Showing 1–30 of 30 results for author: Toma, A

Enhanced Terahertz Spectroscopy of a Monolayer Transition Metal Dichalcogenide

Authors: Xin **, Vincenzo Aglieri, Young-Gyun Jeong, Atiye Pezeshki, Lilian Skokan, Mostafa Shagar, Yuechen Jia, Pablo Bianucci, Andreas Ruediger, Emanuele Orgiu, Andrea Toma, Luca Razzari

Abstract: Two-dimensional materials, including transition metal dichalcogenides, are attractive for a variety of applications in electronics as well as photonics and have recently been envisioned as an appealing platform for phonon polaritonics. However, their direct characterization in the terahertz spectral region, of interest for retrieving, e.g., their phonon response, represents a major challenge, due… ▽ More Two-dimensional materials, including transition metal dichalcogenides, are attractive for a variety of applications in electronics as well as photonics and have recently been envisioned as an appealing platform for phonon polaritonics. However, their direct characterization in the terahertz spectral region, of interest for retrieving, e.g., their phonon response, represents a major challenge, due to the limited sensitivity of typical terahertz spectroscopic tools and the weak interaction of such long-wavelength radiation with sub-nanometer systems. In this work, by exploiting an ad-hoc engineered metallic surface enabling a ten-thousand-fold local absorption boost, we perform enhanced terahertz spectroscopy of a monolayer transition metal dichalcogenide (tungsten diselenide) and extract its dipole-active phonon resonance features. In addition, we use these data to obtain the monolayer effective permittivity around its phonon resonance. Via the direct terahertz characterization of the phonon response of such two-dimensional systems, this method opens the path to the rational design of phonon polariton devices exploiting monolayer transition metal dichalcogenides. △ Less

Submitted 14 May, 2024; originally announced May 2024.

WangLab at MEDIQA-M3G 2024: Multimodal Medical Answer Generation using Large Language Models

Authors: Ronald Xie, Steven Palayew, Augustin Toma, Gary Bader, Bo Wang

Abstract: This paper outlines our submission to the MEDIQA2024 Multilingual and Multimodal Medical Answer Generation (M3G) shared task. We report results for two standalone solutions under the English category of the task, the first involving two consecutive API calls to the Claude 3 Opus API and the second involving training an image-disease label joint embedding in the style of CLIP for image classificati… ▽ More This paper outlines our submission to the MEDIQA2024 Multilingual and Multimodal Medical Answer Generation (M3G) shared task. We report results for two standalone solutions under the English category of the task, the first involving two consecutive API calls to the Claude 3 Opus API and the second involving training an image-disease label joint embedding in the style of CLIP for image classification. These two solutions scored 1st and 2nd place respectively on the competition leaderboard, substantially outperforming the next best solution. Additionally, we discuss insights gained from post-competition experiments. While the performance of these two solutions have significant room for improvement due to the difficulty of the shared task and the challenging nature of medical visual question answering in general, we identify the multi-stage LLM approach and the CLIP image classification approach as promising avenues for further investigation. △ Less

Submitted 22 April, 2024; originally announced April 2024.

WangLab at MEDIQA-CORR 2024: Optimized LLM-based Programs for Medical Error Detection and Correction

Authors: Augustin Toma, Ronald Xie, Steven Palayew, Patrick R. Lawler, Bo Wang

Abstract: Medical errors in clinical text pose significant risks to patient safety. The MEDIQA-CORR 2024 shared task focuses on detecting and correcting these errors across three subtasks: identifying the presence of an error, extracting the erroneous sentence, and generating a corrected sentence. In this paper, we present our approach that achieved top performance in all three subtasks. For the MS dataset,… ▽ More Medical errors in clinical text pose significant risks to patient safety. The MEDIQA-CORR 2024 shared task focuses on detecting and correcting these errors across three subtasks: identifying the presence of an error, extracting the erroneous sentence, and generating a corrected sentence. In this paper, we present our approach that achieved top performance in all three subtasks. For the MS dataset, which contains subtle errors, we developed a retrieval-based system leveraging external medical question-answering datasets. For the UW dataset, reflecting more realistic clinical notes, we created a pipeline of modules to detect, localize, and correct errors. Both approaches utilized the DSPy framework for optimizing prompts and few-shot examples in large language model (LLM) based programs. Our results demonstrate the effectiveness of LLM based programs for medical error correction. However, our approach has limitations in addressing the full diversity of potential errors in medical documentation. We discuss the implications of our work and highlight future research directions to advance the robustness and applicability of medical error detection and correction systems. △ Less

Submitted 22 April, 2024; originally announced April 2024.

GPT-4V(ision) Unsuitable for Clinical Care and Education: A Clinician-Evaluated Assessment

Authors: Senthujan Senkaiahliyan, Augustin Toma, Jun Ma, An-Wen Chan, Andrew Ha, Kevin R. An, Hrishikesh Suresh, Barry Rubin, Bo Wang

Abstract: OpenAI's large multimodal model, GPT-4V(ision), was recently developed for general image interpretation. However, less is known about its capabilities with medical image interpretation and diagnosis. Board-certified physicians and senior residents assessed GPT-4V's proficiency across a range of medical conditions using imaging modalities such as CT scans, MRIs, ECGs, and clinical photographs. Alth… ▽ More OpenAI's large multimodal model, GPT-4V(ision), was recently developed for general image interpretation. However, less is known about its capabilities with medical image interpretation and diagnosis. Board-certified physicians and senior residents assessed GPT-4V's proficiency across a range of medical conditions using imaging modalities such as CT scans, MRIs, ECGs, and clinical photographs. Although GPT-4V is able to identify and explain medical images, its diagnostic accuracy and clinical decision-making abilities are poor, posing risks to patient safety. Despite the potential that large language models may have in enhancing medical education and delivery, the current limitations of GPT-4V in interpreting medical images reinforces the importance of appropriate caution when using it for clinical decision-making. △ Less

Submitted 14 November, 2023; originally announced March 2024.

