Publications

Order by: publication date number of citations

Exact full-RSB SAT/UNSAT transition in infinitely wide two-layer neural networks

Brandon L. Annesi, Enrico M. Malatesta, Francesco Zamponi

SciPost Phys. 18, 118 (2025) · published 2 April 2025 | · pdf

We analyze the problem of storing random pattern-label associations using two classes of continuous non-convex weights models, namely the perceptron with negative margin and an infinite-width two-layer neural network with non-overlapping receptive fields and generic activation function. Using a full-RSB Ansatz we compute the exact value of the SAT/UNSAT transition. Furthermore, in the case of the negative perceptron we show that the overlap distribution of typical states displays an overlap gap (a disconnected support) in certain regions of the phase diagram defined by the value of the margin and the density of patterns to be stored. This implies that some recent theorems that ensure convergence of Approximate Message Passing (AMP) based algorithms to capacity are not applicable. Finally, we show that Gradient Descent is not able to reach the maximal capacity, irrespectively of the presence of an overlap gap for typical states. This finding, similarly to what occurs in binary weight models, suggests that gradient-based algorithms are biased towards highly atypical states, whose inaccessibility determines the algorithmic threshold.
Full event particle-level unfolding with variable-length latent variational diffusion

Alexander Shmakov, Kevin Greif, Michael James Fenton, Aishik Ghosh, Pierre Baldi, Daniel Whiteson

SciPost Phys. 18, 117 (2025) · published 1 April 2025 | · pdf

The measurements performed by particle physics experiments must account for the imperfect response of the detectors used to observe the interactions. One approach, unfolding, statistically adjusts the experimental data for detector effects. Recently, generative machine learning models have shown promise for performing unbinned unfolding in a high number of dimensions. However, all current generative approaches are limited to unfolding a fixed set of observables, making them unable to perform full-event unfolding in the variable dimensional environment of collider data. A novel modification to the variational latent diffusion model (VLD) approach to generative unfolding is presented, which allows for unfolding of high- and variable-dimensional feature spaces. The performance of this method is evaluated in the context of semi-leptonic $t\bar{t}$ production at the Large Hadron Collider.
Photon counting statistics in Gaussian bosonic networks

Kalle Sulo Ukko Kansanen, Pedro Portugal, Christian Flindt, Peter Samuelsson

SciPost Phys. 18, 116 (2025) · published 1 April 2025 | · pdf

The statistics of transmitted photons in microwave cavities play a foundational role in microwave quantum optics and its technological applications. By utilizing quantum mechanical phase-space methods, we here develop a general theory of the photon counting statistics in Gaussian bosonic networks consisting of driven cavities with beamsplitter interactions and two-mode-squeezing. The dynamics of the network can be captured by a Lyapunov equation for the covariance matrix of the cavity fields, which generalizes to a Riccati equation, when counting fields are included. By solving the Riccati equation, we obtain the statistics of emitted and absorbed photons as well as the time-dependent correlations encoded in waiting time distributions and second-order coherence functions. To illustrate our theoretical framework, we first apply it to a simple linear network consisting of two coupled cavities, for which we evaluate the photon cross-correlations and discuss connections between the photon emission statistics and the entanglement between the cavities. We then consider a bosonic circulator consisting of three coupled cavities, for which we investigate how a synthetic flux may affect the direction of the photon flow, similarly to recent experiments. Our general framework paves the way for systematic investigations of the photon counting statistics in Gaussian bosonic networks.
How to escape atypical regions in the symmetric binary perceptron: A journey through connected-solutions states

Damien Barbier

SciPost Phys. 18, 115 (2025) · published 31 March 2025 | · pdf

We study the binary symmetric perceptron model, and in particular its atypical solutions. While the solution-space of this problem is dominated by isolated configurations, it is also solvable for a certain range of constraint density $\alpha$ and threshold $\kappa$ . We provide in this paper a statistical measure probing sequences of solutions, where two consecutive elements shares a strong overlap. After simplifications, we test its predictions by comparing it to Monte-Carlo simulations. We obtain good agreement and show that connected states with a Markovian correlation profile can fully decorrelate from their initialization only for $\kappa>\kappa_{\rm no-mem.\, state}$ ( $\kappa_{\rm no-mem.\, state}\sim \sqrt{0.91\log(N)}$ for $\alpha=0.5$ and $N$ being the dimension of the problem). For $\kappa<\kappa_{\rm no-mem.\, state}$ , we show that decorrelated sequences still exist but have a non-trivial correlations profile. To study this regime we introduce an Ansatz for the correlations that we label as the nested Markov chain.
SymTFT for (3+1)d gapless SPTs and obstructions to confinement

Andrea Antinucci, Christian Copetti, Sakura Schäfer-Nameki

SciPost Phys. 18, 114 (2025) · published 31 March 2025 | · pdf

We study gapless phases in (3+1)d in the presence of 1-form and non-invertible duality symmetries. Using the Symmetry Topological Field Theory (SymTFT) approach, we classify the gapless symmetry-protected (gSPT) phases in these setups, with particular focus on intrinsically gSPTs (igSPTs). These are symmetry protected critical points which cannot be deformed to a trivially gapped phase without spontaneously breaking the symmetry. Although these are by now well-known in (1+1)d, we demonstrate their existence in (3+1)d gauge theories. Here, they have a clear physical interpretation in terms of an obstruction to confinement, even though the full 1-form symmetry does not suffer from 't Hooft anomalies. These igSPT phases provide a new way to realize 1-form symmetries in CFTs, that has no analog for gapped phases. The SymTFT approach allows for a direct generalization from invertible symmetries to non-invertible duality symmetries, for which we study gSPT and igSPT phases as well. We accompany these theoretical results with concrete physical examples realizing such phases and explain how obstruction to confinement is detected at the level of symmetric deformations.
Renormalisation group effects in SMEFT for di-Higgs production

