arXiv subjects

Sandra Nestler

Publications and source records attributed to Sandra Nestler.

9 recordsLinked to original sources

arXiv2026

Power-Law Suppression of Superfluid Stiffness in High-Kinetic-Inductance NbN Films

Disorder is a powerful route to high kinetic inductance in superconducting ultrathin films, enabling compact high-impedance quantum circuits. This functionality, however, comes at the cost of reduced phase rigidity and potentially anomalous electrodynamics. Here, we use NbN microwave resonators with thicknesses down to 2.8 nm and sheet kinetic inductance up to 300 pH per square to probe how this trade-off reshapes the superconducting response. In the thinnest films, transport shows signatures of a Berezinskii-Kosterlitz-Thouless transition, while the microwave response reveals a pronounced low-temperature power-law suppression of the superfluid stiffness, inconsistent with Mattis-Bardeen theory. With increasing thickness, this anomalous regime is progressively suppressed, marking a continuous crossover toward conventional, gap-dominated electrodynamics. Cross-sectional transmission electron microscopy reveals a nanocrystalline twin-domain structure, pointing to oriented microstructural disorder as a crucial factor in the observed response. Overall, the crossover is governed by the ratio of superfluid stiffness to pairing energy, Theta(0)/Tc, identifying this ratio as a parameter governing the boundary between phase-fluctuation-dominated and gap-dominated superconducting electrodynamics in disordered nanofilms.

cond-mat.supr-con↗

arXiv2025

Characterizing Neural Manifolds' Properties and Curvatures using Normalizing Flows

Neuronal activity is found to lie on low-dimensional manifolds embedded within the high-dimensional neuron space. Variants of principal component analysis are frequently employed to assess these manifolds. These methods are, however, limited by assuming a Gaussian data distribution and a flat manifold. In this study, we introduce a method designed to satisfy three core objectives: (1) extract coordinated activity across neurons, described either statistically as correlations or geometrically as manifolds; (2) identify a small number of latent variables capturing these structures; and (3) offer an analytical and interpretable framework characterizing statistical properties by a characteristic function and describing manifold geometry through a collection of charts. To this end, we employ Normalizing Flows (NFs), which learn an underlying probability distribution of data by an invertible mapping between data and latent space. Their simplicity and ability to compute exact likelihoods distinguish them from other generative networks. We adjust the NF's training objective to distinguish between relevant (in manifold) and noise dimensions (out of manifold). Additionally, we find that different behavioral states align with the components of the latent Gaussian mixture model, enabling their treatment as distinct curved manifolds. Subsequently, we approximate the network for each mixture component with a quadratic mapping, allowing us to characterize both neural manifold curvature and non-Gaussian correlations among recording channels. Applying the method to recordings in macaque visual cortex, we demonstrate that state-dependent manifolds are curved and exhibit complex statistical dependencies. Our approach thus enables an expressive description of neural population activity, uncovering non-linear interactions among groups of neurons.

q-bio.NC↗

arXiv2025

Terahertz field effect in a two-dimensional semiconductor MoS2

Layered two-dimensional (2D) materials, with their atomic-scale thickness and tunable electronic, optical, and mechanical properties, open many promising pathways to significantly advance modern electronics. The field effect caused by a strong electric field, typically of MV/cm level, applied perpendicular to the material layers, is a highly effective method for controlling these properties. Field effect allows the regulation of the electron flow in transistor channels, improves the photodetector efficiency and spectral range, and facilitates the exploration of novel exotic quantum phenomena in 2D materials. However, existing approaches to induce the field effect in 2D materials utilize circuit-based electrical gating methods fundamentally limited to microwave response rates. Device-compatible ultrafast, sub-picosecond control needed for modern technology and basic science applications still remains a challenge. In this study, we demonstrate such an ultrafast field effect in atomically thin MoS2, an archetypal 2D semiconductor, embedded in a hybrid 3D-2D terahertz nanoantenna structure. This nanoantenna efficiently converts an incident terahertz electric field into the vertical ultrafast gating field in MoS2 while simultaneously enhancing it to the required MV/cm level. We observe the terahertz field effect optically as coherent terahertz-induced Stark shift of characteristic exciton resonances in MoS2. Our results enable novel developments in technology and the fundamental science of 2D materials, where the terahertz field effect is crucial.

cond-mat.mtrl-sci↗

arXiv2025

Bright quantum dot light sources using monolithic microlenses on gold back-reflectors

