Search

This work presents an unsupervised method for automatically constructing and expanding topic taxonomies using instruction-based fine-tuned LLMs (Large Language Models). We apply topic modeling and keyword extraction techniques to create initial topic taxonomies and LLMs to post-process the resulting terms and create a hierarchy. To expand an existing taxonomy with new terms, we use zero-shot prompting to find out where to add new nodes, which, to our knowledge, is the first work to present such an approach to taxonomy tasks. We use the resulting taxonomies to assign tags that characterize merchants from a retail bank dataset. To evaluate our work, we asked 12 volunteers to answer a two-part form in which we first assessed the quality of the taxonomies created and then the tags assigned to merchants based on that taxonomy. The evaluation revealed a coherence rate exceeding 90% for the chosen taxonomies. The taxonomies' expansion with LLMs also showed exciting results for parent node prediction, with an f1-score above 70% in our taxonomies.

This work proposes the Two-headed DragoNet, a Transformer-based model for hierarchical multi-label classification of financial transactions. Our model is based on a stack of Transformers encoder layers that generate contextual embeddings from two short textual descriptors (merchant name and business activity), followed by a Context Fusion layer and two output heads that classify transactions according to a hierarchical two-level taxonomy (macro and micro categories). Finally, our proposed Taxonomy-aware Attention Layer corrects predictions that break categorical hierarchy rules defined in the given taxonomy. Our proposal outperforms classical machine learning methods in experiments of macro-category classification by achieving an F1-score of 93\% on a card dataset and 95% on a current account dataset.

Saturn is an innovative platform that assists Foundation Model (FM) building and its integration with IT operations (Ops). It is custom-made to meet the requirements of data scientists, enabling them to effectively create and implement FMs while enhancing collaboration within their technical domain. By offering a wide range of tools and features, Saturn streamlines and automates different stages of FM development, making it an invaluable asset for data science teams. This white paper introduces prospective applications of generative AI models derived from FMs in the financial sector.

SPIDER is the full-scale prototype of the ion source of the ITER Heating Neutral Beam Injector, where negative ions of Hydrogen or Deuterium are produced by a RF generated plasma and accelerated with a set of grids up to ~100 keV. The Power Supply System is composed of high voltage dc power supplies capable of handling frequent grid breakdowns, high current dc generators for the magnetic filter field and RF generators for the plasma generation. During the first 3 years of SPIDER operation different electrical issues were discovered, understood and addressed thanks to deep analyses of the experimental results supported by modelling activities. The paper gives an overview on the observed phenomena and relevant analyses to understand them, on the effectiveness of the short-term modifications provided to SPIDER to face the encountered issues and on the design principle of long-term solutions to be introduced during the currently ongoing long shutdown., Comment: 8 pages, 12 figures. Presented at SOFT 2022

ITER envisages the use of two heating neutral beam injectors plus an optional one as part of the auxiliary heating and current drive system. The 16.5 MW expected neutral beam power per injector is several notches higher than worldwide existing facilities. A Neutral Beam Test Facility (NBTF) was established at Consorzio RFX, exploiting the synergy of two test beds, SPIDER and MITICA. SPIDER is dedicated to developing and characterizing large efficient negative ion sources at relevant parameters in ITER-like conditions: source and accelerator located in the same vacuum where the beam propagates, immunity to electromagnetic interferences of multiple radio-frequency (RF) antennas, avoidance of RF-induced discharges on the outside of the source. Three years of experiments on SPIDER have addressed to the necessary design modifications to enable full performances. The source is presently under a long shut-down phase to incorporate learnings from the experimental campaign. Parallelly, developments on MITICA, the full-scale prototype of the ITER NBI featuring a 1 MV accelerator and ion neutralization, are underway including manufacturing of in-vessel components, while power supplies and auxiliary plants are already under final testing and commissioning. Integration, commissioning and tests of the 1MV power supplies are essential for this first-of-kind system, unparalleled both in research and industry field. The integrated test to confirm 1MV output by combining invertor systems, DC generators and transmission lines extracted errors/accidents in some components. To realize a concrete system for ITER, solutions for the repair and the improvement of the system were developed. Hence, NBTF is emerging as a necessary facility, due to the large gap with existing injectors, effectively dedicated to identify issues and find solutions to enable successful ITER NBI operations in a time bound fashion.

