-
Implicit Gaussian process representation of vector fields over arbitrary latent manifolds
Authors:
Robert L. Peach,
Matteo Vinao-Carl,
Nir Grossman,
Michael David,
Emma Mallas,
David Sharp,
Paresh A. Malhotra,
Pierre Vandergheynst,
Adam Gosztolai
Abstract:
Gaussian processes (GPs) are popular nonparametric statistical models for learning unknown functions and quantifying the spatiotemporal uncertainty in data. Recent works have extended GPs to model scalar and vector quantities distributed over non-Euclidean domains, including smooth manifolds appearing in numerous fields such as computer vision, dynamical systems, and neuroscience. However, these a…
▽ More
Gaussian processes (GPs) are popular nonparametric statistical models for learning unknown functions and quantifying the spatiotemporal uncertainty in data. Recent works have extended GPs to model scalar and vector quantities distributed over non-Euclidean domains, including smooth manifolds appearing in numerous fields such as computer vision, dynamical systems, and neuroscience. However, these approaches assume that the manifold underlying the data is known, limiting their practical utility. We introduce RVGP, a generalisation of GPs for learning vector signals over latent Riemannian manifolds. Our method uses positional encoding with eigenfunctions of the connection Laplacian, associated with the tangent bundle, readily derived from common graph-based approximation of data. We demonstrate that RVGP possesses global regularity over the manifold, which allows it to super-resolve and inpaint vector fields while preserving singularities. Furthermore, we use RVGP to reconstruct high-density neural dynamics derived from low-density EEG recordings in healthy individuals and Alzheimer's patients. We show that vector field singularities are important disease markers and that their reconstruction leads to a comparable classification accuracy of disease states to high-density recordings. Thus, our method overcomes a significant practical limitation in experimental and clinical applications.
△ Less
Submitted 17 January, 2024; v1 submitted 28 September, 2023;
originally announced September 2023.
-
Surface Protection and Activation of Mid-IR Plasmonic Waveguides for Spectroscopy of Liquids
Authors:
Mauro David,
Ismael C. Doganlar,
Daniele Nazzari,
Elena Arigliani,
Dominik Wacht,
Masiar Sistani,
Hermann Detz,
Georg Ramer,
Bernhard Lendl,
Walter M. Weber,
Gottfried Strasser,
Borislav Hinkov
Abstract:
Liquid spectroscopy in the mid-infrared spectral range is a very powerful, yet premature technique for selective and sensitive molecule detection. Due to the lack of suitable concepts and materials for versatile miniaturized sensors, it is often still limited to bulky systems and offline analytics. Mid-infrared plasmonics is a promising field of current research for such compact and surface-sensit…
▽ More
Liquid spectroscopy in the mid-infrared spectral range is a very powerful, yet premature technique for selective and sensitive molecule detection. Due to the lack of suitable concepts and materials for versatile miniaturized sensors, it is often still limited to bulky systems and offline analytics. Mid-infrared plasmonics is a promising field of current research for such compact and surface-sensitive structures, enabling new pathways for much-needed photonic integrated sensors. In this work, we focus on extending the concept of Ge/Au-based mid-infrared plasmonic waveguides to enable broadband liquid detection. Through the implementation of high-quality dielectric passivation layers deposited by atomic layer deposition (ALD), we cover the weak and water-soluble Ge native oxide. We show that approximately 10 nm of e.g. Al2O3 or ZrO2 can already protect the plasmonic waveguides for up to 90 min of direct water exposure. This unlocks integrated sensing schemes for broadband molecule detection based on mid-infrared plasmonics. In a proof-of-concept experiment, we further demonstrate that the ZrO2 coated waveguides can be activated by surface functionalization, allowing the selective measurement of diethyl ether at a wavelength of 9.38 μm.
△ Less
Submitted 25 May, 2023;
originally announced May 2023.
-
Advanced Mid-Infrared Plasmonic Waveguides For On-Chip Integrated Photonics
Authors:
Mauro David,
Davide Disnan,
Elena Arigliani,
Anna Lardschneider,
Georg Marschick,
Hanh T. Hoang,
Hermann Detz,
Bernhard Lendl,
Ulrich Schmid,
Gottfried Strasser,
Borislav Hinkov
Abstract:
Long-wave infrared (LWIR, 8-14 um) photonics is a rapidly growing research field within the mid-IR with applications in molecular spectroscopy and optical free-space communication. LWIR-applications are often addressed using rather bulky tabletop-sized free-space optical systems, preventing advanced photonic applications such as rapid-time-scale experiments. Here, device miniaturization into photo…
▽ More
Long-wave infrared (LWIR, 8-14 um) photonics is a rapidly growing research field within the mid-IR with applications in molecular spectroscopy and optical free-space communication. LWIR-applications are often addressed using rather bulky tabletop-sized free-space optical systems, preventing advanced photonic applications such as rapid-time-scale experiments. Here, device miniaturization into photonic integrated circuits (PICs) with maintained optical capabilities is key to revolutionize mid-IR photonics. Sub-wavelength mode confinement in plasmonic structures enabled such miniaturization approaches in the visible-to-near-IR spectral range. However, adopting plasmonics for the LWIR needs suitable low-loss and -dispersion materials with compatible integration strategies to existing mid-IR technology. In this work we further unlock the field of LWIR/mid-IR PICs, by combining photolithographic patterning of organic polymers with dielectric-loaded surface plasmon polariton (DLSPP) waveguides. In particular, polyethylene shows favorable optical properties, including low refractive index and broad transparency between ~2-200 um. We investigate the whole value chain, including design, fabrication, and characterization of polyethylene-based DLSPP waveguides and demonstrate their first-time plasmonic operation and mode guiding capabilities along s-bend structures. Low bending losses of ~1.3 dB and straight-section propagation lengths of ~1 mm, pave the way for unprecedented, complex on-chip mid-IR photonic devices. Moreover, DLSPPs allow full control of the mode parameters (propagation length and guiding capabilities) for precisely addressing advanced sensing and telecommunication applications with chip-scale devices.
