HAL - L’archive ouverte pluridisciplinaire

Articles

• [hal-05381671] Application of Convolution Neural Network for Unfolding Simulated Neutron Spectra of an Activation Spectrometer

  27 novembre 2025, par ano.nymous@ccsd.cnrs.fr.invalid (Rodayna Hmede), Rodayna Hmede
  Neutron studies are of significant interest in fields such as radiation protection, nuclear reactor physics, and criticality safety, where accurate determination of the neutron field is essential. Determining the neutron field typically involves unfolding the detector signals, such as those obtained from the Activation and Counting Neutron Spectrometer (SNAC) detector. Traditional methods, as Bayesian approaches, have been widely integrated for the neutron spectrum unfolding. However, these methods rely on an initial solution estimation, introducing biases or uncertainties. Recent studies in Artificial Intelligence (AI) have demonstrated its potential to address challenges as hysteresis regression. This work is based on our novel convolutional neural network (CNN) architecture to overcome the hysteresis problem in neutron spectrum unfolding. The CNN model predicts the neutron spectrum directly from detector counts, eliminating the need for prior solution predictions. The proposed architecture was trained on a large simulation dataset and validated through a combination of Serpent simulations of various Californium-252 (Cf) spectra and Monte Carlo N-Particles (MCNP) simulations of the Silene reactor. These two complementary simulation approaches are used to evaluate the CNN's evaluation in realistic neutron environments. The results demonstrate the model's high efficiency and accuracy, as evidenced by key performance metrics and the quality of the predicted spectrum (SQ). This approach represents a significant step forward in optimizing and validating AI-based methods for neutron field, especially in criticality dosimetry and radiation protection applications. Preliminary comparisons with Bayesian unfolding codes already indicate than CNN based predictions can capture fine spectral features. A benchmark against MAXED, GRAVEL and Nubay remains a key perspective, together with validation campaigns on neutron facilities.

• [hal-04979046] Optimal Almost Sure Rate of Convergence for the Wavelets Estimator in the Partially Linear Additive Models

  7 novembre 2025, par ano.nymous@ccsd.cnrs.fr.invalid (Khalid Chokri), Khalid Chokri
  In this article, we examine a class of partially linear additive models (PLAM) defined via a measurable mapping Ψ:Rq→R. More precisely, we consider Ψ(Yi):=Yi=Zi⊤β+∑l=1dml(Xl,i)+εi,i=1,…,n, where Zi=(Zi,1,…,Zi,p)⊤ and Xi=(X1,i,…,Xd,i)⊤ denote vectors of explanatory variables. The unknown parameter vector is β=(β1,…,βp)⊤, and m1,…,md are real-valued functions of a single variable whose forms are not specified. The error terms ε1,…,εn are identically distributed with mean zero and finite variance σε, and they fulfill the condition E(ε∣X,Z)=0 almost surely. These models are broadly applicable in finance, biology, and engineering, where capturing intricate nonlinear effects is essential. We propose an estimation method that leverages marginal integration in conjunction with linear wavelet-based techniques to obtain estimators for the unknown components m1,…,md. Under suitable regularity conditions, we establish strong uniform convergence of these estimators, demonstrating that they achieve practically relevant convergence rates. Our theoretical results indicate that these estimators converge uniformly at rates that are favorable for practical applications, underscoring the adaptability and scope of this partially linear additive model.

• [hal-04946880] A nonparametric distribution-free test of independence among continuous random vectors based on L1-norm

  5 novembre 2025, par ano.nymous@ccsd.cnrs.fr.invalid (Nour-Eddine Berrahou), Nour-Eddine Berrahou
  We propose a novel statistical test to assess the mutual independence of multidimensional random vectors. Our approach is based on the L 1 -distance between the joint density function and the product of the marginal densities associated with the presumed independent vectors. Under the null hypothesis, we employ Poissonization techniques to establish the asymptotic normal approximation of the corresponding test statistic, without imposing any regularity assumptions on the underlying Lebesgue density function, denoted as f ( ⋅ ) . Remarkably, we observe that the limiting distribution of the L 1 -based statistics remains unaffected by the specific form of f ( ⋅ ) . This unexpected result contributes to the robustness and versatility of our method. Moreover, our tests exhibit nontrivial local power against a subset of local alternatives, which converge to the null hypothesis at a rate of n − 1 ∕ 2 h − d ∕ 4 n , d ≥ 2 , where n represents the sample size and h n denotes the bandwidth. Finally, the theory is supported by a comprehensive simulation study to investigate the finite-sample performance of our proposed test. The results demonstrate that our testing procedure generally outperforms existing approaches across various examined scenarios.

• [hal-04844328] A finite volume scheme for the local sensing chemotaxis model

  6 octobre 2025, par ano.nymous@ccsd.cnrs.fr.invalid (Maxime Herda), Maxime Herda
  In this paper we design, analyze and simulate a finite volume scheme for a cross-diffusion system which models chemotaxis with local sensing. This system has the same Lyapunov function (or entropy) as the celebrated minimal Keller-Segel system, but unlike the latter, its solutions are known to exist globally in 2D. The long-time behavior of solutions is only partially understood which motivates numerical exploration with a reliable numerical method. We propose a linearly implicit, two-point flux finite volume approximation of the system. We show that the scheme preserves, at the discrete level, the main features of the continuous system, namely mass conservation, non-negativity of solution, entropy dissipation, and duality estimates. These properties allow us to prove the well-posedness, unconditional stability and convergence of the scheme. We also show rigorously that the scheme possesses an asymptotic preserving (AP) property in the quasi-stationary limit. We complement our analysis with thorough numerical experiments investigating convergence and AP properties of the scheme as well as its reliability with respect to stability properties of steady solutions.

0 | 5 | 10