Showing 101–150 of 476 results for author: Evans, D

Understanding Intrinsic Robustness Using Label Uncertainty

Authors: Xiao Zhang, David Evans

Abstract: A fundamental question in adversarial machine learning is whether a robust classifier exists for a given task. A line of research has made some progress towards this goal by studying the concentration of measure, but we argue standard concentration fails to fully characterize the intrinsic robustness of a classification problem since it ignores data labels which are essential to any classification… ▽ More A fundamental question in adversarial machine learning is whether a robust classifier exists for a given task. A line of research has made some progress towards this goal by studying the concentration of measure, but we argue standard concentration fails to fully characterize the intrinsic robustness of a classification problem since it ignores data labels which are essential to any classification task. Building on a novel definition of label uncertainty, we empirically demonstrate that error regions induced by state-of-the-art models tend to have much higher label uncertainty than randomly-selected subsets. This observation motivates us to adapt a concentration estimation algorithm to account for label uncertainty, resulting in more accurate intrinsic robustness measures for benchmark image classification problems. △ Less

Submitted 17 March, 2022; v1 submitted 7 July, 2021; originally announced July 2021.

Comments: ICLR 2022; 23 pages, 8 figures, 1 table

Topological digestion drives time-varying rheology of entangled DNA fluids

Authors: Davide Michieletto, Philip Neill, Simon Weir, David Evans, Natalie Crist, Vincent Martinez, Rae Robertson-Anderson

Abstract: Understanding and controlling the rheology of polymeric complex fluids that are pushed out-of-equilibrium is a fundamental problem in both industry and biology. For example, to package, repair, and replicate DNA, cells use enzymes to constantly manipulate DNA topology, length, and structure. Inspired by this, here we engineer and study DNA-based complex fluids that undergo enzymatically-driven top… ▽ More Understanding and controlling the rheology of polymeric complex fluids that are pushed out-of-equilibrium is a fundamental problem in both industry and biology. For example, to package, repair, and replicate DNA, cells use enzymes to constantly manipulate DNA topology, length, and structure. Inspired by this, here we engineer and study DNA-based complex fluids that undergo enzymatically-driven topological and architectural alterations via restriction endonuclease (RE) reactions. We show that these systems display time-dependent rheological properties that depend on the concentrations and properties of the comprising DNA and REs. Through time-resolved microrheology experiments and Brownian Dynamics simulations, we show that conversion of supercoiled to linear DNA topology leads to a monotonic increase in viscosity. On the other hand, the viscosity of entangled linear DNA undergoing fragmentation displays a universal decrease that we rationalize using living polymer theory. Finally, to showcase the tunability of these behaviours, we design a DNA fluid that exhibits a time-dependent increase, followed by a temporally-gated decrease, of its viscosity. Our results present a class of polymeric fluids that leverage naturally occurring enzymes to drive diverse time-varying rheology by performing architectural alterations to the constituents. △ Less

Submitted 17 June, 2022; v1 submitted 16 June, 2021; originally announced June 2021.

ALICE Central Trigger System for LHC Run 3

Authors: Jakub Kvapil, Anju Bhasin, Marek Bombara, David Evans, Anton Jusko, Alexander Kluge, Marian Krivda, Ivan Kralik, Roman Lietava, Sanket Kumar Nayak, Simone Ragoni, Orlando Villalobos Baillie

Abstract: A major upgrade of the ALICE experiment is in progress and will result in high-rate data taking during LHC Run 3 (2022-2024). The LHC interaction rate at Point 2 where the ALICE experiment is located will be increased to $50\ \mathrm{kHz}$ in Pb--Pb collisions and $1\ \mathrm{MHz}$ in pp collisions. The ALICE experiment will be able to read out data at these interaction rates leading to an increas… ▽ More A major upgrade of the ALICE experiment is in progress and will result in high-rate data taking during LHC Run 3 (2022-2024). The LHC interaction rate at Point 2 where the ALICE experiment is located will be increased to $50\ \mathrm{kHz}$ in Pb--Pb collisions and $1\ \mathrm{MHz}$ in pp collisions. The ALICE experiment will be able to read out data at these interaction rates leading to an increase of the collected luminosity by a factor of up to about 100 with respect to LHC Runs 1 and 2. To satisfy these requirements, a new readout system has been developed for most of the ALICE detectors, allowing the full readout of the data at the required interaction rates without the need for a hardware trigger selection. A novel trigger and timing distribution system will be implemented, based on Passive Optical Network (PON) and GigaBit Transceiver (GBT) technology. To assure backward compatibility a triggered mode based on RD12 Trigger-Timing-Control (TTC) technology, as used in the previous LHC runs, will be maintained and re-implemented under the new Central Trigger System (CTS). A new universal ALICE Trigger Board (ATB) based on the Xilinx Kintex Ultrascale FPGA has been designed to function as a Central Trigger Processor (CTP), Local Trigger Unit (LTU), and monitoring interfaces. In this paper, this new hybrid multilevel system with continuous readout will be described, together with the triggering mechanism and algorithms. An overview of the CTS, the design of the ATB and the different communication protocols will be presented. △ Less

Submitted 30 June, 2021; v1 submitted 15 June, 2021; originally announced June 2021.

Comments: 10 pages, 6 figures, submitted as a proceedings paper for 25th International Conference on Computing in High-Energy and Nuclear Physics (vCHEP 2021) to European Physical Journal (EPJ) Web of Conferences, May 17th-2th 2021, v2: corrected institution list

Formalizing Distribution Inference Risks

Authors: Anshuman Suri, David Evans

Abstract: Property inference attacks reveal statistical properties about a training set but are difficult to distinguish from the primary purposes of statistical machine learning, which is to produce models that capture statistical properties about a distribution. Motivated by Yeom et al.'s membership inference framework, we propose a formal and generic definition of property inference attacks. The proposed… ▽ More Property inference attacks reveal statistical properties about a training set but are difficult to distinguish from the primary purposes of statistical machine learning, which is to produce models that capture statistical properties about a distribution. Motivated by Yeom et al.'s membership inference framework, we propose a formal and generic definition of property inference attacks. The proposed notion describes attacks that can distinguish between possible training distributions, extending beyond previous property inference attacks that infer the ratio of a particular type of data in the training data set. In this paper, we show how our definition captures previous property inference attacks as well as a new attack that reveals the average degree of nodes of a training graph and report on experiments giving insight into the potential risks of property inference attacks. △ Less

Submitted 24 September, 2021; v1 submitted 7 June, 2021; originally announced June 2021.