Category theory plays a special character in mathematics - it unifies distinct branches under the same formalism. Despite this integrative power in math, it also seems to provide the proper foundations to the experimental physicist. In this work, we present another application of category in physics, related to the principle of relativity. The operational construction of (inertial) frames of reference indicates that only the movement between one and another frame is enough to differentiate both of them. This fact is hidden when one applies only group theory to connect frames. In fact, rotations and translations only change coordinates, keeping the frame inert. The change of frames is only attainable by boosts in the classical and relativistic regimes for both Galileo and Lorentz (Poincar\'e) groups. Besides providing a non-trivial example of application of category theory in physics, we also fulfill the presented gap when one directly applies group theory for connecting frames., Comment: 13 pages, adjusted version according to reviewer's suggestions - to appear in Reports on Mathematical Physics

Recently, random lasing in complex networks has shown efficient lasing over more than 50 localised modes, promoted by multiple scattering over the underlying graph. If controlled, these network lasers can lead to fast-switching multifunctional light sources with synthesised spectrum. Here, we observe both in experiment and theory high sensitivity of the network laser to the spatial shape of the pump profile, with mode intensity variation of up to 280% for a non-homogeneous 7% pump decrease. We solve the nonlinear equations within the steady state ab-initio laser theory (SALT) approximation over a graph and we show selective lasing of around 90% of the top modes, effectively programming the spectrum of the lasing networks. In our experiments with polymer networks, this high sensitivity enables control of the lasing spectrum through non-uniform pump patterns. We propose the underlying complexity of the network modes as the key element behind efficient spectral control opening the way for the development of optical devices with wide impact for on-chip photonics for communication, sensing and computation., Comment: 22 pages, 14 figures

This paper is devoted to discussing the topological structure of the arrow of time. In the literature, it is often accepted that its algebraic and topological structures are that of a one-dimensional Euclidean space $\mathbb{E}^1$, although a critical review on the subject is not easy to be found. Hence, leveraging on an operational approach, we collect evidences to identify it structurally as a normed vector space $(\mathbb{Q}, |\cdot|)$, and take a leap of abstraction to complete it, up to isometries, to the real line. During the development of the paper, the space-time is recognized as a fibration, with the fibers being the sets of simultaneous events. The corresponding topology is also exposed: open sets naturally arise within our construction, showing that the classical space-time is non-Hausdorff. The transition from relativistic to classical regimes is explored too., Comment: 26 pages, 9 figures. Misprints corrected and minor adjustments according to the review process

Lasers based on biological materials are attracting an increasing interest in view of their use in integrated and transient photonics. DNA as optical biopolymer in combination with highly-emissive dyes has been reported to have excellent potential in this respect, however achieving miniaturized lasing systems based on solid-state DNA shaped in different geometries to confine and enhance emission is still a challenge, and physico-chemical mechanisms originating fluorescence enhancement are not fully understood. Herein, a class of wavelength-tunable lasers based on DNA nanofibers is demonstrated, for which optical properties are highly controlled through the system morphology. A synergistic effect is highlighted at the basis of lasing action. Through a quantum chemical investigation, we show that the interaction of DNA with the encapsulated dye leads to hindered twisting and suppressed channels for the non-radiative decay. This is combined with effective waveguiding, optical gain, and tailored mode confinement to promote morphologically-controlled lasing in DNA-based nanofibers. The results establish design rules for the development of bright and tunable nanolasers and optical networks based on DNA nanostructures., Comment: 31 pages, 11 figures and 2 tables

We realise an antibacterial nanomaterial based on the self-limited assembly of patchy plasmonic colloids, obtained by adsorption of lysozyme to gold nanoparticles. The possibility of selecting the size of the assemblies within several hundred nanometres allows for tuning their optical response in a wide range of frequencies from visible to near infrared. We also demonstrate an aggregation-dependent modulation of the catalytic activity, which results in an enhancement of the antibacterial performances for assemblies of the proper size. The gained overall control on structure, optical properties and biological activity of such nanomaterial paves the way for the development of novel antibacterial nanozymes with promising applications in treating multi drug resistant bacteria.

Quantum mechanically, photoionization can be fully described by the complex photoionization amplitudes that describe the transition between the ground state and the continuum state. Knowledge of the value of the phase of these amplitudes has been a central interest in photoionization studies and newly developing attosecond science, since the phase can reveal important information about phenomena such as electron correlation. We present a new attosecond-precision interferometric method of angle-resolved measurement for the phase of the photoionization amplitudes, using two phase-locked Extreme Ultraviolet pulses of frequency $\omega$ and $2\omega$, from a Free-Electron Laser. Phase differences $\Delta \tilde \eta$ between one- and two-photon ionization channels, averaged over multiple wave packets, are extracted for neon $2p$ electrons as a function of emission angle at photoelectron energies 7.9, 10.2, and 16.6 eV. $\Delta \tilde \eta$ is nearly constant for emission parallel to the electric vector but increases at 10.2 eV for emission perpendicular to the electric vector. We model our observations with both perturbation and \textit{ab initio} theory, and find excellent agreement. In the existing method for attosecond measurement, Reconstruction of Attosecond Beating By Interference of Two-photon Transitions (RABBITT), a phase difference between two-photon pathways involving absorption and emission of an infrared photon is extracted. Our method can be used for extraction of a phase difference between single-photon and two-photon pathways and provides a new tool for attosecond science, which is complementary to RABBITT.

