Search

We propose a sampling algorithm relying on a collective variable (CV) of mid-size dimension modelled by a normalizing flow and using non-equilibrium dynamics to propose full configurational moves from the proposition of a refreshed value of the CV made by the flow. The algorithm takes the form of a Markov chain with non-local updates, allowing jumps through energy barriers across metastable states. The flow is trained throughout the algorithm to reproduce the free energy landscape of the CV. The output of the algorithm are a sample of thermalized configurations and the trained network that can be used to efficiently produce more configurations. We show the functioning of the algorithm first on a test case with a mixture of Gaussians. Then we successfully test it on a higher dimensional system consisting in a polymer in solution with a compact and an extended stable state separated by a high free energy barrier.

Bound states in the continuum (BICs) garnered significant for their potential to create new types of nanophotonic devices. Most prior demonstrations were based on arrays of dielectric resonators, which cannot be miniaturized beyond the diffraction limit, reducing the applicability of BICs for advanced functions. Here, we demonstrate BICs and quasi-BICs based on high-quality factor phonon-polariton resonances in isotopically pure h11BN and how these states can be supported by periodic arrays of nanoresonators with sizes much smaller than the wavelength. We theoretically illustrate how BICs emerge from the band structure of the arrays and verify both numerically and experimentally the presence of these states and enhanced quality factor. Furthermore, we identify and characterize simultaneously quasi-BICs and bright states. Our method can be generalized to create a large number of optical states and to tune their coupling with the environment, paving the way to miniaturized nanophotonic devices with more advanced functions., Comment: 13 pages, 5 figures, early version

Accurate and rapid situation analysis during humanitarian crises is critical to delivering humanitarian aid efficiently and is fundamental to humanitarian imperatives and the Leave No One Behind (LNOB) principle. This data analysis can highly benefit from language processing systems, e.g., by classifying the text data according to a humanitarian ontology. However, approaching this by simply fine-tuning a generic large language model (LLM) involves considerable practical and ethical issues, particularly the lack of effectiveness on data-sparse and complex subdomains, and the encoding of societal biases and unwanted associations. In this work, we aim to provide an effective and ethically-aware system for humanitarian data analysis. We approach this by (1) introducing a novel architecture adjusted to the humanitarian analysis framework, (2) creating and releasing a novel humanitarian-specific LLM called HumBert, and (3) proposing a systematic way to measure and mitigate biases. Our experiments' results show the better performance of our approach on zero-shot and full-training settings in comparison with strong baseline models, while also revealing the existence of biases in the resulting LLMs. Utilizing a targeted counterfactual data augmentation approach, we significantly reduce these biases without compromising performance., Comment: Accepted at IJCAI 2023

Timely and effective response to humanitarian crises requires quick and accurate analysis of large amounts of text data - a process that can highly benefit from expert-assisted NLP systems trained on validated and annotated data in the humanitarian response domain. To enable creation of such NLP systems, we introduce and release HumSet, a novel and rich multilingual dataset of humanitarian response documents annotated by experts in the humanitarian response community. The dataset provides documents in three languages (English, French, Spanish) and covers a variety of humanitarian crises from 2018 to 2021 across the globe. For each document, HUMSET provides selected snippets (entries) as well as assigned classes to each entry annotated using common humanitarian information analysis frameworks. HUMSET also provides novel and challenging entry extraction and multi-label entry classification tasks. In this paper, we take a first step towards approaching these tasks and conduct a set of experiments on Pre-trained Language Models (PLM) to establish strong baselines for future research in this domain. The dataset is available at https://blog.thedeep.io/humset/., Comment: Published at Findings of EMNLP 2022

Optical singularities play a major role in modern optics and are frequently deployed in structured light, super-resolution microscopy, and holography. While phase singularities are uniquely defined as locations of undefined phase, polarization singularities studied thus far are either partial, i.e., bright points of well-defined polarization, or unstable for small field perturbations. We demonstrate for the first time a complete, topologically protected polarization singularity; it is located in the 4D space spanned by the three spatial dimensions and the wavelength and is created in the focus of a cascaded metasurface-lens system. The field Jacobian plays a key role in the design of such higher-dimensional singularities, which can be extended to multidimensional wave phenomena, and pave the way to novel applications in topological photonics and precision sensing.

The growth of domain-specific applications of semantic models, boosted by the recent achievements of unsupervised embedding learning algorithms, demands domain-specific evaluation datasets. In many cases, content-based recommenders being a prime example, these models are required to rank words or texts according to their semantic relatedness to a given concept, with particular focus on top ranks. In this work, we give a threefold contribution to address these requirements: (i) we define a protocol for the construction, based on adaptive pairwise comparisons, of a relatedness-based evaluation dataset tailored on the available resources and optimized to be particularly accurate in top-rank evaluation; (ii) we define appropriate metrics, extensions of well-known ranking correlation coefficients, to evaluate a semantic model via the aforementioned dataset by taking into account the greater significance of top ranks. Finally, (iii) we define a stochastic transitivity model to simulate semantic-driven pairwise comparisons, which confirms the effectiveness of the proposed dataset construction protocol., Comment: This is a pre-print of an article published in the proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)

Semiconductor ring lasers are miniaturized devices that operate on clockwise and counterclockwise modes. These modes are not coupled in the absence of intracavity reflectors, which prevents the formation of a standing wave in the cavity and, consequently, of a population inversion grating. This should inhibit the onset of multimode emission driven by spatial hole burning. Here we show that, despite this notion, ring quantum cascade lasers inherently operate in phase-locked multimode states, that switch on and off as the pumping level is progressively increased. By rewriting the master equation of lasers with fast gain media in the form of the complex Ginzburg-Landau equation, we show that ring frequency combs stem from a phase instability---a phenomenon also known in superconductors and Bose-Einstein condensates. The instability is due to coupling of the amplitude and phase modulation of the optical field in a semiconductor laser, which plays the role of a Kerr nonlinearity, highlighting a connection between laser and microresonator frequency combs.