Bound states in the continuum and long-range coupling of polaritons in hexagonal boron nitride nanoresonators

Authors: Harsh Gupta, Giacomo Venturi, Tatiana Contino, Eli Janzen, James H. Edgar, Francesco de Angelis, Andrea Toma, Antonio Ambrosio, Michele Tamagnone

Abstract: 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… ▽ More 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. △ Less

Submitted 21 September, 2023; originally announced September 2023.

Comments: 13 pages, 5 figures, early version

Plasmon-enhanced circular dichroism spectroscopy of chiral drug solutions

Authors: Matteo Venturi, Raju Adhikary, Ambaresh Sahoo, Carino Ferrante, Isabella Daidone, Francesco Di Stasio, Andrea Toma, Francesco Tani, Hatice Altug, Antonio Mecozzi, Massimiliano Aschi, Andrea Marini

Abstract: We investigate the potential of surface plasmon polaritons at noble metal interfaces for surface-enhanced chiroptical sensing of dilute chiral drug solutions. The high quality factor of surface plasmon resonances in both Otto and Kretschmann configurations enables the enhancement of circular dichroism differenatial absorption thanks to the large near-field intensity of such plasmonic excitations.… ▽ More We investigate the potential of surface plasmon polaritons at noble metal interfaces for surface-enhanced chiroptical sensing of dilute chiral drug solutions. The high quality factor of surface plasmon resonances in both Otto and Kretschmann configurations enables the enhancement of circular dichroism differenatial absorption thanks to the large near-field intensity of such plasmonic excitations. Furthermore, the subwavelength confinement of surface plasmon polaritons is key to attain chiroptical sensitivity to small amounts of drug volumes placed around $\simeq 100$ nm by the metal surface. Our calculations focus on reparixin, a pharmaceutical molecule currently used in clinical studies for patients with community-acquired pneumonia, including COVID-19 and acute respiratory distress syndrome. Considering realistic dilute solutions of reparixin dissolved in water with concentration $\leq 5$ mg$/$ml, we find a circular-dichroism differential absorption enhancement factor of the order $\simeq 20$ and chirality-induced polarization distortion upon surface plasmon polariton excitation. △ Less

Submitted 27 September, 2023; v1 submitted 16 August, 2023; originally announced August 2023.

Clinical Camel: An Open Expert-Level Medical Language Model with Dialogue-Based Knowledge Encoding

Authors: Augustin Toma, Patrick R. Lawler, Jimmy Ba, Rahul G. Krishnan, Barry B. Rubin, Bo Wang

Abstract: We present Clinical Camel, an open large language model (LLM) explicitly tailored for clinical research. Fine-tuned from LLaMA-2 using QLoRA, Clinical Camel achieves state-of-the-art performance across medical benchmarks among openly available medical LLMs. Leveraging efficient single-GPU training, Clinical Camel surpasses GPT-3.5 in five-shot evaluations on all assessed benchmarks, including 64.3… ▽ More We present Clinical Camel, an open large language model (LLM) explicitly tailored for clinical research. Fine-tuned from LLaMA-2 using QLoRA, Clinical Camel achieves state-of-the-art performance across medical benchmarks among openly available medical LLMs. Leveraging efficient single-GPU training, Clinical Camel surpasses GPT-3.5 in five-shot evaluations on all assessed benchmarks, including 64.3% on the USMLE Sample Exam (compared to 58.5% for GPT-3.5), 77.9% on PubMedQA (compared to 60.2%), 60.7% on MedQA (compared to 53.6%), and 54.2% on MedMCQA (compared to 51.0%). In addition to these benchmarks, Clinical Camel demonstrates its broader capabilities, such as synthesizing plausible clinical notes. This work introduces dialogue-based knowledge encoding, a novel method to synthesize conversational data from dense medical texts. While benchmark results are encouraging, extensive and rigorous human evaluation across diverse clinical scenarios is imperative to ascertain safety before implementation. By openly sharing Clinical Camel, we hope to foster transparent and collaborative research, working towards the safe integration of LLMs within the healthcare domain. Significant challenges concerning reliability, bias, and the potential for outdated knowledge persist. Nonetheless, the transparency provided by an open approach reinforces the scientific rigor essential for future clinical applications. △ Less

Submitted 17 August, 2023; v1 submitted 19 May, 2023; originally announced May 2023.

Comments: for model weights, see https://huggingface.co/wanglab/

WangLab at MEDIQA-Chat 2023: Clinical Note Generation from Doctor-Patient Conversations using Large Language Models

Authors: John Giorgi, Augustin Toma, Ronald Xie, Sondra S. Chen, Kevin R. An, Grace X. Zheng, Bo Wang

Abstract: This paper describes our submission to the MEDIQA-Chat 2023 shared task for automatic clinical note generation from doctor-patient conversations. We report results for two approaches: the first fine-tunes a pre-trained language model (PLM) on the shared task data, and the second uses few-shot in-context learning (ICL) with a large language model (LLM). Both achieve high performance as measured by… ▽ More This paper describes our submission to the MEDIQA-Chat 2023 shared task for automatic clinical note generation from doctor-patient conversations. We report results for two approaches: the first fine-tunes a pre-trained language model (PLM) on the shared task data, and the second uses few-shot in-context learning (ICL) with a large language model (LLM). Both achieve high performance as measured by automatic metrics (e.g. ROUGE, BERTScore) and ranked second and first, respectively, of all submissions to the shared task. Expert human scrutiny indicates that notes generated via the ICL-based approach with GPT-4 are preferred about as often as human-written notes, making it a promising path toward automated note generation from doctor-patient conversations. △ Less

Submitted 3 June, 2023; v1 submitted 3 May, 2023; originally announced May 2023.

