arXiv subjects

Giuseppe Curci

Publications and source records attributed to Giuseppe Curci.

10 recordsLinked to original sources

arXiv2026

Non-reciprocity and exchange-spring delay of domain-wall Walker breakdown in magnetic nanowires with azimuthal magnetization

Domain-wall (DW) motion is a crucial process involved in magnetization reversal, be it under magnetic field or spin-polarized current stimulus. In most cases the DW speed does not exceed $\approx$100m/s and collapses above a given threshold of the stimulus, an effect known as Walker breakdown. A few specific material properties have been identified to delay the breakdown of speed by increasing the energy barrier preventing internal precession. We show that in a 3D nanomagnetic system, here with vortex-state domains, the topology of the magnetization distribution may intrinsically and robustly delay the Walker breakdown due to an exchange-spring effect. However, in contrast to situations in lower dimension, the motion remains steady-state. In addition, curvature induces a major non-reciprocal effect, delaying or not the Walker breakdown depending on the sequence of chiralities of the azimuthal domains on both sides of the DW, independent of its direction of motion.

cond-mat.mes-hall↗

arXiv2025

Ensembling Pruned Attention Heads For Uncertainty-Aware Efficient Transformers

Uncertainty quantification (UQ) is essential for deploying deep neural networks in safety-critical settings. Although methods like Deep Ensembles achieve strong UQ performance, their high computational and memory costs hinder scalability to large models. We introduce Hydra Ensembles, an efficient transformer-based ensemble that prunes attention heads to create diverse members and merges them via a new multi-head attention with grouped fully-connected layers. This yields a compact model with inference speed close to a single network, matching or surpassing Deep Ensembles in UQ performance without retraining from scratch. We also provide an in-depth analysis of pruning, showing that naive approaches can harm calibration, whereas Hydra Ensembles preserves robust uncertainty. Experiments on image and text classification tasks, with various architectures, show consistent gains over Deep Ensembles. Remarkably, in zero-shot classification on ImageNet-1k, our approach surpasses state of the art methods, even without requiring additional training.

cs.LG↗

arXiv1998

Simulation of Supersymmetric Models with a Local Nicolai Map

We study the numerical simulation of supersymmetric models having a local Nicolai map. The mapping can be regarded as a stochastic equation and its numerical integration provides an algorithm for the simulation of the original model. In this paper, the method is discussed in details and applied to examples in 0+1 and 1+1 dimensions.

hep-lat↗

arXiv1997

Adaptive Identification of VIRGO-like Noise Spectrum

The aim of this work is to show how it is possible to build an on line whitening filter in an adaptive way. We have modeled the VIRGO noise spectrum as an autoregressive stochastic process, after a pre-filtering of the theoretical curve which flattens the low frequency part of the spectrum. We have tested some very popular adaptive algorithms, based on the gradient methods and on the least squares methods with a lattice structure filter.

gr-qc↗

arXiv1994

QCD corrections to electroweak processes in an unconventional scheme: application to the b -> s gamma decay

In this work we give a detailed description of a method for the calculation of QCD corrections to electroweak processes in dimensional regularization that does not require any definition of the gamma_5 matrix in d dimensions. This method appears particularly convenient in order to limit the algebraic complexity of higher order calculations. As an example, we compute the leading logarithmic corrections to the b -> s gamma decay.

hep-ph↗

arXiv1994

Numerical simulation of the Kardar-Parisi-Zhang equation

We simulate the Kardar-Parisi-Zhang equation in 2+1 dimensions. The Hopf-Cole transformation is used in order to obtain a stable numerical scheme. The two relevant critical exponents are precisely measured. (2 PostScript figures available from the authors)

hep-lat↗

arXiv1993

The Kramers equation simulation algorithm I. Operator analysis

Using an operatorial formalism, we study the Kramers equation and its applications to numerical simulations. We obtain classes of algorithms which may be made precise at every desired order in the time step $ε$ and with a set of free parameters which can be used to reduce autocorrelations. We show that it is possible to use a global Metropolis test to restore Detailed Balance.

hep-lat↗

arXiv1993

The Kramers equation simulation algorithm II. An application to the Gross-Neveu model

We continue the investigation on the applications of the Kramers equation to the numerical simulation of field theoretic models. In a previous paper we have described the theory and proposed various algorithms. Here, we compare the simplest of them with the Hybrid Monte Carlo algorithm studying the two-dimensional lattice Gross-Neveu model. We used a Symanzik improved action with dynamical Wilson fermions. Both the algorithms allow for the determination of the critical mass. Their performances in the definite phase simulations are comparable with the Hybrid Monte Carlo. For the two methods, the numerical values of the measured quantities agree within the errors and are compatible with the theoretical predictions; moreover, the Kramers algorithm is safer from the point of view of the numerical precision.

hep-lat↗

arXiv1993

Scaling, asymptotic scaling and Symanzik improvement. Deconfinement temperature in SU(2) pure gauge theory

We report on a high statistics simulation of SU(2) pure gauge field theory at finite temperature, using Symanzik action. We determine the critical coupling for the deconfinement phase transition on lattices up to 8 x 24, using Finite Size Scaling techniques. We find that the pattern of asymptotic scaling violation is essentially the same as the one observed with conventional, not improved action. On the other hand, the use of effective couplings defined in terms of plaquette expectation values shows a precocious scaling, with respect to an analogous analysis of data obtained by the use of Wilson action, which we interpret as an effect of improvement.

hep-lat↗

arXiv1992

Radiative Correction Effects of a Very Heavy Top

If the top is very heavy, m_t >> M_Z, the dominant radiative correction effects in all electroweak precision tests can be exactly characterized in terms of two quantities, the rho-parameter and the GIM violating Z -> b bbar coupling. These quantities can be computed using the Standard Model Lagrangian with vanishing gauge couplings. This is done here up to two loops for arbitrary values of the Higgs mass.

hep-ph↗