Showing 1–10 of 10 results for author: Annunziata, A

Enhancing Q&A with Domain-Specific Fine-Tuning and Iterative Reasoning: A Comparative Study

Authors: Zooey Nguyen, Anthony Annunziata, Vinh Luong, Sang Dinh, Quynh Le, Anh Hai Ha, Chanh Le, Hong An Phan, Shruti Raghavan, Christopher Nguyen

Abstract: This paper investigates the impact of domain-specific model fine-tuning and of reasoning mechanisms on the performance of question-answering (Q&A) systems powered by large language models (LLMs) and Retrieval-Augmented Generation (RAG). Using the FinanceBench SEC financial filings dataset, we observe that, for RAG, combining a fine-tuned embedding model with a fine-tuned LLM achieves better accura… ▽ More This paper investigates the impact of domain-specific model fine-tuning and of reasoning mechanisms on the performance of question-answering (Q&A) systems powered by large language models (LLMs) and Retrieval-Augmented Generation (RAG). Using the FinanceBench SEC financial filings dataset, we observe that, for RAG, combining a fine-tuned embedding model with a fine-tuned LLM achieves better accuracy than generic models, with relatively greater gains attributable to fine-tuned embedding models. Additionally, employing reasoning iterations on top of RAG delivers an even bigger jump in performance, enabling the Q&A systems to get closer to human-expert quality. We discuss the implications of such findings, propose a structured technical design space capturing major technical components of Q&A AI, and provide recommendations for making high-impact technical choices for such components. We plan to follow up on this work with actionable guides for AI teams and further investigations into the impact of domain-specific augmentation in RAG and into agentic AI capabilities such as advanced planning and reasoning. △ Less

Submitted 19 April, 2024; v1 submitted 17 April, 2024; originally announced April 2024.

Comments: Fixed typo of OODA's score on harder-question set in Table 2

High-order Harmonic Generation in Thermotropic Liquid Crystals

Authors: Andrea Annunziata, Luise Becker, Marta Luisa Murillo-Sánchez, Patrick Friebel, Salvatore Stagira, Davide Faccialà, Caterina Vozzi, Laura Cattaneo

Abstract: Thermotropic liquid crystals are versatile optical materials that exhibit a state of matter intermediate between liquids and solids. Their properties can change significantly with temperature, pressure, or other external factors, leading to different phases. The transport properties within these materials in different phases are still largely unexplored and their understanding would enable excitin… ▽ More Thermotropic liquid crystals are versatile optical materials that exhibit a state of matter intermediate between liquids and solids. Their properties can change significantly with temperature, pressure, or other external factors, leading to different phases. The transport properties within these materials in different phases are still largely unexplored and their understanding would enable exciting prospects for innovative technological advancements. High-order harmonic spectroscopy proved to be a powerful spectroscopic tool for investigating the electronic and nuclear dynamics in matter. Here we report the first experimental observation of high-order harmonic generation in thermotropic liquid crystals in two different phase states, nematic and isotropic. We found the harmonic emission in the nematic phase to be strongly dependent on the relative orientation of the driving field polarization with respect to the liquid crystal alignment. Specifically, the harmonic yield has a maximum when the molecules are aligned perpendicularly to the polarization of the incoming radiation. Our results establish the first step for applying high-order harmonic spectroscopy as a tool for resolving ultrafast electron dynamics in liquid crystals with unprecedented temporal and spatial resolution. △ Less

Submitted 16 November, 2023; originally announced November 2023.