Comments: ICML 2021 Workshop on Theory and Practice of Differential Privacy. Longer version of work available at arXiv:2109.06024 Update: Labelling error for Census[Race], where graphs were mirror-images because of 1-ratio being used instead of the ratio. Comparison with SOTA also updated; conclusions remain unchanged

Internal calibration of Gaia BP/RP low-resolution spectra

Authors: J. M. Carrasco, M. Weiler, C. Jordi, C. Fabricius, F. De Angeli, D. W. Evans, F. van Leeuwen, M. Riello, P. Montegriffo

Abstract: The full third Gaia data release will provide the calibrated spectra obtained with the blue and red Gaia slit-less spectrophotometers. The main challenge when facing Gaia spectral calibration is that no lamp spectra or flat fields are available during the mission. Also, the significant size of the line spread function with respect to the dispersion of the prisms produces alien photons contaminatin… ▽ More The full third Gaia data release will provide the calibrated spectra obtained with the blue and red Gaia slit-less spectrophotometers. The main challenge when facing Gaia spectral calibration is that no lamp spectra or flat fields are available during the mission. Also, the significant size of the line spread function with respect to the dispersion of the prisms produces alien photons contaminating neighbouring positions of the spectra. This makes the calibration special and different from standard approaches. This work gives a detailed description of the internal calibration model to obtain the spectrophotometric data in the Gaia catalogue. The main purpose of the internal calibration is to bring all the epoch spectra onto a common flux and pixel (pseudo-wavelength) scale, taking into account variations over the focal plane and with time, producing a mean spectrum from all the observations of the same source. In order to describe all observations in a common mean flux and pseudo-wavelength scale, we construct a suitable representation of the internally calibrated mean spectra via basis functions and we describe the transformation between non calibrated epoch spectra and calibrated mean spectra via a discrete convolution, parametrising the convolution kernel to recover the relevant coefficients. The model proposed here is able to combine all observations into a mean instrument to allow the comparison of different sources and observations obtained with different instrumental conditions along the mission and the generation of mean spectra from a number of observations of the same source. The output of this model provides the internal mean spectra, not as a sampled function (flux and wavelength), but as a linear combination of basis functions, although sampled spectra can easily be derived from them. △ Less

Submitted 30 June, 2021; v1 submitted 3 June, 2021; originally announced June 2021.

Comments: 20 pages, 22 figures

Journal ref: A&A 652, A86 (2021)

Gaia Photometric Science Alerts

Authors: S. T. Hodgkin, D. L. Harrison, E. Breedt, T. Wevers, G. Rixon, A. Delgado, A. Yoldas, Z. Kostrzewa-Rutkowska, Ł. Wyrzykowski, M. van Leeuwen, N. Blagorodnova, H. Campbell, D. Eappachen, M. Fraser, N. Ihanec, S. E. Koposov, K. Kruszyńska, G. Marton, K. A. Rybicki, A. G. A. Brown, P. W. Burgess, G. Busso, S. Cowell, F. De Angeli, C. Diener , et al. (86 additional authors not shown)

Abstract: Since July 2014, the Gaia mission has been engaged in a high-spatial-resolution, time-resolved, precise, accurate astrometric, and photometric survey of the entire sky. Aims: We present the Gaia Science Alerts project, which has been in operation since 1 June 2016. We describe the system which has been developed to enable the discovery and publication of transient photometric events as seen by G… ▽ More Since July 2014, the Gaia mission has been engaged in a high-spatial-resolution, time-resolved, precise, accurate astrometric, and photometric survey of the entire sky. Aims: We present the Gaia Science Alerts project, which has been in operation since 1 June 2016. We describe the system which has been developed to enable the discovery and publication of transient photometric events as seen by Gaia. Methods: We outline the data handling, timings, and performances, and we describe the transient detection algorithms and filtering procedures needed to manage the high false alarm rate. We identify two classes of events: (1) sources which are new to Gaia and (2) Gaia sources which have undergone a significant brightening or fading. Validation of the Gaia transit astrometry and photometry was performed, followed by testing of the source environment to minimise contamination from Solar System objects, bright stars, and fainter near-neighbours. Results: We show that the Gaia Science Alerts project suffers from very low contamination, that is there are very few false-positives. We find that the external completeness for supernovae, $C_E=0.46$, is dominated by the Gaia scanning law and the requirement of detections from both fields-of-view. Where we have two or more scans the internal completeness is $C_I=0.79$ at 3 arcsec or larger from the centres of galaxies, but it drops closer in, especially within 1 arcsec. Conclusions: The per-transit photometry for Gaia transients is precise to 1 per cent at $G=13$, and 3 per cent at $G=19$. The per-transit astrometry is accurate to 55 milliarcseconds when compared to Gaia DR2. The Gaia Science Alerts project is one of the most homogeneous and productive transient surveys in operation, and it is the only survey which covers the whole sky at high spatial resolution (subarcsecond), including the Galactic plane and bulge. △ Less

Submitted 2 June, 2021; originally announced June 2021.

Comments: 26 pages, 26 figures, accepted for publication in Astronomy & Astrophysics

Journal ref: A&A 652, A76 (2021)

Statistical tests based on Rényi entropy estimation

Authors: Mehmet Siddik Cadirci, Dafydd Evans, Nikolai Leonenko, Oleg Seleznjev

Abstract: Entropy and its various generalizations are important in many fields, including mathematical statistics, communication theory, physics and computer science, for characterizing the amount of information associated with a probability distribution. In this paper we propose goodness-of-fit statistics for the multivariate Student and multivariate Pearson type II distributions, based on the maximum entr… ▽ More Entropy and its various generalizations are important in many fields, including mathematical statistics, communication theory, physics and computer science, for characterizing the amount of information associated with a probability distribution. In this paper we propose goodness-of-fit statistics for the multivariate Student and multivariate Pearson type II distributions, based on the maximum entropy principle and a class of estimators for Rényi entropy based on nearest neighbour distances. We prove the L^2-consistency of these statistics using results on the subadditivity of Euclidean functionals on nearest neighbour graphs, and investigate their rate of convergence and asymptotic distribution using Monte Carlo methods. In addition we present a novel iterative method for estimating the shape parameter of the multivariate Student and multivariate Pearson type II distributions. △ Less

Submitted 1 June, 2021; originally announced June 2021.

First Dark Matter Search Results From Coherent CAPTAIN-Mills

Authors: A. A. Aguilar-Arevalo, S. Biedron, J. Boissevain, M. Borrego, M. Chavez-Estrada, A. Chavez, J. M. Conrad, R. L. Cooper, A. Diaz, J. R. Distel, J. D'Olivo, E. Dunton, B. Dutta, A. Elliott, D. Evans, D. Fields, J. Greenwood, M. Gold, J. Gordon, E. D. Guarincerri, E. C. Huang, N. Kamp, C. Kelsey, K. Knickerbocker, R. Lake , et al. (25 additional authors not shown)

Abstract: This paper describes the operation of the Coherent CAPTAIN-Mills (CCM) detector located at the Lujan Neutron Science Center (LANSCE) at Los Alamos National Laboratory (LANL). CCM is a 10-ton liquid argon (LAr) detector located 20 meters from a high flux neutron/neutrino source and is designed to search for sterile neutrinos ($ν_s$) and light dark matter (LDM). An engineering run was performed in F… ▽ More This paper describes the operation of the Coherent CAPTAIN-Mills (CCM) detector located at the Lujan Neutron Science Center (LANSCE) at Los Alamos National Laboratory (LANL). CCM is a 10-ton liquid argon (LAr) detector located 20 meters from a high flux neutron/neutrino source and is designed to search for sterile neutrinos ($ν_s$) and light dark matter (LDM). An engineering run was performed in Fall 2019 to study the characteristics of the CCM120 detector by searching for coherent scattering signals consistent with $ν_s$'s and LDM resulting from $π^+$ and $π^0$ decays in the tungsten target. New parameter space in a leptophobic dark matter model was excluded for DM masses between $\sim2.0$ and 30 MeV. The lessons learned from this run have guided the development and construction of the new CCM200 detector that will begin operations in 2021 and significantly improve on these searches. △ Less

Submitted 19 May, 2022; v1 submitted 28 May, 2021; originally announced May 2021.