Comments: Camera-ready submission to ClinicalNLP @ ACL 2023

A Cloud-based Deep Learning Framework for Early Detection of Pushing at Crowded Event Entrances

Authors: Ahmed Alia, Mohammed Maree, Mohcine Chraibi, Anas Toma, Armin Seyfried

Abstract: Crowding at the entrances of large events may lead to critical and life-threatening situations, particularly when people start pushing each other to reach the event faster. Automatic and timely identification of pushing behavior would help organizers and security forces to intervene early and mitigate dangerous situations. In this paper, we propose a cloud-based deep learning framework for automat… ▽ More Crowding at the entrances of large events may lead to critical and life-threatening situations, particularly when people start pushing each other to reach the event faster. Automatic and timely identification of pushing behavior would help organizers and security forces to intervene early and mitigate dangerous situations. In this paper, we propose a cloud-based deep learning framework for automatic early detection of pushing in crowded event entrances. The proposed framework initially modifies and trains the EfficientNetV2B0 Convolutional Neural Network model. Subsequently, it integrates the adapted model with an accurate and fast pre-trained deep optical flow model with the color wheel method to analyze video streams and identify pushing patches in real-time. Moreover, the framework uses live capturing technology and a cloud-based environment to collect video streams of crowds in real-time and provide early-stage results. A novel dataset is generated based on five real-world experiments and their associated ground truth data to train the adapted EfficientNetV2B0 model. The experimental setups simulated a crowded event entrance, while the ground truths for each video experiment was generated manually by social psychologists. Several experiments on the videos and the generated dataset are carried out to evaluate the accuracy and annotation delay time of the proposed framework. The experimental results show that the proposed framework identified pushing behaviors with an accuracy rate of 87% within a reasonable delay time. △ Less

Submitted 9 June, 2023; v1 submitted 16 February, 2023; originally announced February 2023.

Journal ref: 2023, IEEE Access

Meta-Optics with Lithium Niobate

Authors: Anna Fedotova, Luca Carletti, Attilio Zilli, Frank Setzpfandt, Isabelle Staude, Andrea Toma, Marco Finazzi, Costantino De Angelis, Thomas Pertsch, Dragomir N. Neshev, Michele Celebrano

Abstract: The rapid development of metasurfaces - 2D ensembles of engineered nanostructures - is presently fostering a steady drive towards the miniaturization of many optical functionalities and devices to a subwavelength size. The material platforms for optical metasurfaces are rapidly expanding and for the past few years, we are seeing a surge in establishing meta-optical elements from high-index, highly… ▽ More The rapid development of metasurfaces - 2D ensembles of engineered nanostructures - is presently fostering a steady drive towards the miniaturization of many optical functionalities and devices to a subwavelength size. The material platforms for optical metasurfaces are rapidly expanding and for the past few years, we are seeing a surge in establishing meta-optical elements from high-index, highly transparent materials with strong nonlinear and electro-optic properties. Crystalline lithium niobate (LN), a prime material of choice in integrated photonics, has shown great promise for future meta-optical components, thanks to its large electro-optical coefficient, second-order nonlinear response and broad transparency window ranging from the visible to the mid-infrared. Recent advances in nanofabrication technology have indeed marked a new milestone in the miniaturization of LN platforms, hence enabling the first demonstrations of LN-based metasurfaces. These seminal works set the first step**stone towards the realization of ultra-flat monolithic nonlinear light sources with emission ranging from the visible to the infrared, efficient sources of correlated photon pairs, as well as electro-optical devices. Here, we review these recent advances, discussing potential perspectives for applications in light conversion and modulation sha** as well as quantum optics, with a critical eye on the potential setbacks and limitations of this emerging field. △ Less

Submitted 1 November, 2022; originally announced November 2022.

Comments: 45 Pages, 6 Figures

On the Automated Detection of Corneal Edema with Second Harmonic Generation Microscopy and Deep Learning

Authors: Stefan R. Anton, Rosa M. Martínez-Ojeda, Radu Hristu, George A. Stanciu, Antonela Toma, Cosmin K. Banica, Enrique J. Fernández, Mikko Huttunen, Juan M. Bueno, Stefan G. Stanciu

Abstract: When the cornea becomes hydrated above its physiologic level it begins to significantly scatter light, loosing transparency and thus impairing eyesight. This condition, known as corneal edema, can be associated with different causes, such as corneal scarring, corneal infection, corneal inflammation, and others, making it difficult to diagnose and quantify. Previous works have shown that Second Har… ▽ More When the cornea becomes hydrated above its physiologic level it begins to significantly scatter light, loosing transparency and thus impairing eyesight. This condition, known as corneal edema, can be associated with different causes, such as corneal scarring, corneal infection, corneal inflammation, and others, making it difficult to diagnose and quantify. Previous works have shown that Second Harmonic Generation Microscopy (SHG) represents a valuable non-linear optical imaging tool to non-invasively identify and monitor changes in the collagen architecture of the cornea, potentially playing a pivotal role in future in-vivo cornea diagnostic methods. However, the interpretation of SHG data can pose significant problems when transferring such approaches to clinical settings, given the low availability of public data sets, and training resources. In this work we explore the use of three Deep Learning models, the highly popular InceptionV3 and ResNet50, alongside FLIMBA, a custom developed architecture, requiring no pre-training, to automatically detect corneal edema in SHG images of porcine cornea. We discuss and evaluate data augmentation strategies tuned to the specifics of the herein addressed application and observe that Deep Learning models building on different architectures provide complementary results. Importantly, we observe that the combined use of such complementary models boosts the overall classification performance in the case of differentiating edematous and healthy corneal tissues, up to an AU-ROC=0.98. These results have potential to be extrapolated to other diagnostics scenarios, such as differentiation of corneal edema in different stages, automated extraction of hydration level of cornea, or automated identification of corneal edema causes, and thus pave the way for novel methods for cornea diagnostics with Deep-Learning assisted non-linear optical imaging. △ Less

Submitted 14 November, 2022; v1 submitted 1 October, 2022; originally announced October 2022.