Comments: The paper is composed by 12 pages including references. Five images are present

Breakdown of Scaling and Friction Weakening in the Critical Granular Flow

Authors: Andrea Baldassarri, Mario A. Annunziata, Andrea Gnoli, Giorgio Pontuale, Alberto Petri

Abstract: The way granular materials response to an applied shear stress is of the utmost relevance to both human activities and natural environment. One of the their most intriguing and less understood behavior, is the stick-instability, whose most dramatic manifestation are earthquakes, ultimately governed by the dynamics of rocks and debris jammed within the fault gauge. Many of the features of earthquak… ▽ More The way granular materials response to an applied shear stress is of the utmost relevance to both human activities and natural environment. One of the their most intriguing and less understood behavior, is the stick-instability, whose most dramatic manifestation are earthquakes, ultimately governed by the dynamics of rocks and debris jammed within the fault gauge. Many of the features of earthquakes, i.e. intermittency, broad times and energy scale involved, are mimicked by a very simple experimental set-up, where small beads of glass under load are slowly sheared by an elastic medium. Analyzing data from long lasting experiments, we identify a critical dynamical regime, that can be related to known theoretical models used for "crackling-noise" phenomena. In particular, we focus on the average shape of the slip velocity, observing a "breakdown of scaling": while small slips show a self-similar shape, large does not, in a way that suggests the presence of subtle inertial effects within the granular system. In order to characterise the crossover between the two regimes, we investigate the frictional response of the system, which we trat as a stochastic quantity. Computing different averages, we evidence a weakening effect, whose Stribeck threshold velocity can be related to the aforementioned breaking of scaling. △ Less

Submitted 6 November, 2018; originally announced November 2018.

Comments: 11 pages, 14 figures

How log-normal is your country? An analysis of the statistical distribution of the exported volumes of products

Authors: Mario Alberto Annunziata, Alberto Petri, Giorgio Pontuale, Andrea Zaccaria

Abstract: We have considered the statistical distributions of the volumes of the different products exported by 148 countries. We have found that the form of these distributions is not unique but heavily depends on the level of development of the nation, as expressed by macroeconomic indicators like GDP, GDP per capita, total export and a recently introduced measure for countries' economic complexity called… ▽ More We have considered the statistical distributions of the volumes of the different products exported by 148 countries. We have found that the form of these distributions is not unique but heavily depends on the level of development of the nation, as expressed by macroeconomic indicators like GDP, GDP per capita, total export and a recently introduced measure for countries' economic complexity called fitness. We have identified three major classes: a) an incomplete log-normal shape, truncated on the left side, for the less developed countries, b) a complete log-normal, with a wider range of volumes, for nations characterized by intermediate economy, and c) a strongly asymmetric shape for countries with a high degree of development. The ranking curves of the exported volumes from each country seldom cross each other, showing a clear hierarchy of export volumes. Finally, the log-normality hypothesis has been checked for the distributions of all the 148 countries through different tests, Kolmogorov-Smirnov and Cramer-Von Mises, confirming that it cannot be rejected only for the countries of intermediate economy. △ Less

Submitted 11 June, 2015; originally announced June 2015.

Comments: 10 pages, 5 figures, submitted to IWcee15 conference

Hardening transition in a one-dimensional model for ferrogels

Authors: M. A. Annunziata, A. M. Menzel, H. Löwen

Abstract: We introduce and investigate a coarse-grained model for quasi one-dimensional ferrogels. In our description the magnetic particles are represented by hard spheres with a magnetic dipole moment in their centers. Harmonic springs connecting these spheres mimic the presence of a cross-linked polymer matrix. A special emphasis is put on the coupling of the dipolar orientations to the elastic deformati… ▽ More We introduce and investigate a coarse-grained model for quasi one-dimensional ferrogels. In our description the magnetic particles are represented by hard spheres with a magnetic dipole moment in their centers. Harmonic springs connecting these spheres mimic the presence of a cross-linked polymer matrix. A special emphasis is put on the coupling of the dipolar orientations to the elastic deformations of the matrix, where a memory effect of the orientations is included. Although the particles are displaced along one spatial direction only, the system already shows rich behavior: as a function of the magnetic dipole moment, we find a phase transition between "soft-elastic" states with finite interparticle separation and finite compressive elastic modulus on the one hand, and "hardened" states with touching particles and therefore diverging compressive elastic modulus on the other hand. Corresponding phase diagrams are derived neglecting thermal fluctuations of the magnetic particles. In addition, we consider a situation in which a spatially homogeneous magnetization is initially imprinted into the material. Depending on the strength of the magneto-mechanical coupling between the dipole orientations and the elastic deformations, the system then relaxes to a uniaxially ferromagnetic, an antiferromagnetic, or a spiral state of magnetization to minimize its energy. One purpose of our work is to provide a largely analytically solvable approach that can provide a benchmark to test future descriptions of higher complexity. From an applied point of view, our results could be exploited, for example, for the construction of novel dam** devices of tunable shock absorbance. △ Less

Submitted 2 June, 2013; originally announced June 2013.