Report number: LA-UR-21-24983

Journal ref: Physical Review D Vol. 106, No. 1 (2022)

The ultra-hot-Jupiter KELT-16 b: Dynamical Evolution and Atmospheric Properties

Authors: L. Mancini, J. Southworth, L. Naponiello, O. Basturk, D. Barbato, F. Biagiotti, I. Bruni, L. Cabona, G. D'Ago, M. Damasso, A. Erdem, D. Evans, Th. Henning, O. Ozturk, D. Ricci, A. Sozzetti, J. Tregloan-Reed, S. Yalcinkayaz

Abstract: We present broad-band photometry of 30 planetary transits of the ultra-hot Jupiter KELT-16b, using five medium-class telescopes. The transits were monitored through standard B, V, R, I filters and four were simultaneously observed from different places, for a total of 36 new light curves. We used these new photometric data and those from the TESS space telescope to review the main physical propert… ▽ More We present broad-band photometry of 30 planetary transits of the ultra-hot Jupiter KELT-16b, using five medium-class telescopes. The transits were monitored through standard B, V, R, I filters and four were simultaneously observed from different places, for a total of 36 new light curves. We used these new photometric data and those from the TESS space telescope to review the main physical properties of the KELT-16 planetary system. Our results agree with previous measurements but are more precise. We estimated the mid-transit times for each of these transits and combined them with others from the literature to obtain 69 epochs, with a time baseline extending over more than four years, and searched for transit time variations. We found no evidence for a period change, suggesting a lower limit for orbital decay at 8 Myr, with a lower limit on the reduced tidal quality factor of $Q^{\prime}_{\star}>(1.9 \pm 0.8) \times 10^5$ with $95\%$ confidence. We built up an observational, low-resolution transmission spectrum of the planet, finding evidence of the presence of optical absorbers, although with a low significance. Using TESS data, we reconstructed the phase curve finding that KELT-16b has a phase offset of $25.25 \pm 14.03$ $^{\circ}$E, a day- and night-side brightness temperature of $3190 \pm 61$ K and $2668 \pm 56$ K, respectively. Finally, we compared the flux ratio of the planet over its star at the TESS and Spitzer wavelengths with theoretical emission spectra, finding evidence of a temperature inversion in the planet's atmosphere, the chemical composition of which is preferably oxygen-rich rather than carbon-rich. △ Less

Submitted 5 November, 2021; v1 submitted 3 May, 2021; originally announced May 2021.

Comments: 17 pages, 16 figures, Accepted for publication in Monthly Notices of the Royal Astronomical Society

Journal ref: MNRAS 509, 1447-1464 (2022)

Stealthy Backdoors as Compression Artifacts

Authors: Yulong Tian, Fnu Suya, Fengyuan Xu, David Evans

Abstract: In a backdoor attack on a machine learning model, an adversary produces a model that performs well on normal inputs but outputs targeted misclassifications on inputs containing a small trigger pattern. Model compression is a widely-used approach for reducing the size of deep learning models without much accuracy loss, enabling resource-hungry models to be compressed for use on resource-constrained… ▽ More In a backdoor attack on a machine learning model, an adversary produces a model that performs well on normal inputs but outputs targeted misclassifications on inputs containing a small trigger pattern. Model compression is a widely-used approach for reducing the size of deep learning models without much accuracy loss, enabling resource-hungry models to be compressed for use on resource-constrained devices. In this paper, we study the risk that model compression could provide an opportunity for adversaries to inject stealthy backdoors. We design stealthy backdoor attacks such that the full-sized model released by adversaries appears to be free from backdoors (even when tested using state-of-the-art techniques), but when the model is compressed it exhibits highly effective backdoors. We show this can be done for two common model compression techniques -- model pruning and model quantization. Our findings demonstrate how an adversary may be able to hide a backdoor as a compression artifact, and show the importance of performing security tests on the models that will actually be deployed not their precompressed version. △ Less

Submitted 30 April, 2021; originally announced April 2021.

Comments: 20 pages, 9 figures, 14 tables

Symplectic cohomology of compound Du Val singularities

Authors: Jonathan David Evans, Yanki Lekili

Abstract: We compute symplectic cohomology for Milnor fibres of certain compound Du Val singularities that admit small resolution by using homological mirror symmetry. Our computations suggest a new conjecture that the existence of a small resolution has strong implications for the symplectic cohomology and conversely. We also use our computations to give a contact invariant of the link of the singularities… ▽ More We compute symplectic cohomology for Milnor fibres of certain compound Du Val singularities that admit small resolution by using homological mirror symmetry. Our computations suggest a new conjecture that the existence of a small resolution has strong implications for the symplectic cohomology and conversely. We also use our computations to give a contact invariant of the link of the singularities and thereby distinguish many contact structures on connected sums of $S^2 \times S^3$. △ Less

Submitted 8 June, 2022; v1 submitted 23 April, 2021; originally announced April 2021.

Comments: 41 pages; v3 incorporated referees' comments. Accepted for publication in Annales Henri Lebesgue

Chiral Polarization Textures Induced by the Flexoelectric Effect in Ferroelectric Nanocylinders

Authors: Anna N. Morozovska, Riccardo Hertel, Salia Cherifi-Hertel, Victor Yu. Reshetnyak, Eugene A. Eliseev, Dean R. Evans

Abstract: Polar chiral structures have recently attracted much interest within the scientific community, as they pave the way towards innovative device concepts similar to the developments achieved in nanomagnetism. Despite the growing interest, many fundamental questions related to the mechanisms controlling the appearance and stability of ferroelectric topological structures remain open. In this context,… ▽ More Polar chiral structures have recently attracted much interest within the scientific community, as they pave the way towards innovative device concepts similar to the developments achieved in nanomagnetism. Despite the growing interest, many fundamental questions related to the mechanisms controlling the appearance and stability of ferroelectric topological structures remain open. In this context, ferroelectric nanoparticles provide a flexible playground for such investigations. Here, we present a theoretical study of ferroelectric polar textures in a cylindrical core-shell nanoparticle. The calculations reveal a chiral polarization structure containing two oppositely oriented diffuse axial domains located near the cylinder ends, separated by a region with a zero-axial polarization. We name this polarization configuration "flexon" to underline the flexoelectric nature of its axial polarization. Analytical calculations and numerical simulation results show that the flexon's chirality can be switched by reversing the sign of the flexoelectric coefficient. Furthermore, the anisotropy of the flexoelectric coupling is found to critically influence the polarization texture and domain morphology. The flexon rounded shape, combined with its distinct chiral properties and the localization nature near the surface, are reminiscent of Chiral Bobber structures in magnetism. In the azimuthal plane, the flexon displays the polarization state of a vortex with an axially polarized core region, i.e., a meron. The flexoelectric effect, which couples the electric polarization and elastic strain gradients, plays a determining role in the stabilization of these chiral states. We discuss similarities between this interaction and the recently predicted ferroelectric Dyzaloshinskii-Moriya interaction leading to chiral polarization states. △ Less

Submitted 27 August, 2021; v1 submitted 1 April, 2021; originally announced April 2021.