Comments: 19 pages, 7 figures, 1 Table

Spread of SARS-CoV-2 in a SIS model with vaccination and breakthrough infection

Authors: Ariel Félix Gualtieri, Carolina de la Cal, Augusto Francisco Toma, Pedro Hecht

Abstract: Although previous infection and vaccination provide protection against SARS-CoV-2 infection, both reinfection and breakthrough infection are possible events whose occurrence would increase with time after first exposure to the antigen and with the emergence of new variants of the virus. Periodic vaccination could counteract this decline in protection. In the present work, our aim was to develop an… ▽ More Although previous infection and vaccination provide protection against SARS-CoV-2 infection, both reinfection and breakthrough infection are possible events whose occurrence would increase with time after first exposure to the antigen and with the emergence of new variants of the virus. Periodic vaccination could counteract this decline in protection. In the present work, our aim was to develop and explore a model of SARS-CoV-2 spread with vaccination, reinfection and breakthrough infection. A modified deterministic SIS (Susceptible-Infected-Susceptible) model represented by a system of differential equations was designed. As in any SIS model, the population was divided into susceptible and infected individuals. But in our design, susceptible individuals were, in turn, grouped into three consecutive categories whose susceptibility increases with time after infection or vaccination. The model was studied by means of computer simulations, which were analysed qualitatively. The results obtained show that the prevalence, after oscillating between peaks and valleys, reaches a plateau phase. Moreover, as might be expected, the magnitude of the peaks and plateaus increases as the infection rate rises, the vaccination rate decreases and the rate of decay of protection conferred by vaccination or previous infection increases. Therefore, the present study suggests that, at least under certain conditions, the spread of SARS-CoV-2, although it could experience fluctuations, would finally evolve into an endemic form, with a more or less stable prevalence that would depend on the levels of infection and vaccination, and on the kinetics of post-infection and post-vaccination protection. However, it should be kept in mind that our development is a theoretical scheme with many limitations. For this reason, its predictions should be considered with great care. △ Less

Submitted 3 June, 2022; originally announced June 2022.

Comments: 24 pages with 12 figures

Radially and azimuthally pure vortex beams from phase-amplitude metasurfaces

Authors: Michael de Oliveira, Marco Piccardo, Sahand Eslami, Vincenzo Aglieri, Andrea Toma, Antonio Ambrosio

Abstract: To exploit the full potential of the transverse spatial structure of light using the Laguerre-Gaussian basis, it is necessary to control the azimuthal and radial components of the photons. Vortex phase elements are commonly used to generate these modes of light, offering precise control over the azimuthal index but neglect the radially dependent amplitude term which defines their associated corres… ▽ More To exploit the full potential of the transverse spatial structure of light using the Laguerre-Gaussian basis, it is necessary to control the azimuthal and radial components of the photons. Vortex phase elements are commonly used to generate these modes of light, offering precise control over the azimuthal index but neglect the radially dependent amplitude term which defines their associated corresponding transverse profile. Here we experimentally demonstrate the generation of high purity Laguerre-Gaussian beams with a single step on-axis transformation implemented with a dielectric phase-amplitude metasurface. By vectorially structuring the input beam and projecting it onto an orthogonal polarisation basis, we can sculpt any vortex beam in phase and amplitude. We characterize the azimuthal and radial purity of the generated vortex beams, reaching a purity of 98% for a vortex beam with $\ell=50$ and $p=0$. Furthermore, we comparatively show that the purity of the generated vortex beams outperform those generated with other well-established phase-only metasurface approaches. In addition, we highlight the formation of 'ghost' orbital angular momentum orders from azimuthal gratings (analogous to ghost orders in ruled gratings), which have not been widely studied to date. Our work brings higher-order vortex beams and their unlimited potential within reach of wide adoption. △ Less

Submitted 31 May, 2022; originally announced May 2022.

Comments: article (19 pages, 5 figures) and supplemental (7 pages, 8 figures)

Systematic Comparison of Path Planning Algorithms using PathBench

Authors: Hao-Ya Hsueh, Alexandru-Iosif Toma, Hussein Ali Jaafar, Edward Stow, Riku Murai, Paul H. J. Kelly, Sajad Saeedi

Abstract: Path planning is an essential component of mobile robotics. Classical path planning algorithms, such as wavefront and rapidly-exploring random tree (RRT) are used heavily in autonomous robots. With the recent advances in machine learning, development of learning-based path planning algorithms has been experiencing rapid growth. An unified path planning interface that facilitates the development an… ▽ More Path planning is an essential component of mobile robotics. Classical path planning algorithms, such as wavefront and rapidly-exploring random tree (RRT) are used heavily in autonomous robots. With the recent advances in machine learning, development of learning-based path planning algorithms has been experiencing rapid growth. An unified path planning interface that facilitates the development and benchmarking of existing and new algorithms is needed. This paper presents PathBench, a platform for develo**, visualizing, training, testing, and benchmarking of existing and future, classical and learning-based path planning algorithms in 2D and 3D grid world environments. Many existing path planning algorithms are supported; e.g. A*, Dijkstra, waypoint planning networks, value iteration networks, gated path planning networks; and integrating new algorithms is easy and clearly specified. The benchmarking ability of PathBench is explored in this paper by comparing algorithms across five different hardware systems and three different map types, including built-in PathBench maps, video game maps, and maps from real world databases. Metrics, such as path length, success rate, and computational time, were used to evaluate algorithms. Algorithmic analysis was also performed on a real world robot to demonstrate PathBench's support for Robot Operating System (ROS). PathBench is open source. △ Less

Submitted 6 March, 2022; originally announced March 2022.

Comments: Accepted to Advanced Robotics Journal; 23 pages, 9 figures, 4 tables. arXiv admin note: substantial text overlap with arXiv:2105.01777

All-optical Reconfiguration of Ultrafast Dichroism in Gold Metasurfaces

Authors: Andrea Schirato, Andrea Toma, Remo Proietti Zaccaria, Alessandro Alabastri, Giulio Cerullo, Giuseppe Della Valle, Margherita Maiuri

Abstract: Optical metasurfaces have come into the spotlight as a promising platform for light manipulation at the nanoscale, including ultrafast all-optical control via excitation with femtosecond laser pulses. Recently, dichroic metasurfaces have been exploited to modulate the polarization state of light with unprecedented speed. Here, we theoretically predict and experimentally demonstrate by pump-probe s… ▽ More Optical metasurfaces have come into the spotlight as a promising platform for light manipulation at the nanoscale, including ultrafast all-optical control via excitation with femtosecond laser pulses. Recently, dichroic metasurfaces have been exploited to modulate the polarization state of light with unprecedented speed. Here, we theoretically predict and experimentally demonstrate by pump-probe spectroscopy the capability to reconfigure the ultrafast dichroic signal of a gold metasurface by simply acting on the polarization of the pump pulse, which is shown to reshape the spatio-temporal distribution of the optical perturbation. The photoinduced anisotropic response, driven by out-of-equilibrium carriers and extinguished in a sub-picosecond temporal window, is readily controlled in intensity by tuning the polarization direction of the excitation up to a full sign reversal. This work proves that nonlinear metasurfaces offer the flexibility to tailor their ultrafast optical response in a fully all-optically reconfigurable platform. △ Less

Submitted 10 January, 2022; originally announced January 2022.