Comments: 12 pages, 12 figures

Journal ref: J. Chem. Phys. 138, 204906 (2013)

Colloids and polymers in random colloidal matrices: demixing under good-solvent conditions

Authors: Mario Alberto Annunziata, Andrea Pelissetto

Abstract: We consider a simplified coarse-grained model for colloid-polymer mixtures, in which polymers are represented as monoatomic molecules interacting by means of pair potentials. We use it to study polymer-colloid segregation in the presence of a quenched matrix of colloidal hard spheres. We fix the polymer-to-colloid size ratio to 0.8 and consider matrices such that the fraction f of the volume that… ▽ More We consider a simplified coarse-grained model for colloid-polymer mixtures, in which polymers are represented as monoatomic molecules interacting by means of pair potentials. We use it to study polymer-colloid segregation in the presence of a quenched matrix of colloidal hard spheres. We fix the polymer-to-colloid size ratio to 0.8 and consider matrices such that the fraction f of the volume that is not accessible to the colloids due to the matrix is equal to 40%. As in the Asakura-Oosawa-Vrij (AOV) case, we find that binodal curves in the polymer and colloid volume-fraction plane have a small dependence on disorder. As for the position of the critical point, the behavior is different from that observed in the AOV case: while the critical colloid volume fraction is essentially the same in the bulk and in the presence of the matrix, the polymer volume fraction at criticality increases as f increases. At variance with the AOV case, no capillary colloid condensation or evaporation is generically observed. △ Less

Submitted 1 October, 2012; originally announced October 2012.

Comments: 27 pages, 10 figures

Fluid-fluid demixing curves for colloid-polymer mixtures in a random colloidal matrix

Authors: Mario Alberto Annunziata, Andrea Pelissetto

Abstract: We study fluid-fluid phase separation in a colloid-polymer mixture adsorbed in a colloidal porous matrix close to the θ-point. For this purpose we consider the Asakura-Oosawa model in the presence of a quenched matrix of colloidal hard spheres. We study the dependence of the demixing curve on the parameters that characterize the quenched matrix, fixing the polymer-to-colloid size ratio to 0.8. We… ▽ More We study fluid-fluid phase separation in a colloid-polymer mixture adsorbed in a colloidal porous matrix close to the θ-point. For this purpose we consider the Asakura-Oosawa model in the presence of a quenched matrix of colloidal hard spheres. We study the dependence of the demixing curve on the parameters that characterize the quenched matrix, fixing the polymer-to-colloid size ratio to 0.8. We find that, to a large extent, demixing curves depend only on a single parameter f, which represents the volume fraction which is unavailable to the colloids. We perform Monte Carlo simulations for volume fractions f equal to 40% and 70%, finding that the binodal curves in the polymer and colloid packing-fraction plane have a small dependence on disorder. The critical point instead changes significantly: for instance, the colloid packing fraction at criticality increases with increasing f. Finally, we observe for some values of the parameters capillary condensation of the colloids: a bulk colloid-poor phase is in chemical equilibrium with a colloid-rich phase in the matrix. △ Less

Submitted 7 September, 2011; originally announced September 2011.