Comments: 53 pages, 5 figures, supplementary materials on 31 pages

Journal ref: Phys. Rev. B 104, 054118 (2021)

Improved Estimation of Concentration Under $\ell_p$-Norm Distance Metrics Using Half Spaces

Authors: Jack Prescott, Xiao Zhang, David Evans

Abstract: Concentration of measure has been argued to be the fundamental cause of adversarial vulnerability. Mahloujifar et al. presented an empirical way to measure the concentration of a data distribution using samples, and employed it to find lower bounds on intrinsic robustness for several benchmark datasets. However, it remains unclear whether these lower bounds are tight enough to provide a useful app… ▽ More Concentration of measure has been argued to be the fundamental cause of adversarial vulnerability. Mahloujifar et al. presented an empirical way to measure the concentration of a data distribution using samples, and employed it to find lower bounds on intrinsic robustness for several benchmark datasets. However, it remains unclear whether these lower bounds are tight enough to provide a useful approximation for the intrinsic robustness of a dataset. To gain a deeper understanding of the concentration of measure phenomenon, we first extend the Gaussian Isoperimetric Inequality to non-spherical Gaussian measures and arbitrary $\ell_p$-norms ($p \geq 2$). We leverage these theoretical insights to design a method that uses half-spaces to estimate the concentration of any empirical dataset under $\ell_p$-norm distance metrics. Our proposed algorithm is more efficient than Mahloujifar et al.'s, and our experiments on synthetic datasets and image benchmarks demonstrate that it is able to find much tighter intrinsic robustness bounds. These tighter estimates provide further evidence that rules out intrinsic dataset concentration as a possible explanation for the adversarial vulnerability of state-of-the-art classifiers. △ Less

Submitted 23 March, 2021; originally announced March 2021.

Comments: ICLR 2021

The Gaia spectrophotometric standard stars survey -- V. Preliminary flux tables for the calibration of Gaia DR2 and (E)DR3

Authors: E. Pancino, N. Sanna, G. Altavilla, S. Marinoni, M. Rainer, G. Cocozza, S. Ragaini, S. Galleti, M. Bellazzini, A. Bragaglia, G. Tessicini, H. Voss, J. M. Carrasco, C. Jordi, D. L. Harrison, F. De Angeli, D. W. Evans, G. Fanari

Abstract: We present the flux tables of the spectro-photometric standard stars (SPSS) used to calibrate in flux the Gaia DR2 and (E)DR3 data releases. The latest SPSS grid version contains 112 stars, whose flux tables agree to better than 1% with the CALSPEC spectra of 11 flux standards for the calibration of the Hubble Space Telescope. The synthetic magnitudes computed on the SPSS spectra also agree to bet… ▽ More We present the flux tables of the spectro-photometric standard stars (SPSS) used to calibrate in flux the Gaia DR2 and (E)DR3 data releases. The latest SPSS grid version contains 112 stars, whose flux tables agree to better than 1% with the CALSPEC spectra of 11 flux standards for the calibration of the Hubble Space Telescope. The synthetic magnitudes computed on the SPSS spectra also agree to better than 1% with the Landolt magnitudes of 37 stars in common. The typical spreads in both comparisons are of the order of 1%. These uncertainties already meet the initial requirements for the Gaia SPSS project, but further improvements are expected in the next SPSS versions, that will be used to calibrate future Gaia releases. We complement the SPSS flux tables with literature spectra of 60 additional stars that did not pass all the criteria to be SPSS, the Passband Validation Library (PVL). The PVL contains stars of extreme spectral types, such as bright O and B stars and late M stars and brown dwarfs, and was useful to investigate systematic effects in the previous Gaia DR2 release and to minimize them in the EDR3 one. The PVL literature spectra are recalibrated as accurately as possible onto the SPSS reference scale, so that the two sets together can be used in a variety of validation and comparison studies. △ Less

Submitted 15 March, 2021; v1 submitted 12 March, 2021; originally announced March 2021.

Comments: 17 pages, 17 figures, accepted for publication by MNRAS

Gaia Early Data Release 3: The Galactic anticentre

Authors: Gaia Collaboration, T. Antoja, P. McMillan, G. Kordopatis, P. Ramos, A. Helmi, E. Balbinot, T. Cantat-Gaudin, L. Chemin, F. Figueras, C. Jordi, S. Khanna, M. Romero-Gomez, G. Seabroke, A. G. A. Brown, A. Vallenari, T. Prusti, J. H. J. de Bruijne, C. Babusiaux, M. Biermann, O. L. Creevey, D. W. Evans, L. Eyer, A. Hutton, F. Jansen , et al. (395 additional authors not shown)

Abstract: We aim to demonstrate the scientific potential of the Gaia Early Data Release 3 (EDR3) for the study of the Milky Way structure and evolution. We used astrometric positions, proper motions, parallaxes, and photometry from EDR3 to select different populations and components and to calculate the distances and velocities in the direction of the anticentre. We explore the disturbances of the current d… ▽ More We aim to demonstrate the scientific potential of the Gaia Early Data Release 3 (EDR3) for the study of the Milky Way structure and evolution. We used astrometric positions, proper motions, parallaxes, and photometry from EDR3 to select different populations and components and to calculate the distances and velocities in the direction of the anticentre. We explore the disturbances of the current disc, the spatial and kinematical distributions of early accreted versus in-situ stars, the structures in the outer parts of the disc, and the orbits of open clusters Berkeley 29 and Saurer 1. We find that: i) the dynamics of the Galactic disc are very complex with vertical asymmetries, and new correlations, including a bimodality with disc stars with large angular momentum moving vertically upwards from below the plane, and disc stars with slightly lower angular momentum moving preferentially downwards; ii) we resolve the kinematic substructure (diagonal ridges) in the outer parts of the disc for the first time; iii) the red sequence that has been associated with the proto-Galactic disc that was present at the time of the merger with Gaia-Enceladus-Sausage is currently radially concentrated up to around 14 kpc, while the blue sequence that has been associated with debris of the satellite extends beyond that; iv) there are density structures in the outer disc, both above and below the plane, most probably related to Monoceros, the Anticentre Stream, and TriAnd, for which the Gaia data allow an exhaustive selection of candidate member stars and dynamical study; and v) the open clusters Berkeley~29 and Saurer~1, despite being located at large distances from the Galactic centre, are on nearly circular disc-like orbits. We demonstrate how, once again, the Gaia are crucial for our understanding of the different pieces of our Galaxy and their connection to its global structure and history. △ Less

Submitted 26 April, 2021; v1 submitted 14 January, 2021; originally announced January 2021.