Optical vortex crystals with dynamic topologies

Authors: Marco Piccardo, Michael de Oliveira, Andrea Toma, Vincenzo Aglieri, Andrew Forbes, Antonio Ambrosio

Abstract: Vortex crystals are geometric arrays of vortices found in various physics fields, owing their regular internal structure to mutual interactions within a spatially confined system. In optics, vortex crystals may form spontaneously within a nonlinear resonator but their usefulness is limited by the lack of control over their topology. On the other hand, programmable devices used in free space, like… ▽ More Vortex crystals are geometric arrays of vortices found in various physics fields, owing their regular internal structure to mutual interactions within a spatially confined system. In optics, vortex crystals may form spontaneously within a nonlinear resonator but their usefulness is limited by the lack of control over their topology. On the other hand, programmable devices used in free space, like spatial light modulators, allow the design of nearly arbitrary vortex distributions but without any intrinsic dynamics. By combining non-Hermitian optics with on-demand topological transformations enabled by metasurfaces, we report a solid-state laser that generates vortex crystals with mutual interactions and actively-tunable topologies. We demonstrate 10x10 coherent vortex arrays with nonlocal coupling networks that are not limited to nearest-neighbor coupling but rather dictated by the crystal's topology. The vortex crystals exhibit sharp Bragg diffraction peaks, witnessing their coherence and high topological charge purity, which we resolve spatially over the whole lattice by introducing a parallelized analysis technique. By structuring light at the source, we enable complex transformations that allow to arbitrarily partition the orbital angular momentum inside the cavity and to heal topological charge defects, making these resonators a robust and versatile tool for advanced applications in topological optics. △ Less

Submitted 22 July, 2021; originally announced July 2021.

Nonlinear Hall effect as a local probe of plasmonic magnetic hot spots

Authors: Karina A. Guerrero-Becerra, Andrea Tomadin, Andrea Toma, Remo Proietti Zaccaria, Francesco De Angelis, Marco Polini

Abstract: Recently developed plasmonic nanostructures are able to generate intense and localized magnetic hot spots in a large spectral range from the terahertz to the visible. However, a direct measurement of the magnetic field at the hot spot has not been performed yet, due to the absence of magnetic field detectors that work at those high frequencies and that fit the hot-spot area. We propose to place a… ▽ More Recently developed plasmonic nanostructures are able to generate intense and localized magnetic hot spots in a large spectral range from the terahertz to the visible. However, a direct measurement of the magnetic field at the hot spot has not been performed yet, due to the absence of magnetic field detectors that work at those high frequencies and that fit the hot-spot area. We propose to place a graphene ribbon in the hot spot of a plasmonic nanostructure driven by a laser beam, such that a current is generated due to both the magnetic field at the hot spot and the electric field of the laser. We demonstrate that a nonlinear Hall voltage, which can be measured by standard electrical means, builds up across the ribbon, making it possible to directly probe the magnetic field at the hot spot. △ Less

Submitted 7 June, 2021; originally announced June 2021.

Comments: 9 pages, 7 figures

Journal ref: Phys. Rev. B 104, 075434 (2021)

PathBench: A Benchmarking Platform for Classical and Learned Path Planning Algorithms

Authors: Alexandru-Iosif Toma, Hao-Ya Hsueh, Hussein Ali Jaafar, Riku Murai, Paul H. J. Kelly, Sajad Saeedi

Abstract: Path planning is a key component in mobile robotics. A wide range of path planning algorithms exist, but few attempts have been made to benchmark the algorithms holistically or unify their interface. Moreover, with the recent advances in deep neural networks, there is an urgent need to facilitate the development and benchmarking of such learning-based planning algorithms. This paper presents PathB… ▽ More Path planning is a key component in mobile robotics. A wide range of path planning algorithms exist, but few attempts have been made to benchmark the algorithms holistically or unify their interface. Moreover, with the recent advances in deep neural networks, there is an urgent need to facilitate the development and benchmarking of such learning-based planning algorithms. This paper presents PathBench, a platform for develo**, visualizing, training, testing, and benchmarking of existing and future, classical and learned 2D and 3D path planning algorithms, while offering support for Robot Oper-ating System (ROS). Many existing path planning algorithms are supported; e.g. A*, wavefront, rapidly-exploring random tree, value iteration networks, gated path planning networks; and integrating new algorithms is easy and clearly specified. We demonstrate the benchmarking capability of PathBench by comparing implemented classical and learned algorithms for metrics, such as path length, success rate, computational time and path deviation. These evaluations are done on built-in PathBench maps and external path planning environments from video games and real world databases. PathBench is open source. △ Less

Submitted 4 May, 2021; originally announced May 2021.