Abstract axiomatic formulation of mathematical structures are extensively used to describe our physical world. We take here the reverse way. By making basic assumptions as starting point, we reconstruct some features of both geometry and topology in a fully operational manner. Curiously enough, primitive concepts such as points, spaces, straight lines, planes are all defined within our formalism. Our construction breaks down with the usual literature, as our axioms have deep connection with nature. Besides that, we hope this operational approach could also be of pedagogical interest., Comment: 12 pages; 15 figures

An understanding of collective effects is of fundamental importance for the design and optimisation of the performance of modern accelerators. In particular, the design of an accelerator with strict requirements on the beam quality, such as a free electron laser (FEL), is highly dependent on a correspondence between simulation, theory and experiments in order to correctly account for the effect of coherent synchrotron radiation (CSR), and other collective effects. A traditional approach in accelerator simulation codes is to utilise an analytic one-dimensional approximation to the CSR force. We present an extension of the 1D CSR theory in order to correctly account for the CSR force at the entrance and exit of a bending magnet. A limited range of applicability to this solution, in particular in bunches with a large transverse spot size or offset from the nominal axis, is recognised. More recently developed codes calculate the CSR effect in dispersive regions directly from the Lienard-Wiechert potentials, albeit with approximations to improve the computational time. A new module of the General Particle Tracer (GPT) code was developed for simulating the effects of CSR, and benchmarked against other codes. We experimentally demonstrate departure from the commonly used 1D CSR theory for more extreme bunch length compression scenarios at the FERMI FEL facility. Better agreement is found between experimental data and the codes which account for the transverse extent of the bunch, particularly in more extreme compression scenarios., Comment: 33 pages, 19 figures, submitted to New J. Phys

Quantum emitters radiate light omni-directionally, making it hard to collect and use the generated photons. Here we propose a 3D metal-dielectric parabolic antenna surrounding an individual quantum dot as a source of collimated single photons which can then be easily extracted and manipulated. Our fabrication method relies on a single optically-induced polymerization step, once the selected emitter has been localized by confocal microscopy. Compared to conventional nano-antennas, our geometry does not require near-field coupling and it is therefore very robust against misalignment issues, and minimally affected by absorption in the metal. The parabolic antenna provides one of the largest reported experimental directivities (D=106) and the lowest beam divergences ({\Theta}=13.5 deg), a broadband operation over all the visible and near-IR range, together with more than 96% extraction efficiency, offering a practical advantage for quantum technological applications., Comment: accepted for publication in Nano Letters

The Gamma Beam System (GBS) is a high brightness LINAC to be installed in Magurele (Bucharest) at the new ELI-NP (Extreme Light Infrastructure - Nuclear Physics) laboratory. The accelerated electrons, with energies ranging from 280 to 720 MeV, will collide with a high power laser to produce tunable high energy photons (0.2-20MeV ) with high intensity (10e13 photons/s), high brilliance and spectral purity (0.1% BW), through the Compton backscattering process. This light source will be open to users for nuclear photonics and nuclear physics advanced experiments. Tested high level applications will play an important role in commissioning and operation. In this paper we report the progress and status of the development of dedicated high level applications. We also present the results of the test on the FERMI LINAC of the electron trajectory control method based on Dispersion Free Steering.

Conventional nano-photonic schemes minimise multiple scattering to realise a miniaturised version of beam-splitters, interferometers and optical cavities for light propagation and lasing. Here instead, we introduce a nanophotonic network built from multiple paths and interference, to control and enhance light-matter interaction via light localisation. The network is built from a mesh of subwavelength waveguides, and can sustain localised modes and mirror-less light trapping stemming from interference over hundreds of nodes. With optical gain, these modes can easily lase, reaching $\sim$100 pm linewidths. We introduce a graph solution to the Maxwell's equation which describes light on the network, and predicts lasing action. In this framework, the network optical modes can be designed via the network connectivity and topology, and lasing can be tailored and enhanced by the network shape. Nanophotonic networks pave the way for new laser device architectures, which can be used for sensitive biosensing and on-chip optical information processing.