Comments: Gaia EDR3 performance verification paper, version 2 closer to published version in A&A, complete list of authors

Journal ref: A&A 649, A8 (2021)

Gaia Early Data Release 3: Modelling and calibration of Gaia's point and line spread functions

Authors: N. Rowell, M. Davidson, L. Lindegren, F. van Leeuwen, J. Castañeda, C. Fabricius, U. Bastian, N. C. Hambly, J. Hernández, A. Bombrun, D. W. Evans, F. De Angeli, M. Riello, D. Busonero, C. Crowley, A. Mora, U. Lammers, G. Gracia, J. Portell, M. Biermann, A. G. A. Brown

Abstract: Context: The unprecedented astrometric precision of the Gaia mission relies on accurate estimates of the locations of sources in the Gaia data stream. This is ultimately performed by point spread function (PSF) fitting, which in turn requires an accurate reconstruction of the PSF. Gaia Early Data Release 3 (EDR3) will, for the first time, use a PSF calibration that models several of the strongest… ▽ More Context: The unprecedented astrometric precision of the Gaia mission relies on accurate estimates of the locations of sources in the Gaia data stream. This is ultimately performed by point spread function (PSF) fitting, which in turn requires an accurate reconstruction of the PSF. Gaia Early Data Release 3 (EDR3) will, for the first time, use a PSF calibration that models several of the strongest dependences, leading to signficantly reduced systematic errors. Aims: We describe the PSF model and calibration pipeline implemented for Gaia EDR3, including an analysis of the calibration results over the 34 months of data. We include a discussion of the limitations of the current pipeline and directions for future releases. This will be of use both to users of Gaia data and as a reference for other precision astrometry missions. Methods: We develop models of the 1D line spread function (LSF) and 2D PSF profiles based on a linear combination of basis components. We fit the models to selected primary sources in independent time ranges, using simple parameterisations for the colour and other dependences. Variation in time is smoothed by merging the independent calibrations in a square root information filter, with resets at certain mission events that induce a discontinuous change in the PSF. Results: The PSF calibration shows strong time and colour dependences that accurately reproduce the varying state of the Gaia astrometric instrument. Analysis of the residuals reveals both the performance and the limitations of the current models and calibration pipeline, and indicates the directions for future development. Conclusions: The PSF modelling and calibration carried out for Gaia EDR3 represents a major step forwards in the data processing and will lead to reduced systematic errors in the core mission data products. Further significant improvements are expected in the future data releases. △ Less

Submitted 3 December, 2020; originally announced December 2020.

Comments: Accepted by A&A for inclusion in Gaia EDR3 special issue

Journal ref: A&A 649, A11 (2021)

Gaia Early Data Release 3: The Gaia Catalogue of Nearby Stars

Authors: Gaia Collaboration, R. L. Smart, L. M. Sarro, J. Rybizki, C. Reylé, A. C. Robin, N. C. Hambly, U. Abbas, M. A. Barstow, J. H. J. de Bruijne, B. Bucciarelli, J. M. Carrasco, W. J. Cooper, S. T. Hodgkin, E. Masana, D. Michalik, J. Sahlmann, A. Sozzetti, A. G. A. Brown, A. Vallenari, T. Prusti, C. Babusiaux, M. Biermann, O. L. Creevey, D. W. Evans , et al. (398 additional authors not shown)

Abstract: We produce a clean and well-characterised catalogue of objects within 100\,pc of the Sun from the \G\ Early Data Release 3. We characterise the catalogue through comparisons to the full data release, external catalogues, and simulations. We carry out a first analysis of the science that is possible with this sample to demonstrate its potential and best practices for its use. The selection of obj… ▽ More We produce a clean and well-characterised catalogue of objects within 100\,pc of the Sun from the \G\ Early Data Release 3. We characterise the catalogue through comparisons to the full data release, external catalogues, and simulations. We carry out a first analysis of the science that is possible with this sample to demonstrate its potential and best practices for its use. The selection of objects within 100\,pc from the full catalogue used selected training sets, machine-learning procedures, astrometric quantities, and solution quality indicators to determine a probability that the astrometric solution is reliable. The training set construction exploited the astrometric data, quality flags, and external photometry. For all candidates we calculated distance posterior probability densities using Bayesian procedures and mock catalogues to define priors. Any object with reliable astrometry and a non-zero probability of being within 100\,pc is included in the catalogue. We have produced a catalogue of \NFINAL\ objects that we estimate contains at least 92\% of stars of stellar type M9 within 100\,pc of the Sun. We estimate that 9\% of the stars in this catalogue probably lie outside 100\,pc, but when the distance probability function is used, a correct treatment of this contamination is possible. We produced luminosity functions with a high signal-to-noise ratio for the main-sequence stars, giants, and white dwarfs. We examined in detail the Hyades cluster, the white dwarf population, and wide-binary systems and produced candidate lists for all three samples. We detected local manifestations of several streams, superclusters, and halo objects, in which we identified 12 members of \G\ Enceladus. We present the first direct parallaxes of five objects in multiple systems within 10\,pc of the Sun. △ Less

Submitted 3 December, 2020; originally announced December 2020.

Comments: 45 Pages, 39 figures in main part and 18 in appendix, tables on CDS

Journal ref: A&A 649, A6 (2021)

Gaia Early Data Release 3: Acceleration of the solar system from Gaia astrometry

Authors: Gaia Collaboration, S. A. Klioner, F. Mignard, L. Lindegren, U. Bastian, P. J. McMillan, J. Hernández, D. Hobbs, M. Ramos-Lerate, M. Biermann, A. Bombrun, A. de Torres, E. Gerlach, R. Geyer, T. Hilger, U. Lammers, H. Steidelmüller, C. A. Stephenson, A. G. A. Brown, A. Vallenari, T. Prusti, J. H. J. de Bruijne, C. Babusiaux, O. L. Creevey, D. W. Evans , et al. (392 additional authors not shown)

Abstract: Context. Gaia Early Data Release 3 (Gaia EDR3) provides accurate astrometry for about 1.6 million compact (QSO-like) extragalactic sources, 1.2 million of which have the best-quality five-parameter astrometric solutions. Aims. The proper motions of QSO-like sources are used to reveal a systematic pattern due to the acceleration of the solar system barycentre with respect to the rest frame of the… ▽ More Context. Gaia Early Data Release 3 (Gaia EDR3) provides accurate astrometry for about 1.6 million compact (QSO-like) extragalactic sources, 1.2 million of which have the best-quality five-parameter astrometric solutions. Aims. The proper motions of QSO-like sources are used to reveal a systematic pattern due to the acceleration of the solar system barycentre with respect to the rest frame of the Universe. Apart from being an important scientific result by itself, the acceleration measured in this way is a good quality indicator of the Gaia astrometric solution. Methods. The effect of the acceleration is obtained as a part of the general expansion of the vector field of proper motions in Vector Spherical Harmonics (VSH). Various versions of the VSH fit and various subsets of the sources are tried and compared to get the most consistent result and a realistic estimate of its uncertainty. Additional tests with the Gaia astrometric solution are used to get a better idea on possible systematic errors in the estimate. Results. Our best estimate of the acceleration based on Gaia EDR3 is $(2.32 \pm 0.16) \times 10^{-10}$ m s${}^{-2}$ (or $7.33 \pm 0.51$ km s$^{-1}$ Myr${}^{-1}$) towards $α= 269.1^\circ \pm 5.4^\circ$, $δ= -31.6^\circ \pm 4.1^\circ$, corresponding to a proper motion amplitude of $5.05 \pm 0.35$ $μ$as yr${}^{-1}$. This is in good agreement with the acceleration expected from current models of the Galactic gravitational potential. We expect that future Gaia data releases will provide estimates of the acceleration with uncertainties substantially below 0.1 $μ$as yr${}^{-1}$. △ Less

Submitted 3 December, 2020; originally announced December 2020.