Comments: The Conference on Robots and Vision (CRV2021), Supplementary Website: https://sites.google.com/view/pathbench/

Waypoint Planning Networks

Authors: Alexandru-Iosif Toma, Hussein Ali Jaafar, Hao-Ya Hsueh, Stephen James, Daniel Lenton, Ronald Clark, Sajad Saeedi

Abstract: With the recent advances in machine learning, path planning algorithms are also evolving; however, the learned path planning algorithms often have difficulty competing with success rates of classic algorithms. We propose waypoint planning networks (WPN), a hybrid algorithm based on LSTMs with a local kernel - a classic algorithm such as A*, and a global kernel using a learned algorithm. WPN produc… ▽ More With the recent advances in machine learning, path planning algorithms are also evolving; however, the learned path planning algorithms often have difficulty competing with success rates of classic algorithms. We propose waypoint planning networks (WPN), a hybrid algorithm based on LSTMs with a local kernel - a classic algorithm such as A*, and a global kernel using a learned algorithm. WPN produces a more computationally efficient and robust solution. We compare WPN against A*, as well as related works including motion planning networks (MPNet) and value iteration networks (VIN). In this paper, the design and experiments have been conducted for 2D environments. Experimental results outline the benefits of WPN, both in efficiency and generalization. It is shown that WPN's search space is considerably less than A*, while being able to generate near optimal results. Additionally, WPN works on partial maps, unlike A* which needs the full map in advance. The code is available online. △ Less

Submitted 1 May, 2021; originally announced May 2021.

Comments: The Conference on Robots and Vision (CRV2021) Supplementary Website: https://sites.google.com/view/waypoint-planning-networks

Transient optical symmetry breaking for ultrafast broadband dichroism in plasmonic metasurfaces

Authors: Andrea Schirato, Margherita Maiuri, Andrea Toma, Silvio Fugattini, Remo Proietti Zaccaria, Paolo Laporta, Peter Nordlander, Giulio Cerullo, Alessandro Alabastri, Giuseppe Della Valle

Abstract: Ultrafast nanophotonics is an emerging research field aimed at the development of nanodevices capable of light modulation with unprecedented speed. A promising approach exploits the optical nonlinearity of nanostructured materials (either metallic or dielectric) to modulate their effective permittivity via interaction with intense ultrashort laser pulses. While the ultrafast temporal dynamics of s… ▽ More Ultrafast nanophotonics is an emerging research field aimed at the development of nanodevices capable of light modulation with unprecedented speed. A promising approach exploits the optical nonlinearity of nanostructured materials (either metallic or dielectric) to modulate their effective permittivity via interaction with intense ultrashort laser pulses. While the ultrafast temporal dynamics of such nanostructures following photoexcitation has been studied in depth, sub-ps transient spatial inhomogeneities taking place at the nanoscale have been so far almost ignored. Here we theoretically predict and experimentally demonstrate that the inhomogeneous space-time distribution of photogenerated hot carriers induces a transient symmetry breaking in a plasmonic metasurface made of highly symmetric metaatoms. The process is fully reversible, and results in a broadband transient dichroic optical response with a recovery of the initial isotropic state in less than 1 picosecond, overcoming the speed bottleneck caused by slower relaxation processes, such as electron-phonon and phonon-phonon scattering. Our results pave the way to the development of ultrafast dichroic devices, capable of Tera bit/s modulation of light polarization. △ Less

Submitted 31 July, 2020; originally announced August 2020.

Photoinduced Temperature Gradients in Sub-wavelength Plasmonic Structures: The Thermoplasmonics of Nanocones

Authors: Joao Cunha, Tian-Long Guo, Alemayehu Nana Koya, Andrea Toma, Mirko Prato, Giuseppe Della Valle, Alessandro Alabastri, Remo Proietti Zaccaria

Abstract: Plasmonic structures are renowned for their capability to efficiently convert light into heat at the nanoscale. However, despite the possibility to generate deep sub-wavelength electromagnetic hot spots, the formation of extremely localized thermal hot spots is an open challenge of research, simply because of the diffusive spread of heat along the whole metallic nanostructure. Here we tackle this… ▽ More Plasmonic structures are renowned for their capability to efficiently convert light into heat at the nanoscale. However, despite the possibility to generate deep sub-wavelength electromagnetic hot spots, the formation of extremely localized thermal hot spots is an open challenge of research, simply because of the diffusive spread of heat along the whole metallic nanostructure. Here we tackle this challenge by exploiting single gold nanocones. We theoretically show how these structures can indeed realize extremely high temperature gradients within the metal, leading to deep sub-wavelength thermal hot spots, owing to their capability of concentrating light at the apex under resonant conditions even under continuous wave illumination. A three-dimensional Finite Element Method model is employed to study the electromagnetic field in the structure and subsequent thermoplasmonic behaviour, in terms of the three-dimensional temperature distribution. We show how the latter is affected by nanocone size, shape, and composition of the surrounding environment. Finally, we anticipate the use of photoinduced temperature gradients in nanocones for applications in optofluidics and thermoelectrics or for thermally induced nanofabrication. △ Less

Submitted 29 May, 2020; originally announced May 2020.

Boosting infrared energy transfer in 3D nanoporous gold antennas

Authors: D. Garoli, E. Calandrini, A. Bozzola, M. Ortolani, S. Cattarin, S. Barison, A. Toma, F. De Angelis

Abstract: The applications of plasmonics to energy transfer from free-space radiation to molecules are currently limited to the visible region of the electromagnetic spectrum due to the intrinsic optical properties of bulk noble metals that support strong electromagnetic field confinement only close to their plasma frequency in the visible/ultraviolet range. In this work, we show that nanoporous gold can be… ▽ More The applications of plasmonics to energy transfer from free-space radiation to molecules are currently limited to the visible region of the electromagnetic spectrum due to the intrinsic optical properties of bulk noble metals that support strong electromagnetic field confinement only close to their plasma frequency in the visible/ultraviolet range. In this work, we show that nanoporous gold can be exploited as plasmonic material for the mid-infrared region to obtain strong electromagnetic field confinement, co-localized with target molecules into the nanopores and resonant with their vibrational frequency. The effective optical response of the nanoporous metal enables the penetration of optical fields deep into the nanopores, where molecules can be loaded thus achieving a more efficient light-matter coupling if compared to bulk gold. In order to realize plasmonic resonators made of nanoporous gold, we develop a nanofabrication method based on polymeric templates for metal deposition and we obtain antenna arrays resonating at mid-infrared wavelengths selected by design. We then coat the antennas with a thin (3 nm) silica layer acting as target dielectric layer for optical energy transfer. We study the strength of the light-matter coupling at the vibrational absorption frequency of silica at 1250 cm-1 through the analysis of the experimental Fano lineshape that is benchmarked against identical structures made of bulk gold. The boost of optical energy transfer from free-space mid-infrared radiation to molecular vibrations in nanoporous 3D nanoantenna arrays can open new application routes for plasmon-enhanced physical-chemical reactions. △ Less

Submitted 31 March, 2018; originally announced April 2018.