We present high quality factor optical nanoresonators operating in the mid-IR to far-IR based on phonon polaritons in van der Waals materials. The nanoresonators are disks patterned from isotopically pure hexagonal boron nitride (isotopes 10B and 11B) and {\alpha}-molybdenum trioxide. We experimentally achieved quality factors of nearly 400, the highest ever observed in nano-resonators at these wavelengths. The excited modes are deeply subwavelength, and the resonators are 10 to 30 times smaller than the exciting wavelength. These results are very promising for the realization of nano-photonics devices such as optical bio-sensors and miniature optical components such as polarizers and filters., Comment: 8 pages, 3 figures 3 tables

Polaritons formed by the coupling of light and material excitations such as plasmons, phonons, or excitons enable light-matter interactions at the nanoscale beyond what is currently possible with conventional optics. Recently, significant interest has been attracted by polaritons in van der Waals materials, which could lead to applications in sensing, integrated photonic circuits and detectors. However, novel techniques are required to control the propagation of polaritons at the nanoscale and to implement the first practical devices. Here we report the experimental realization of polariton refractive and meta-optics in the mid-infrared by exploiting the properties of low-loss phonon polaritons in isotopically pure hexagonal boron nitride (hBN), which allow it to interact with the surrounding dielectric environment comprising the low-loss phase change material, Ge$_3$Sb$_2$Te$_6$ (GST). We demonstrate waveguides which confine polaritons in a 1D geometry, and refractive optical elements such as lenses and prisms for phonon polaritons in hBN, which we characterize using scanning near field optical microscopy. Furthermore, we demonstrate metalenses, which allow for polariton wavefront engineering and sub-wavelength focusing. Our method, due to its sub-diffraction and planar nature, will enable the realization of programmable miniaturized integrated optoelectronic devices, and will lay the foundation for on-demand biosensors., Comment: 15 pages, 4 figures, typos corrected in v2

Since the days of Hertz, radio transmitters have evolved from rudimentary circuits emitting around 50 MHz to modern ubiquitous Wi-Fi devices operating at gigahertz radio bands. As wireless data traffic continues to increase there is a need for new communication technologies capable of high-frequency operation for high-speed data transfer. Here we give a proof of concept of a new compact radio frequency transmitter based on a semiconductor laser frequency comb. In this laser, the beating among the coherent modes oscillating inside the cavity generates a radio frequency current, which couples to the electrodes of the device. We show that redesigning the top contact of the laser allows one to exploit the internal oscillatory current to drive an integrated dipole antenna, which radiates into free space. In addition, direct modulation of the laser current permits encoding a signal in the radiated radio frequency carrier. Working in the opposite direction, the antenna can receive an external radio frequency signal, couple it to the active region and injection lock the laser. These results pave the way to new applications and functionality in optical frequency combs, such as wireless radio communication and wireless synchronization to a reference source.

Hyperbolic media have attracted much attention in the photonics community, thanks to their ability to confine light to arbitrarily small volumes and to their use for super-resolution applications. The 2D counterpart of these media can be achieved with hyperbolic metasurfaces, which support in-plane hyperbolic guided modes thanks to nanopatterns which, however, pose significant fabrication challenges and limit the achievable confinement. We show that thin flakes of the van der Waals material {\alpha}-MoO3 can support naturally in-plane hyperbolic polariton guided modes at mid-infrared frequencies without any patterning. This is possible because {\alpha}-MoO3 is a biaxial hyperbolic crystal, with three different Restrahlen bands, each for a different crystal axis. Our findings can pave the way towards new paradigm to manipulate and confine light in planar photonic devices.

Van der Waals heterostructures assembled from layers of 2D materials have attracted considerable interest due to their novel optical and electrical properties. Here we report a scattering-type scanning near field optical microscopy study of hexagonal boron nitride on black phosphorous (h-BN/BP) heterostructures, demonstrating the first direct observation of in-plane anisotropic phonon polariton modes in vdW heterostructures. Strikingly, the measured in-plane optical anisotropy along armchair and zigzag crystal axes exceeds the ratio of refractive indices of BP in the x-y plane. We explain that this enhancement is due to the high confinement of the phonon polaritons in h-BN. We observe a maximum in-plane optical anisotropy of {\alpha}_max=1.25 in the 1405-1440 cm-1 frequency spectrum. These results provide new insights on the behavior of polaritons in vdW heterostructures, and the observed anisotropy enhancement paves the way to novel nanophotonic devices and to a new way to characterize optical anisotropy in thin films.

We report a THz reflectarray metasurface which uses graphene as active element to achieve beam steering, shaping and broadband phase modulation. This is based on the creation of a voltage controlled reconfigurable phase hologram, which can impart different reflection angles and phases to an incident beam, replacing bulky and fragile rotating mirrors used for terahertz imaging. This can also find applications in other regions of the electromagnetic spectrum, paving the way to versatile optical devices including light radars, adaptive optics, electro-optical modulators and screens., Comment: 16 pages, 10 figures