Comments: A&A, accepted

Journal ref: A&A 649, A9 (2021)

Gaia Early Data Release 3: Photometric content and validation

Authors: M. Riello, F. De Angeli, D. W. Evans, P. Montegriffo, J. M. Carrasco, G. Busso, L. Palaversa, P. W. Burgess, C. Diener, M. Davidson, N. Rowell, C. Fabricius, C. Jordi, M. Bellazzini, E. Pancino, D. L. Harrison, C. Cacciari, F. van Leeuwen, N. C. Hambly, S. T. Hodgkin, P. J. Osborne, G. Altavilla, M. A. Barstow, A. G. A. Brown, M. Castellani , et al. (17 additional authors not shown)

Abstract: Gaia Early Data Release 3 contains astrometry and photometry results for about 1.8 billion sources based on observations collected by the ESA Gaia satellite during the first 34 months of operations. This paper focuses on the photometric content, describing the input data, the algorithms, the processing, and the validation of the results. Particular attention is given to the quality of the data and… ▽ More Gaia Early Data Release 3 contains astrometry and photometry results for about 1.8 billion sources based on observations collected by the ESA Gaia satellite during the first 34 months of operations. This paper focuses on the photometric content, describing the input data, the algorithms, the processing, and the validation of the results. Particular attention is given to the quality of the data and to a number of features that users may need to take into account to make the best use of the EDR3 catalogue. The treatment of the BP and RP background has been updated to include a better estimation of the local background, and the detection of crowding effects has been used to exclude affected data from the calibrations. The photometric calibration models have also been updated to account for flux loss over the whole magnitude range. Significant improvements in the modelling and calibration of the point and line spread functions have also helped to reduce a number of instrumental effects that were still present in DR2. EDR3 contains 1.806 billion sources with G-band photometry and 1.540 billion sources with BP and RP photometry. The median uncertainty in the G-band photometry, as measured from the standard deviation of the internally calibrated mean photometry for a given source, is 0.2 mmag at magnitude G=10 to 14, 0.8 mmag at G=17, and 2.6 mmag at G=19. The significant magnitude term found in the Gaia DR2 photometry is no longer visible, and overall there are no trends larger than 1 mmag/mag. Using one passband over the whole colour and magnitude range leaves no systematics above the 1% level in magnitude in any of the bands, and a larger systematic is present for a very small sample of bright and blue sources. A detailed description of the residual systematic effects is provided. Overall the quality of the calibrated mean photometry in EDR3 is superior with respect to DR2 for all bands. △ Less

Submitted 3 December, 2020; originally announced December 2020.

Gaia Early Data Release 3: Structure and properties of the Magellanic Clouds

Authors: Gaia Collaboration, X. Luri, L. Chemin, G. Clementini, H. E. Delgado, P. J. McMillan, M. Romero-Gómez, E. Balbinot, A. Castro-Ginard, R. Mor, V. Ripepi, L. M. Sarro, M. -R. L. Cioni, C. Fabricius, A. Garofalo, A. Helmi, T. Muraveva, A. G. A. Brown, A. Vallenari, T. Prusti, J. H. J. de, C. Babusiaux, M. Biermann, O. L. Creevey, D. W. Evans , et al. (395 additional authors not shown)

Abstract: We compare the Gaia DR2 and Gaia EDR3 performances in the study of the Magellanic Clouds and show the clear improvements in precision and accuracy in the new release. We also show that the systematics still present in the data make the determination of the 3D geometry of the LMC a difficult endeavour; this is at the very limit of the usefulness of the Gaia EDR3 astrometry, but it may become feasib… ▽ More We compare the Gaia DR2 and Gaia EDR3 performances in the study of the Magellanic Clouds and show the clear improvements in precision and accuracy in the new release. We also show that the systematics still present in the data make the determination of the 3D geometry of the LMC a difficult endeavour; this is at the very limit of the usefulness of the Gaia EDR3 astrometry, but it may become feasible with the use of additional external data. We derive radial and tangential velocity maps and global profiles for the LMC for the several subsamples we defined. To our knowledge, this is the first time that the two planar components of the ordered and random motions are derived for multiple stellar evolutionary phases in a galactic disc outside the Milky Way, showing the differences between younger and older phases. We also analyse the spatial structure and motions in the central region, the bar, and the disc, providing new insights into features and kinematics. Finally, we show that the Gaia EDR3 data allows clearly resolving the Magellanic Bridge, and we trace the density and velocity flow of the stars from the SMC towards the LMC not only globally, but also separately for young and evolved populations. This allows us to confirm an evolved population in the Bridge that is slightly shift from the younger population. Additionally, we were able to study the outskirts of both Magellanic Clouds, in which we detected some well-known features and indications of new ones. △ Less

Submitted 4 January, 2021; v1 submitted 3 December, 2020; originally announced December 2020.

Comments: This paper is part of the "demonstration papers" released with Gaia EDR3: https://www.cosmos.esa.int/web/gaia/earlydr3

Journal ref: A&A 649, A7 (2021)

Gaia Early Data Release 3: Summary of the contents and survey properties

Authors: Gaia Collaboration, A. G. A Brown, A. Vallenari, T. Prusti, J. H. J. de Bruijne, C. Babusiaux, M. Biermann, O. L. Creevey, D. W. Evans, L. Eyer, A. Hutton, F. Jansen, C. Jordi, S. A. Klioner, U. Lammers, L. Lindegren, X. Luri, F. Mignard, C. Panem, D. Pourbaix, S. Randich, P. Sartoretti, C. Soubiran, N. A. Walton, F. Arenou , et al. (401 additional authors not shown)

Abstract: We present the early installment of the third Gaia data release, Gaia EDR3, consisting of astrometry and photometry for 1.8 billion sources brighter than magnitude 21, complemented with the list of radial velocities from Gaia DR2. Gaia EDR3 contains celestial positions and the apparent brightness in G for approximately 1.8 billion sources. For 1.5 billion of those sources, parallaxes, proper motio… ▽ More We present the early installment of the third Gaia data release, Gaia EDR3, consisting of astrometry and photometry for 1.8 billion sources brighter than magnitude 21, complemented with the list of radial velocities from Gaia DR2. Gaia EDR3 contains celestial positions and the apparent brightness in G for approximately 1.8 billion sources. For 1.5 billion of those sources, parallaxes, proper motions, and the (G_BP-G_RP) colour are also available. The passbands for G, G_BP, and G_RP are provided as part of the release. For ease of use, the 7 million radial velocities from Gaia DR2 are included in this release, after the removal of a small number of spurious values. New radial velocities will appear as part of Gaia DR3. Finally, Gaia EDR3 represents an updated materialisation of the celestial reference frame (CRF) in the optical, the Gaia-CRF3, which is based solely on extragalactic sources. The creation of the source list for Gaia EDR3 includes enhancements that make it more robust with respect to high proper motion stars, and the disturbing effects of spurious and partially resolved sources. The source list is largely the same as that for Gaia DR2, but it does feature new sources and there are some notable changes. The source list will not change for Gaia DR3. Gaia EDR3 represents a significant advance over Gaia DR2, with parallax precisions increased by 30 percent, proper motion precisions increased by a factor of 2, and the systematic errors in the astrometry suppressed by 30--40 percent for the parallaxes and by a factor ~2.5 for the proper motions. The photometry also features increased precision, but above all much better homogeneity across colour, magnitude, and celestial position. A single passband for G, G_BP, and G_RP is valid over the entire magnitude and colour range, with no systematics above the 1 percent level. △ Less

Submitted 9 June, 2021; v1 submitted 2 December, 2020; originally announced December 2020.