Journal ref: Nanoscale 9 (2), 915-922, 2017

Fractal-like plasmonic self-similar material with a tailorable plasma frequency in the near-infrared

Authors: Denis Garoli, Eugenio Calandrini, Angelo Bozzola, Andrea Toma, Sandro Cattarin, Michele Ortolani, Francesco De Angelis

Abstract: In this work, we show that modulating the fractal dimension of nanoporous gold allows its effective dielectric response to be tailored over a wide spectral range of infrared wavelengths. In particular, the plasma edge and effective plasma frequency depend linearly on the fractal dimension, which can be controlled by varying the pore and ligament sizes. Importantly, the fractal porous metal exhibit… ▽ More In this work, we show that modulating the fractal dimension of nanoporous gold allows its effective dielectric response to be tailored over a wide spectral range of infrared wavelengths. In particular, the plasma edge and effective plasma frequency depend linearly on the fractal dimension, which can be controlled by varying the pore and ligament sizes. Importantly, the fractal porous metal exhibits superior plasmonic properties compared to its bulk counterpart. These properties, combined with a longer skin depth on the order of 100-200 nm, enables the penetration of optical energy deep into the nanopores where molecules can be loaded, thus achieving more effective light-matter coupling. These findings may open new pathways to engineering the optical response of fractal-like or self-similar metamaterials without the need for sophisticated lithographic patterning. △ Less

Submitted 21 March, 2018; originally announced March 2018.

Comments: 21 pages, 10 figures

Magnetic Control of Transmission and Helicity of Nano-Structured Optical Beams in Magnetoplasmonic Vortex Lenses

Authors: Nicolò Maccaferri, Yuri Gorodetski, Andrea Toma, Pierfrancesco Zilio, Francesco De Angelis, Denis Garoli

Abstract: We theoretically investigate the generation of far-field propagating optical beams with a desired orbital angular momentum by using an archetypical magnetoplasmonic tip surrounded by a gold spiral slit. The use of a magnetic material can lead to important implications once magneto-optical activity is activated through the application of an external magnetic field. The physical model and the numeri… ▽ More We theoretically investigate the generation of far-field propagating optical beams with a desired orbital angular momentum by using an archetypical magnetoplasmonic tip surrounded by a gold spiral slit. The use of a magnetic material can lead to important implications once magneto-optical activity is activated through the application of an external magnetic field. The physical model and the numerical study presented here introduce the concept of magnetically tunable plasmonic vortex lens, namely a magnetoplasmonic vortex lens, which ensures a tunable selectivity in the polarization state of the generated nanostructured beam. The presented system provides a promising platform for a localized excitation of plasmonic vortices followed by their beaming in the far-field with an active modulation of both light's transmittance and helicity. △ Less

Submitted 23 October, 2017; v1 submitted 7 September, 2017; originally announced September 2017.

Emitter Localisation from Reception Timestamps in Asynchronous Networks

Authors: Nicolo' Facchi, Francesco Gringoli, Fabio Ricciato, Andrea Toma

Abstract: We address the problem of localising a mobile terminal ("blind" node) in unknown position from a set of "anchor" nodes in known positions. The proposed method does not require any form of node synchronisation nor measurement (or control) of the transmission times, which is difficult or anyway costly to achieve in practice. It relies exclusively on reception timestamps collected by the anchor nodes… ▽ More We address the problem of localising a mobile terminal ("blind" node) in unknown position from a set of "anchor" nodes in known positions. The proposed method does not require any form of node synchronisation nor measurement (or control) of the transmission times, which is difficult or anyway costly to achieve in practice. It relies exclusively on reception timestamps collected by the anchor nodes, according to their local clocks, that overhear packets transmitted by the blind node and by (at least one) other transmitting node(s) in known position, e.g., other anchors. The clock differences between the nodes are not eliminated ex ante through clock synchronisation, as in traditional ToA and TDoA methods. Instead, they are counteracted ex post, during the data processing stage, leveraging the data redundancy that is intrinsic to the multiple reception of the same packet by different (anchor) nodes. We validate the proposed method in different experimental settings, indoor and outdoor, using exclusively low-cost Commercial-Off-The-Shelf WiFi devices, achieving sub-meter accuracy in full Line-of-Sight conditions and meter-level accuracy in mild Non-line-of-sight environment. The proposed method does not require that the blind node participate actively to the localisation procedure and can use "opportunistically" any legacy signal or packet available over-the-air for communication purposes. Considering the very minimal requirement on the system - basically, only that anchors in known positions are able to collect and share reception timestamps - the proposed approach can enable practical adoption of opportunistic and/or cooperative localisation on top of existing radio communication systems. △ Less

Submitted 14 July, 2015; v1 submitted 30 July, 2014; originally announced July 2014.

Comments: To appear in Computer Networks

Robust portfolio optimization using pseudodistances

Authors: Aida Toma, Samuela Leoni-Aubin

Abstract: The presence of outliers in financial asset returns is a frequently occuring phenomenon and may lead to unreliable mean-variance optimized portfolios. This fact is due to the unbounded influence that outliers can have on the mean returns and covariance estimators that are inputs in the optimization procedure. In the present paper we consider new robust estimators of location and covariance obtaine… ▽ More The presence of outliers in financial asset returns is a frequently occuring phenomenon and may lead to unreliable mean-variance optimized portfolios. This fact is due to the unbounded influence that outliers can have on the mean returns and covariance estimators that are inputs in the optimization procedure. In the present paper we consider new robust estimators of location and covariance obtained by minimizing an empirical version of a pseudodistance between the assumed model and the true model underlying the data. We prove statistical properties of the new mean and covariance matrix estimators, such as affine equivariance, B-robustness and efficiency. These estimators can be easily used in place of the classical estimators, thereby providing robust optimized portfolios. A Monte Carlo simulation study and an application to real data show the advantages of the proposed approach. △ Less

Submitted 25 May, 2013; originally announced May 2013.