We present the fabrication process of bright $GaAs$ quantum dot (QD) photon sources by non-deterministic embedding into broadband monolithic $Al_{0.15}Ga_{0.85}As$ microlens arrays on gold-coated substrates. Arrays of cylindrical photoresist templates, with diameters ranging from $2$ $μm$ to $5$ $μm$, are thermally reflowed and subsequently transferred into the $Al_{0.15}Ga_{0.85}As$ thin-film semiconductor heterostructure with embedded quantum dots through an optimized anisotropic and three-dimensional shape-preserving reactive ion etching process. This methodology facilitated the fabrication of large-scale ($2$ $mm$ $\times$ $4$ $mm$) and densely packed arrays of uniformly shaped microlenses ($\sim$ $40 \times 10^3$ $mm^{-1}$), with the brightest emissions from QDs embedded in microlenses exhibiting lateral diameters and heights of $2.7$ $μm$ and $1.35$ $μm$, respectively. Finite-difference time-domain simulations of both idealized and fabricated lens shapes provide a comprehensive three-dimensional analysis of the device performance and optimization potentials such as anti-reflection coatings. It is found that free-space extraction (fiber-coupled) efficiencies of up to $62$ $\%$ ($37$ $\%$) are achievable for hemispherical QD-microlenses on gold-coated substrates. A statistical model for the fabrication yield of QD-microlenses is developed and experimentally corroborated by photoluminescence spectroscopy of fabricated microlens arrays. This analysis exhibited a free-space intensity enhancement by factors of up to $\times 200$ in approximately $1$ out of $200$ microlenses, showing good agreement to the theoretical expectations. This scalable fabrication strategy underscores the potential of these compact, high-efficiency sources offering new prospects for applications of these devices in future large-scale quantum networks.

cond-mat.mes-hall↗

arXiv2023

Statistical physics, Bayesian inference and neural information processing

Lecture notes from the course given by Professor Sara A. Solla at the Les Houches summer school on "Statistical physics of Machine Learning". The notes discuss neural information processing through the lens of Statistical Physics. Contents include Bayesian inference and its connection to a Gibbs description of learning and generalization, Generalized Linear Models as a controlled alternative to backpropagation through time, and linear and non-linear techniques for dimensionality reduction.

cond-mat.dis-nn↗

arXiv2023

Learning Interacting Theories from Data

One challenge of physics is to explain how collective properties arise from microscopic interactions. Indeed, interactions form the building blocks of almost all physical theories and are described by polynomial terms in the action. The traditional approach is to derive these terms from elementary processes and then use the resulting model to make predictions for the entire system. But what if the underlying processes are unknown? Can we reverse the approach and learn the microscopic action by observing the entire system? We use invertible neural networks (INNs) to first learn the observed data distribution. By the choice of a suitable nonlinearity for the neuronal activation function, we are then able to compute the action from the weights of the trained model; a diagrammatic language expresses the change of the action from layer to layer. This process uncovers how the network hierarchically constructs interactions via nonlinear transformations of pairwise relations. We test this approach on simulated data sets of interacting theories. The network consistently reproduces a broad class of unimodal distributions; outside this class, it finds effective theories that approximate the data statistics up to the third cumulant. We explicitly show how network depth and data quantity jointly improve the agreement between the learned and the true model. This work shows how to leverage the power of machine learning to transparently extract microscopic models from data.

cond-mat.dis-nn↗

arXiv2023

Neuronal architecture extracts statistical temporal patterns

Neuronal systems need to process temporal signals. We here show how higher-order temporal (co-)fluctuations can be employed to represent and process information. Concretely, we demonstrate that a simple biologically inspired feedforward neuronal model is able to extract information from up to the third order cumulant to perform time series classification. This model relies on a weighted linear summation of synaptic inputs followed by a nonlinear gain function. Training both - the synaptic weights and the nonlinear gain function - exposes how the non-linearity allows for the transfer of higher order correlations to the mean, which in turn enables the synergistic use of information encoded in multiple cumulants to maximize the classification accuracy. The approach is demonstrated both on a synthetic and on real world datasets of multivariate time series. Moreover, we show that the biologically inspired architecture makes better use of the number of trainable parameters as compared to a classical machine-learning scheme. Our findings emphasize the benefit of biological neuronal architectures, paired with dedicated learning algorithms, for the processing of information embedded in higher-order statistical cumulants of temporal (co-)fluctuations.

q-bio.NC↗

arXiv2022

Path classification by stochastic linear recurrent neural networks

We investigate the functioning of a classifying biological neural network from the perspective of statistical learning theory, modelled, in a simplified setting, as a continuous-time stochastic recurrent neural network (RNN) with identity activation function. In the purely stochastic (robust) regime, we give a generalisation error bound that holds with high probability, thus showing that the empirical risk minimiser is the best-in-class hypothesis. We show that RNNs retain a partial signature of the paths they are fed as the unique information exploited for training and classification tasks. We argue that these RNNs are easy to train and robust and back these observations with numerical experiments on both synthetic and real data. We also exhibit a trade-off phenomenon between accuracy and robustness.

stat.ML↗

arXiv2020

Unfolding recurrence by Green's functions for optimized reservoir computing

Cortical networks are strongly recurrent, and neurons have intrinsic temporal dynamics. This sets them apart from deep feed-forward networks. Despite the tremendous progress in the application of feed-forward networks and their theoretical understanding, it remains unclear how the interplay of recurrence and non-linearities in recurrent cortical networks contributes to their function. The purpose of this work is to present a solvable recurrent network model that links to feed forward networks. By perturbative methods we transform the time-continuous, recurrent dynamics into an effective feed-forward structure of linear and non-linear temporal kernels. The resulting analytical expressions allow us to build optimal time-series classifiers from random reservoir networks. Firstly, this allows us to optimize not only the readout vectors, but also the input projection, demonstrating a strong potential performance gain. Secondly, the analysis exposes how the second order stimulus statistics is a crucial element that interacts with the non-linearity of the dynamics and boosts performance.

cond-mat.dis-nn↗