Comments: Accepted for A&A Special Issue on Gaia EDR3, 21 pages, 2 figures. This version includes the updates in the erratum (https://doi.org/10.1051/0004-6361/202039657e)

Journal ref: A&A 650, C3 (2021)

Flexo-Elastic Control Factors of Domain Morphology in Core-Shell Ferroelectric Nanoparticles: Soft and Rigid Shells

Authors: Eugene A. Eliseev, Anna N. Morozovska, Riccardo Hertel, Hanna V. Shevliakova, Yevhen M. Fomichov, Victor Yu. Reshetnyak, Dean R. Evans

Abstract: Within the framework of the Landau-Ginzburg-Devonshire approach we explore the impact of elastic anisotropy, electrostriction, flexoelectric couplings, and mismatch strain on the domain structure morphology in ferroelectric core-shell nanoparticles of spherical shape. We perform finite element modelling (FEM) for multiaxial ferroelectric nanoparticle cores covered with an elastically-isotropic sof… ▽ More Within the framework of the Landau-Ginzburg-Devonshire approach we explore the impact of elastic anisotropy, electrostriction, flexoelectric couplings, and mismatch strain on the domain structure morphology in ferroelectric core-shell nanoparticles of spherical shape. We perform finite element modelling (FEM) for multiaxial ferroelectric nanoparticle cores covered with an elastically-isotropic soft or elastically-anisotropic rigid paraelectric shell, with and without mismatch strains. In the case of a core covered with a soft shell, the FEM results show that at room temperature a single polarization vortex with a dipolar kernel can be stable if the electrostriction coupling is weak. With increasing anisotropic electrostriction coupling, the vortex disappears and is replaced by complex flux-closure structures, which are formed in the equatorial plane and transform into an elongated vortex with a central 180-degree domain wall near the core poles. This complex domain morphology develops in the core due to the anisotropic electrostriction, and the flexoelectric coupling leads to an additional curvature and twist of the polarization isosurfaces. In contrast to this, FEM performed for a core covered with a rigid shell shows that, at room temperature, the anisotropic elastic properties of the shell can stabilize vortex-like structures with three flux-closure domains, which gradually cross in the equatorial plane of the core and transform into 120-degree domains near the core poles. The flexoelectric coupling leads to a noticeable curling of the flux-closure domain walls. A mismatch strain compensates the curling of the flux-closure domains in the core confined by the elastically-anisotropic rigid shell. Our analysis reveals different types of topological defects, namely Bloch point structures (BPS) and Ising lines, that form in a ferroelectric core covered with a soft or rigid shell. △ Less

Submitted 1 December, 2020; originally announced December 2020.

Comments: 61 page,including 11 figures and SI with 9 figures

Journal ref: Acta Materialia 212, 116889 (2021)

Contact-free reversible switching of improper ferroelectric domains by electron and ion irradiation

Authors: Erik D. Roede, Aleksander B. Mosberg, Donald M. Evans, Edith Bourret, Zewu Yan, Antonius T. J. van Helvoort, Dennis Meier

Abstract: Focused ion beam (FIB) and scanning electron microscopy (SEM) are used to reversibly switch improper ferroelectric domains in the hexagonal manganite ErMnO$_3$. Surface charging is achieved by local ion (positive charging) and electron (positive and negative charging) irradiation, which allows controlled polarization switching without the need for electrical contacts. Polarization cycling reveals… ▽ More Focused ion beam (FIB) and scanning electron microscopy (SEM) are used to reversibly switch improper ferroelectric domains in the hexagonal manganite ErMnO$_3$. Surface charging is achieved by local ion (positive charging) and electron (positive and negative charging) irradiation, which allows controlled polarization switching without the need for electrical contacts. Polarization cycling reveals that the domain walls tend to return to the equilibrium configuration obtained in the as-grown state. The electric field response of sub-surface domains is studied by FIB cross-sectioning, revealing the 3D switching behavior. The results clarify how the polarization reversal in hexagonal manganites progresses at the level of domains, resolving both domain wall movements and the nucleation and growth of new domains. Our FIB-SEM based switching approach is applicable to all ferroelectrics where a sufficiently large electric field can be built up via surface charging, facilitating contact-free high-resolution studies of the domain and domain wall response to electric fields in 3D. △ Less

Submitted 1 December, 2020; originally announced December 2020.

Visualizing Temperature-Dependent Phase Stability in High Entropy Alloys

Authors: Daniel Evans, Jiadong Chen, Geoffroy Hautier, Wenhao Sun

Abstract: High Entropy Alloys (HEAs) contain near equimolar amounts of five or more elements and are a compelling space for materials design. Great emphasis is placed on identifying HEAs that form a homogeneous solid-solution, but the design of such HEAs is hindered by the difficulty of navigating stability relationships in high-component spaces. Traditional phase diagrams use barycentric coordinates to rep… ▽ More High Entropy Alloys (HEAs) contain near equimolar amounts of five or more elements and are a compelling space for materials design. Great emphasis is placed on identifying HEAs that form a homogeneous solid-solution, but the design of such HEAs is hindered by the difficulty of navigating stability relationships in high-component spaces. Traditional phase diagrams use barycentric coordinates to represent composition axes, which require D = (N - 1) spatial dimensions to represent an N-component system, meaning that HEA systems with N > 4 components cannot be readily visualized. Here, we propose forgoing barycentric composition axes in favor of two energy axes: a formation-energy axis and a 'reaction energy' axis. These Inverse Hull Webs offer an information-dense 2D representation that successfully capture complex phase stability relationships in N > 4 component systems. We use our new diagrams to visualize the transition of HEA solid-solutions from high-temperature stability to metastability upon quenching, and identify important thermodynamic features that are correlated with the persistence or decomposition of metastable HEAs. △ Less

Submitted 16 April, 2021; v1 submitted 29 November, 2020; originally announced November 2020.

Comments: 18 pages, 5 figures

Effect of pseudo-cubic (111)-oriented orthorhombic substrate facets on perovskite oxide thin film synthesis

Authors: K. Kjærnes, T. Bolstad, D. M. Evans, E. Lysne, B. A. D. Williamson, D. Meier, S. M. Selbach, T. Tybell

Abstract: Strain engineering with different substrate facets is promising for tuning functional properties of thin film perovskite oxides. By choice of facet, different surface symmetries and chemical bond directions for epitaxial interfaces can be tailored. Here, preparation of well-defined pseudo-cubic (111)-oriented orthorhombic substrates of DyScO3 , GdScO3 , and NdGaO3 is reported. The choice of orthor… ▽ More Strain engineering with different substrate facets is promising for tuning functional properties of thin film perovskite oxides. By choice of facet, different surface symmetries and chemical bond directions for epitaxial interfaces can be tailored. Here, preparation of well-defined pseudo-cubic (111)-oriented orthorhombic substrates of DyScO3 , GdScO3 , and NdGaO3 is reported. The choice of orthorhombic facet, (011)o or (101)o , both corresponding to pseudo-cubic (111)pc , gives vicinal surfaces with single or double (111pc layer terrace step heights, respectively, impacting subsequent thin film growth. Orthorhombic LaFeO3 epitaxy on the (101)o facet reveals a distinction between alternating (111)pc layers, both during and after growth. The observed differences are explained based on the oxygen octahedral tilt pattern relative to the two orthorhombic (111)pc surfaces. This robust structural detail in the orthorhombic perovskite oxides enables utilisation of different (111)pc facets for property engineering, through polyhedral connectivity control and cation coordination at epitaxial interfaces. △ Less

Submitted 25 November, 2020; originally announced November 2020.