Comments: 26 pages, 8 figures. In publication in Molecular Physics, special volume dedicated to Luciano Reatto for his 70th birthday

Reset dynamics and latching in niobium superconducting nanowire single-photon detectors

Authors: Anthony J. Annunziata, Orlando Quaranta, Daniel F. Santavicca, Alessandro Casaburi, Luigi Frunzio, Mikkel Ejrnaes, Michael J. Rooks, Roberto Cristiano, Sergio Pagano, Aviad Frydman, Daniel E. Prober

Abstract: We study the reset dynamics of niobium (Nb) superconducting nanowire single-photon detectors (SNSPDs) using experimental measurements and numerical simulations. The numerical simulations of the detection dynamics agree well with experimental measurements, using independently determined parameters in the simulations. We find that if the photon-induced hotspot cools too slowly, the device will latch… ▽ More We study the reset dynamics of niobium (Nb) superconducting nanowire single-photon detectors (SNSPDs) using experimental measurements and numerical simulations. The numerical simulations of the detection dynamics agree well with experimental measurements, using independently determined parameters in the simulations. We find that if the photon-induced hotspot cools too slowly, the device will latch into a dc resistive state. To avoid latching, the time for the hotspot to cool must be short compared to the inductive time constant that governs the resetting of the current in the device after hotspot formation. From simulations of the energy relaxation process, we find that the hotspot cooling time is determined primarily by the temperature-dependent electron-phonon inelastic time. Latching prevents reset and precludes subsequent photon detection. Fast resetting to the superconducting state is therefore essential, and we demonstrate experimentally how this is achieved. △ Less

Submitted 4 August, 2010; originally announced August 2010.

Journal ref: J. Appl. Phys. 108, 084507 (2010)

Tunable superconducting nanoinductors

Authors: Anthony J Annunziata, Daniel F Santavicca, Luigi Frunzio, Gianluigi Catelani, Michael J Rooks, Aviad Frydman, Daniel E Prober

Abstract: We characterize inductors fabricated from ultra-thin, approximately 100 nm wide strips of niobium (Nb) and niobium nitride (NbN). These nanowires have a large kinetic inductance in the superconducting state. The kinetic inductance scales linearly with the nanowire length, with a typical value of 1 nH/um for NbN and 44 pH/um for Nb at a temperature of 2.5 K. We measure the temperature and current d… ▽ More We characterize inductors fabricated from ultra-thin, approximately 100 nm wide strips of niobium (Nb) and niobium nitride (NbN). These nanowires have a large kinetic inductance in the superconducting state. The kinetic inductance scales linearly with the nanowire length, with a typical value of 1 nH/um for NbN and 44 pH/um for Nb at a temperature of 2.5 K. We measure the temperature and current dependence of the kinetic inductance and compare our results to theoretical predictions. We also simulate the self-resonant frequencies of these nanowires in a compact meander geometry. These nanowire inductive elements have applications in a variety of microwave frequency superconducting circuits. △ Less

Submitted 27 September, 2010; v1 submitted 23 July, 2010; originally announced July 2010.

Comments: 11 pages, 5 figures

Journal ref: Nanotechnology 21, 445202 (2010)

Niobium superconducting nanowire single-photon detectors

Authors: Anthony J. Annunziata, Daniel F. Santavicca, Joel D. Chudow, Luigi Frunzio, Michael J. Rooks, Aviad Frydman, Daniel E. Prober

Abstract: We investigate the performance of superconducting nanowire photon detectors fabricated from ultra-thin Nb. A direct comparison is made between these detectors and similar nanowire detectors fabricated from NbN. We find that Nb detectors are significantly more susceptible than NbN to thermal instability (latching) at high bias. We show that the devices can be stabilized by reducing the input resi… ▽ More We investigate the performance of superconducting nanowire photon detectors fabricated from ultra-thin Nb. A direct comparison is made between these detectors and similar nanowire detectors fabricated from NbN. We find that Nb detectors are significantly more susceptible than NbN to thermal instability (latching) at high bias. We show that the devices can be stabilized by reducing the input resistance of the readout. Nb detectors optimized in this way are shown to have approximately 2/3 the reset time of similar large-active-area NbN detectors of the same geometry, with approximately 6% detection efficiency for single photons at 470 nm. △ Less

Submitted 8 January, 2009; originally announced January 2009.

Journal ref: IEEE Trans. Appl. Supercond. 19(3), 327-331 (2009)