△ Less
Submitted 5 May, 2023;
originally announced May 2023.
-
The Belgian repository of fundamental atomic data and stellar spectra (BRASS). I. Cross-matching atomic databases of astrophysical interest
Authors:
Mike Laverick,
Alex Lobel,
Thibault Merle,
Pierre Royer,
Christophe Martayan,
Marc David,
Herman Hensberge,
Emmanuel Thienpont
Abstract:
Fundamental atomic parameters, such as oscillator strengths, play a key role in modelling and understanding the chemical composition of stars in the universe. Despite the significant work underway to produce these parameters for many astrophysically important ions, uncertainties in these parameters remain large and can propagate throughout the entire field of astronomy. The Belgian repository of f…
▽ More
Fundamental atomic parameters, such as oscillator strengths, play a key role in modelling and understanding the chemical composition of stars in the universe. Despite the significant work underway to produce these parameters for many astrophysically important ions, uncertainties in these parameters remain large and can propagate throughout the entire field of astronomy. The Belgian repository of fundamental atomic data and stellar spectra (BRASS) aims to provide the largest systematic and homogeneous quality assessment of atomic data to date in terms of wavelength, atomic and stellar parameter coverage. To prepare for it, we first compiled multiple literature occurrences of many individual atomic transitions, from several atomic databases of astrophysical interest, and assessed their agreement. Several atomic repositories were searched and their data retrieved and formatted in a consistent manner. Data entries from all repositories were cross-matched against our initial BRASS atomic line list to find multiple occurrences of the same transition. Where possible we used a non-parametric cross-match depending only on electronic configurations and total angular momentum values. We also checked for duplicate entries of the same physical transition, within each retrieved repository, using the non-parametric cross-match. We report the cross-matched transitions for each repository and compare their fundamental atomic parameters. We find differences in log(gf) values of up to 2 dex or more. We also find and report that ~2% of our line list and Vienna Atomic Line Database retrievals are composed of duplicate transitions. Finally we provide a number of examples of atomic spectral lines with different log(gf) values, and discuss the impact of these uncertain log(gf) values on quantitative spectroscopy. All cross-matched atomic data and duplicate transitions are available to download at brass.sdf.org.
△ Less
Submitted 20 December, 2017;
originally announced December 2017.
-
Impact of photon cross section systematic uncertainties on Monte Carlo-determined depth-dose distributions
Authors:
Eder Aguirre,
Mario A. Bernal,
Mariano David,
Carlos E. deAlmeida
Abstract:
This work studies the impact of systematic uncertainties associated to interaction cross sections on depth dose curves determined by Monte Carlo simulations. The corresponding sensitivity factors are quantified by changing cross sections in a given amount and determining the variation in the dose. The influence of total cross sections for all particles, photons and only for Compton scattering is a…
▽ More
This work studies the impact of systematic uncertainties associated to interaction cross sections on depth dose curves determined by Monte Carlo simulations. The corresponding sensitivity factors are quantified by changing cross sections in a given amount and determining the variation in the dose. The influence of total cross sections for all particles, photons and only for Compton scattering is addressed. The PENELOPE code was used in all simulations. It was found that photon cross section sensitivity factors depend on depth. In addition, they are positive and negative for depths below and above an equilibrium depth, respectively. At this depth, sensitivity factors are null. The equilibrium depths found in this work agree very well with the mean free path of the corresponding incident photon energy. Using the sensitivity factors reported here, it is possible to estimate the impact of photon cross section uncertainties on the uncertainty of Monte Carlo-determined depth dose curves.
△ Less
Submitted 30 March, 2016; v1 submitted 29 March, 2016;
originally announced March 2016.
-
Weather data analysis based on typical weather sequence analysis. Application: energy building simulation
Authors:
Mathieu David,
Laetitia Adelard,
Francois Garde,
Harry Boyer
Abstract:
In building studies dealing about energy efficiency and comfort, simulation software need relevant weather files with optimal time steps. Few tools generate extreme and mean values of simultaneous hourly data including correlation between the climatic parameters. This paper presents the C++ Runeole software based on typical weather sequences analysis. It runs an analysis process of a stochastic co…
▽ More
In building studies dealing about energy efficiency and comfort, simulation software need relevant weather files with optimal time steps. Few tools generate extreme and mean values of simultaneous hourly data including correlation between the climatic parameters. This paper presents the C++ Runeole software based on typical weather sequences analysis. It runs an analysis process of a stochastic continuous multivariable phenomenon with frequencies properties applied to a climatic database. The database analysis associates basic statistics, PCA (Principal Component Analysis) and automatic classifications. Different ways of applying these methods will be presented. All the results are stored in the Runeole internal database that allows an easy selection of weather sequences. The extreme sequences are used for system and building sizing and the mean sequences are used for the determination of the annual cooling loads as proposed by Audrier-Cros (Audrier-Cros, 1984). This weather analysis was tested with the database of the French weather forecast utility Meteo France. Reunion Island experiences a lot of different micro-climates due to the high altitude (3069m), specific relief, and geographic situation (Tropic of Capricorn). Furthermore Reunion Island has the densest meteorological network in France and is an ideal place to validate the methodology with different climates. To test the efficiency of such analysis, simulations using the resulting weather sequences were carried out with the building simulation software CODYRUN. This analysis is the first step of a more global research concerning weather data generation. Future work will permit whole hourly typical meteorological year generation using neural networks.
△ Less
Submitted 25 September, 2014;
originally announced September 2014.
