in the input sequence. Recent work has shown that state space models (SSMs) provide an efficient alternative, with the S6 module at the core of the Mamba architecture achieving state-of-the-art results on long-sequence benchmarks. In this paper, we introduce the COFFEE (COntext From FEEdback) model, a novel time-varying SSM that incorporates state feedback to enable context-dependent selectivity, while still allowing for parallel implementation. Whereas the selectivity mechanism of S6 only depends on the current input, COFFEE computes it from the internal state, which serves as a compact representation of the sequence history. This shift allows the model to regulate its dynamics based on accumulated context, improving its ability to capture long-range dependencies. In addition to state feedback, we employ an efficient model parametrization that removes redundancies present in S6 and leads to a more compact and trainable formulation. On the induction head task, COFFEE achieves near-perfect accuracy with two orders of magnitude fewer parameters and training sequences compared to S6. On MNIST, COFFEE largely outperforms S6 within the same architecture, reaching 97% accuracy with only 3585 parameters. These results showcase the role of state feedback as a key mechanism for building scalable and efficient sequence models.