IST Austria: PubRep (Institute of Science and Technology)
Not a member yet
6140 research outputs found
Sort by
Efficient identification of wide shallow neural networks with biases
No full textThe identification of the parameters of a neural network from finite samples of input-output pairs is often referred to as the teacher-student model, and this model has represented a popular framework for understanding training and generalization. Even if the problem is NP-complete in the worst case, a rapidly growing literature – after adding suitable distributional assumptions – has established finite sample identification of two-layer networks with a number of neurons (math. formula), D being the input dimension. For the range (math. formula) the problem becomes harder, and truly little is known for networks parametrized by biases as well. This paper fills the gap by providing efficient algorithms and rigorous theoretical guarantees of finite sample identification for such wider shallow networks with biases. Our approach is based on a two-step pipeline: first, we recover the direction of the weights, by exploiting second order information; next, we identify the signs by suitable algebraic evaluations, and we recover the biases by empirical risk minimization via gradient descent. Numerical results demonstrate the effectiveness of our approach
Ionic association and Wien effect in 2D confined electrolytes
No full textRecent experimental advances in nanofluidics have allowed to explore ion transport across molecular-scale pores, in particular, for iontronic applications. Two-dimensional nanochannels—in which a single molecular layer of electrolyte is confined between solid walls—constitute a unique platform to investigate fluid and ion transport in extreme confinement, highlighting unconventional transport properties. In this work, we study ionic association in 2D nanochannels, and its consequences on non-linear ionic transport, using both molecular dynamics simulations and analytical theory. We show that under sufficient confinement, ions assemble into pairs or larger clusters in a process analogous to a Kosterlitz–Thouless transition, here modified by the dielectric confinement. We further show that the breaking of pairs results in an electric-field dependent conduction, a mechanism usually known as the second Wien effect. However the 2D nature of the system results in non-universal, temperature-dependent, scaling of the conductivity with electric field, leading to ionic coulomb blockade in some regimes. A 2D generalization of the Onsager theory fully accounts for the non-linear transport. These results suggest ways to exploit electrostatic interactions between ions to build new nanofluidic devices
¹⁸⁷Os nuclear resonance scattering to explore hyperfine interactions and lattice dynamics for biological applications
No full textOsmium complexes with osmium in different oxidation states (II, III, IV, and VI) have been reported to exhibit antiproliferative activity in cancer cell lines. Herein, we demonstrate unexplored opportunities offered by 187Os nuclear forward scattering (NFS) and nuclear inelastic scattering (NIS) of synchrotron radiation for characterization of hyperfine interactions and lattice dynamics in a benchmark Os(VI) complex, K2[OsO2(OH)4]. We determined the isomer shift [δ = 3.3(1) millimeters per second] relative to [OsIVCl6]2− and quadrupole splitting [ΔEQ = 12.0(2) millimeters per second] with NFS. We estimated the Lamb-Mössbauer factor [0.80(4)], extracted the density of phonon states, and carried out a thermodynamics characterization using the NIS data combined with first-principles calculations. Overall, we provide evidence that 187Os nuclear resonance scattering is a reliable technique for the investigation of hyperfine interactions and Os-specific vibrations in osmium(VI) species and is thus applicable for such measurements in osmium complexes of other oxidation states, including those with anticancer activity such as Os(III) and Os(IV)
Observation of collapse and revival in a superconducting atomic frequency comb
No full textRecent advancements in superconducting circuits have enabled the experimental study of collective behavior of precisely controlled intermediate-scale ensembles of qubits. In this work, we demonstrate an atomic frequency comb formed by individual artificial atoms strongly coupled to a single resonator mode. We observe periodic microwave pulses that originate from a single coherent excitation dynamically interacting with the multiqubit ensemble. We show that this revival dynamics emerges as a consequence of the constructive and periodic rephasing of the five superconducting qubits forming the vacuum Rabi split comb. In the future, similar devices could be used as a memory with in situ tunable storage time or as an on-chip periodic pulse generator with nonclassical photon statistics
Strong charge-photon coupling in planar germanium enabled by granular aluminium superinductors
No full textHigh kinetic inductance superconductors are gaining increasing interest for the realisation of qubits, amplifiers and detectors. Moreover, thanks to their high impedance, quantum buses made of such materials enable large zero-point fluctuations of the voltage, boosting the coupling rates to spin and charge qubits. However, fully exploiting the potential of disordered or granular superconductors is challenging, as their inductance and, therefore, impedance at high values are difficult to control. Here, we report a reproducible fabrication of granular aluminium resonators by developing a wireless ohmmeter, which allows in situ measurements during film deposition and, therefore, control of the kinetic inductance of granular aluminium films. Reproducible fabrication of circuits with impedances (inductances) exceeding 13 kΩ (1 nH per square) is now possible. By integrating a 7.9 kΩ resonator with a germanium double quantum dot, we demonstrate strong charge-photon coupling with a rate of gc/2π = 566 ± 2 MHz. This broadly applicable method opens the path for novel qubits and high-fidelity, long-distance two-qubit gates