Gudrun Heinrich, Jannis Lang

SciPost Phys. 18, 113 (2025) · published 31 March 2025 | · pdf

We study the effects of renormalisation group running for the Wilson coefficients of dimension-6 operators contributing to Higgs boson pair production in gluon fusion within Standard Model Effective Field Theory (SMEFT). The running of these Wilson coefficients has been implemented in the NLO QCD code ggHH_SMEFT, which is publicly available within the Powheg-Box-V2 framework.
Neural quantum state study of fracton models

Marc Machaczek, Lode Pollet, Ke Liu

SciPost Phys. 18, 112 (2025) · published 28 March 2025 | · pdf

Fracton models host unconventional topological orders in three and higher dimensions and provide promising candidates for quantum memory platforms. Understanding their robustness against quantum fluctuations is an important task but also poses great challenges due to the lack of efficient numerical tools. In this work, we establish neural quantum states (NQS) as new tools to study phase transitions in these models. Exact and efficient parametrizations are derived for three prototypical fracton codes — the checkerboard and X-cube model, as well as Haah's code — both in terms of a restricted Boltzmann machine (RBM) and a correlation-enhanced RBM. We then adapt the correlation-enhanced RBM architecture to a perturbed checkerboard model and reveal its strong first-order phase transition between the fracton phase and a trivial field-polarizing phase. To this end, we simulate this highly entangled system on lattices of up to 512 qubits with high accuracy, representing a cutting-edge application of variational neural-network methods. In addition, we reproduce the phase transition of the X-cube model previously obtained with quantum Monte Carlo and high-order series expansion methods. Our work demonstrates the remarkable potential of NQS in studying complicated three-dimensional problems and highlights physics-oriented constructions of NQS architectures.
Scaling laws in jet classification

Joshua Batson, Yonatan Kahn

SciPost Phys. Core 8, 034 (2025) · published 28 March 2025 | · pdf

We demonstrate the emergence of scaling laws in the benchmark top versus QCD jet classification problem in collider physics. Six distinct physically-motivated classifiers exhibit power-law scaling of the binary cross-entropy test loss as a function of training set size, with distinct power law indices. This result highlights the importance of comparing classifiers as a function of dataset size rather than for a fixed training set, as the optimal classifier may change considerably as the dataset is scaled up. We speculate on the interpretation of our results in terms of previous models of scaling laws observed in natural language and image datasets.
Hierarchy of degenerate stationary states in a boundary-driven dipole-conserving spin chain

Apoorv Srivastava, Shovan Dutta

SciPost Phys. 18, 111 (2025) · published 27 March 2025 | · pdf

Kinetically constrained spin chains serve as a prototype for structured ergodicity breaking in isolated quantum systems. We show that such a system exhibits a hierarchy of degenerate steady states when driven by incoherent pump and loss at the boundary. By tuning the relative pump and loss and how local the constraints are, one can stabilize mixed steady states, noiseless subsystems, and various decoherence-free subspaces, all of which preserve large amounts of information. We also find that a dipole-conserving bulk suppresses current in steady state. These exact results based on the flow in Hilbert space hold regardless of the specific Hamiltonian or drive mechanism. Our findings show that a competition of kinetic constraints and local drives can induce different forms of ergodicity breaking in open systems, which should be accessible in quantum simulators.
Engineering unique localization transition with coupled Hatano-Nelson chains

Ritaban Samanta, Aditi Chakrabarty, Sanjoy Datta

SciPost Phys. Core 8, 033 (2025) · published 27 March 2025 | · pdf

The paradigmatic Hatano-Nelson (HN) Hamiltonian induces a delocalization-localization (DL) transition in a one-dimensional (1D) lattice with random disorder, in striking contrast to its Hermitian counterpart. The DL transition also persists in the presence of a quasiperiodic potential separating completely delocalized and localized eigenstates. In this study, we reveal that coupling two 1D quasiperiodic Hatano-Nelson (QHN) lattices signiﬁcantly alters the nature of the DL transition and identify two critical points, $V_{c1} < V_{c2}$ , when the nearest neighbors of the two 1D QHN lattices are cross-coupled with strong hopping amplitudes under periodic boundary conditions (PBC). Complete delocalization occurs below $V_{c1}$ and the states are completely localized above $V_{c2}$ , while two mobility edges symmetrically emerge about Re[E] = 0 between $V_{c1}$ and $V_{c2}$ . Notably, under speciﬁc asymmetric cross-hopping amplitudes, $V_{c1}$ approaches zero, resulting in localized states even for an inﬁnitesimally weak potential. Remarkably, we also ﬁnd that the mobility edges precisely divide the delocalized and localized states in equal proportions. We demonstrate a possible implementation of these findings in a coupled waveguided array which can be exploited to control and manipulate the light localization depending upon the hopping amplitude in the two QHN chains.

Previous 1 2 3 4 5 6 7 8 9 10 ... 324 Next