PAMAM6 dendrimers and DNA: pH dependent "beads-on-a-string" behavior revealed by small angle X-ray scattering

Authors: Rolf Dootz, Adriana Cristina Toma, Thomas Pfohl

Abstract: DNA interactions with polycations are not only important for our understanding of chromatin compaction but also for characterizing DNA-binding proteins involved in transcription, replication and repair. DNA is known to form several types of liquid-crystalline phases depending, among other factors, on polycation structure and charge density. Theoretical studies and simulations have predicted the wr… ▽ More DNA interactions with polycations are not only important for our understanding of chromatin compaction but also for characterizing DNA-binding proteins involved in transcription, replication and repair. DNA is known to form several types of liquid-crystalline phases depending, among other factors, on polycation structure and charge density. Theoretical studies and simulations have predicted the wrap** of DNA around spherical positively charged polycations. As a potential mimic of the histone octamer or other DNA wrap** proteins, poly(amido amine) generation 6 (PAMAM6) dendrimers have been chosen for our study. The self-assembly of DNA induced by PAMAM6 has been investigated using small angle X-ray scattering (SAXS) in order to reveal the assemblies' structure dependence on the pH of the environment and on dendrimers concentration. We demonstrate that at pH 8.5 dense phases are formed and characterized by a 2D-columnar hexagonal lattice which is transformed into a 3D hexagonal lattice with increasing dendrimer concentration (charge ratio N/P). Moreover, a systematic analysis of the scattering data collected from diluted samples at pH 8.5 and 5.5 have led us to propose a wrap** scenario of DNA around PAMAM6 at pH 5.5. △ Less

Submitted 12 April, 2011; originally announced April 2011.

Comments: 12 pages, 7 figures

Decomposable Pseudodistances and Applications in Statistical Estimation

Authors: Michel Broniatowski, Aida Toma, Igor Vajda

Abstract: The aim of this paper is to introduce new statistical criterions for estimation, suitable for inference in models with common continuous support. This proposal is in the direct line of a renewed interest for divergence based inference tools imbedding the most classical ones, such as maximum likelihood, Chi-square or Kullback Leibler. General pseudodistances with decomposable structure are consider… ▽ More The aim of this paper is to introduce new statistical criterions for estimation, suitable for inference in models with common continuous support. This proposal is in the direct line of a renewed interest for divergence based inference tools imbedding the most classical ones, such as maximum likelihood, Chi-square or Kullback Leibler. General pseudodistances with decomposable structure are considered, they allowing to define minimum pseudodistance estimators, without using nonparametric density estimators. A special class of pseudodistances indexed by α>0, leading for α\downarrow0 to the Kulback Leibler divergence, is presented in detail. Corresponding estimation criteria are developed and asymptotic properties are studied. The estimation method is then extended to regression models. Finally, some examples based on Monte Carlo simulations are discussed. △ Less

Submitted 8 April, 2011; originally announced April 2011.

MSC Class: 62F12; 62F10

Structural and dynamic properties of linker histone H1 binding to DNA

Authors: Rolf Dootz, Adriana Cristina Toma, Thomas Pfohl

Abstract: Found in all eukaryotic cells, linker histones H1 are known to bind to and rearrange nucleosomal linker DNA. In vitro, the fundamental nature of H1/DNA interactions has attracted wide interest among research communities - for biologists from a chromatin organization deciphering point of view, and for physicists from the study of polyelectrolyte interactions point of view. Hence, H1/DNA binding pro… ▽ More Found in all eukaryotic cells, linker histones H1 are known to bind to and rearrange nucleosomal linker DNA. In vitro, the fundamental nature of H1/DNA interactions has attracted wide interest among research communities - for biologists from a chromatin organization deciphering point of view, and for physicists from the study of polyelectrolyte interactions point of view. Hence, H1/DNA binding processes, structural and dynamical information about these self-assemblies is of broad importance. Targeting a quantitative understanding of H1 induced DNA compaction mechanisms our strategy is based on using small angle X-ray microdiffraction in combination with microfluidics. The usage of microfluidic hydrodynamic focusing devices facilitate a microscale control of these self-assembly processes. In addition, the method enables time-resolved access to structure formation in situ, in particular to transient intermediate states. The observed time dependent structure evolution shows that the interaction of H1 with DNA can be described as a two step process: an initial unspecific binding of H1 to DNA is followed by a rearrangement of molecules within the formed assemblies. The second step is most likely induced by interactions between the charged side chains of the protein and DNA. This leads to an increase in lattice spacing within the DNA/protein assembly and induces a decrease in the correlation length of the mesophases, probably due to a local bending of the DNA. △ Less

Submitted 11 August, 2010; originally announced August 2010.

Comments: 21 pages, 6 figures

Dual divergence estimators and tests: robustness results

Authors: Aida Toma, Michel Broniatowski

Abstract: The class of dual $φ$-divergence estimators (introduced in Broniatowski and Keziou (2009) is explored with respect to robustness through the influence function approach. For scale and location models, this class is investigated in terms of robustness and asymptotic relative efficiency. Some hypothesis tests based on dual divergence criterions are proposed and their robustness properties are stud… ▽ More The class of dual $φ$-divergence estimators (introduced in Broniatowski and Keziou (2009) is explored with respect to robustness through the influence function approach. For scale and location models, this class is investigated in terms of robustness and asymptotic relative efficiency. Some hypothesis tests based on dual divergence criterions are proposed and their robustness properties are studied. The empirical performances of these estimators and tests are illustrated by Monte Carlo simulation for both noncontaminated and contaminated data. △ Less

Submitted 15 December, 2009; v1 submitted 14 December, 2009; originally announced December 2009.

MSC Class: 62F10; 62F03; 62G35