On the L-polynomials of curves over finite fields
No full textWe discuss, in a non-Archimedean setting, the distribution of the coefficients of L-polynomials of curves of genus g over Fq . Among other results, this allows us to prove that the Q-vector space spanned by such characteristic polynomials has dimension g + 1. We also state a conjecture about the Archimedean distribution of the number of rational points of curves over finite fields
Research data for the publication "Learning reshapes the hippocampal representation hierarchy"
No full textResearch data for the article "Learning reshapes the hippocampal representation hierarchy" from Chiossi et al. (PNAS, 2025). The data includes hippocampal CA1 unit activity and behaviour tracking of 5 Long Evans rats during the learning of an associative memory task. Detailed information can be found in the 'readme.txt' file
DNA methylation-based predictors of metabolic traits in Scottish and Singaporean cohorts
No full textExploring the molecular correlates of metabolic health measures may identify their shared and unique biological processes and pathways. Molecular proxies of these traits may also provide a more objective approach to their measurement. Here, DNA methylation (DNAm) data were used in epigenome-wide association studies (EWASs) and for training epigenetic scores (EpiScores) of six metabolic traits: body mass index (BMI), body fat percentage, waist-hip ratio, and blood-based measures of glucose, high-density lipoprotein cholesterol, and total cholesterol in >17,000 volunteers from the Generation Scotland (GS) cohort. We observed a maximum of 12,033 significant findings (p < 3.6 × 10−8) for BMI in a marginal linear regression EWAS. By contrast, a joint and conditional Bayesian penalized regression approach yielded 27 high-confidence associations with BMI. EpiScores trained in GS performed well in both Scottish and Singaporean test cohorts (Lothian Birth Cohort 1936 [LBC1936] and Health for Life in Singapore [HELIOS]). The EpiScores for BMI and total cholesterol performed best in HELIOS, explaining 20.8% and 7.1% of the variance in the measured traits, respectively. The corresponding results in LBC1936 were 14.4% and 3.2%, respectively. Differences were observed in HELIOS for body fat, where the EpiScore explained ∼9% of the variance in Chinese and Malay -subgroups but ∼3% in the Indian subgroup. The EpiScores also correlated with cognitive function in LBC1936 (standardized βrange: 0.08–0.12, false discovery rate p [pFDR] < 0.05). Accounting for the correlation structure across the methylome can vastly affect the number of lead findings in EWASs. The EpiScores of metabolic traits are broadly applicable across populations and can reflect differences in cognition
Morse predecomposition of an invariant set
No full textMotivated by the study of recurrent orbits and dynamics within a Morse set of a Morse decomposition we introduce the concept of Morse predecomposition of an isolated invariant set within the setting of both combinatorial and classical dynamical systems. While Morse decomposition summarizes solely the gradient part of a dynamical system, the developed generalization extends to the recurrent component as well. In particular, a chain recurrent set, which is indecomposable in terms of Morse decomposition, can be represented more finely in the Morse predecomposition framework. This generalization is achieved by forgoing the poset structure inherent to Morse decomposition and relaxing the notion of connection between Morse sets (elements of Morse decomposition) in favor of what we term ’links’. We prove that a Morse decomposition is a special case of Morse predecomposition indexed by a poset. Additionally, we show how a Morse predecomposition may be condensed back to retrieve a Morse decomposition
Comparison of renormalized interactions using one-dimensional few-body systems as a testbed
No full textEven though the one-dimensional contact interaction requires no regularization, renormalization methods have been shown to improve the convergence of numerical calculations considerably. In this work, we compare and contrast these methods: “the running coupling constant” where the two-body ground-state energy is used as a renormalization condition, and two effective interaction approaches that include information about the ground as well as excited states. In particular, we calculate the energies and densities of few-fermion systems in a harmonic oscillator with the configuration-interaction method and compare the results based upon renormalized and bare interactions. We find that the use of the running coupling constant instead of the bare interaction improves convergence significantly. A comparison with an effective interaction, which is designed to reproduce the relative part of the energy spectrum of two particles, showed a similar improvement. The effective interaction provides an additional improvement if the center-of-mass excitations are included in the construction. Finally, we discuss the transformation of observables alongside the renormalization of the potential, and demonstrate that this might be an essential ingredient for accurate numerical calculations