Insulating improper ferroelectric domain walls as robust barrier layer capacitors

Authors: Lukas Puntigam, Jan Schultheiß, Ana Strinic, Zewu Yan, Edith Bourret, Markus Altthaler, Istvan Kezsmarki, Donald M. Evans, Dennis Meier, Stephan Krohns

Abstract: We report the dielectric properties of improper ferroelectric h-ErMnO$_3$. From the bulk characterisation we observe a temperature and frequency range with two distinct relaxation-like features, leading to high and even 'colossal' values for the dielectric permittivity. One feature trivially originates from the formation of a Schottky barrier at the electrode-sample interface, whereas the second o… ▽ More We report the dielectric properties of improper ferroelectric h-ErMnO$_3$. From the bulk characterisation we observe a temperature and frequency range with two distinct relaxation-like features, leading to high and even 'colossal' values for the dielectric permittivity. One feature trivially originates from the formation of a Schottky barrier at the electrode-sample interface, whereas the second one relates to an internal barrier layer capacitance (BLC). The calculated volume fraction of the internal BLC (of 8 %) is in good agreement with the observed volume fraction of insulating domain walls (DWs). While it is established that insulating DWs can give rise to high dielectric constants, studies typically focused on proper ferroelectrics where electric fields can remove the DWs. In h-ErMnO$_3$, by contrast, the insulating DWs are topologically protected, facilitating operation under substantially higher electric fields. Our findings provide the basis for a conceptually new approach to engineer materials exhibiting colossal dielectric permittivities using domain walls in improper ferroelecctrics with potential applications in electroceramic capacitors. △ Less

Submitted 20 November, 2020; originally announced November 2020.

Comments: 7 pages, 4 figures

Finding Friends and Flip** Frenemies: Automatic Paraphrase Dataset Augmentation Using Graph Theory

Authors: Hannah Chen, Yangfeng Ji, David Evans

Abstract: Most NLP datasets are manually labeled, so suffer from inconsistent labeling or limited size. We propose methods for automatically improving datasets by viewing them as graphs with expected semantic properties. We construct a paraphrase graph from the provided sentence pair labels, and create an augmented dataset by directly inferring labels from the original sentence pairs using a transitivity pr… ▽ More Most NLP datasets are manually labeled, so suffer from inconsistent labeling or limited size. We propose methods for automatically improving datasets by viewing them as graphs with expected semantic properties. We construct a paraphrase graph from the provided sentence pair labels, and create an augmented dataset by directly inferring labels from the original sentence pairs using a transitivity property. We use structural balance theory to identify likely mislabelings in the graph, and flip their labels. We evaluate our methods on paraphrase models trained using these datasets starting from a pretrained BERT model, and find that the automatically-enhanced training sets result in more accurate models. △ Less

Submitted 3 November, 2020; originally announced November 2020.

Comments: EMNLP 2020 (Findings)

Entropy-based test for generalized Gaussian distributions

Authors: Mehmet Siddik Cadirci, Dafydd Evans, Nikolai Leonenko, Vitalii Makogin

Abstract: In this paper, we provide the proof of $L^2$ consistency for the $k$th nearest neighbour distance estimator of the Shannon entropy for an arbitrary fixed $k\geq 1.$ We construct the non-parametric test of goodness-of-fit for a class of introduced generalized multivariate Gaussian distributions based on a maximum entropy principle. The theoretical results are followed by numerical studies on simula… ▽ More In this paper, we provide the proof of $L^2$ consistency for the $k$th nearest neighbour distance estimator of the Shannon entropy for an arbitrary fixed $k\geq 1.$ We construct the non-parametric test of goodness-of-fit for a class of introduced generalized multivariate Gaussian distributions based on a maximum entropy principle. The theoretical results are followed by numerical studies on simulated samples. △ Less

Submitted 13 October, 2020; originally announced October 2020.

MSC Class: 62G05; 28D20; 62G10;

Large-amplitude variables in Gaia Data Release 2. Multi-band variability characterization

Authors: N. Mowlavi, L. Rimoldini, D. W. Evans, M. Riello, F. De Angeli, L. Palaversa, M. Audard, L. Eyer, P. Garcia-Lario, P. Gavras, B. Holl, G. Jevardat de Fombelle, I. Lecœur-Taïbi, K. Nienartowicz

Abstract: The second data release (DR2) of Gaia provides mean photometry in three bands for $\sim$1.4 billion sources, but light curves and variability properties are available for only $\sim$0.5 million of them. Here, we provide a census of large-amplitude variables with amplitudes larger than $\sim$0.2 mag in the $G$ band for objects with mean brightnesses between 5.5 and 19 mag. To achieve this, we rely… ▽ More The second data release (DR2) of Gaia provides mean photometry in three bands for $\sim$1.4 billion sources, but light curves and variability properties are available for only $\sim$0.5 million of them. Here, we provide a census of large-amplitude variables with amplitudes larger than $\sim$0.2 mag in the $G$ band for objects with mean brightnesses between 5.5 and 19 mag. To achieve this, we rely on variability amplitude proxies in $G$, $G_{BP}$ and $G_{RP}$ computed from the uncertainties on the magnitudes published in DR2. We then apply successive filters to identify two subsets containing respectively sources with reliable mean $G_{BP}$ and $G_{RP}$ (for studies using colours) and sources having compatible amplitude proxies in $G$, $G_{BP}$ and $G_{RP}$ (for multi-band variability studies). The full catalogue gathers $23\,315\,874$ large-amplitude variable candidates, and the two subsets with increased levels of purity contain respectively $1\,148\,861$ and $618\,966$ sources. A multi-band variability analysis of the catalogue shows that different types of variable stars can be globally categorized in four groups according to their colour and blue-to-red amplitude ratios as determined from the $G$, $G_{BP}$ and $G_{RP}$ amplitude proxies. The catalogue constitutes the first census of Gaia large-amplitude variable candidates, extracted from the public DR2 archive. The overview presented here illustrates the added-value of the mission for multi-band variability studies even at this stage when epoch photometry is not yet available for all sources. (Abridged abstract) △ Less

Submitted 15 January, 2021; v1 submitted 16 September, 2020; originally announced September 2020.

Comments: Final version, A&A, in press. Main text: 20 pages, 26 figures. Four appendixes

Journal ref: A&A 648, A44 (2021)

A Lagrangian Klein bottle you can't squeeze

Authors: Jonathan David Evans

Abstract: Suppose you have a nonorientable Lagrangian surface L in a symplectic 4-manifold. How far can you deform the symplectic form before the smooth isotopy class of L contains no Lagrangians? I solve this question for a particular Lagrangian Klein bottle. I also discuss some related conjectures. Suppose you have a nonorientable Lagrangian surface L in a symplectic 4-manifold. How far can you deform the symplectic form before the smooth isotopy class of L contains no Lagrangians? I solve this question for a particular Lagrangian Klein bottle. I also discuss some related conjectures. △ Less

Submitted 8 June, 2022; v1 submitted 3 September, 2020; originally announced September 2020.

Comments: 10 pages, 4 figures. This version of the article has been accepted for publication, after peer review, but is not the Version of Record. The Version of Record is available online at: https://doi.org/10.1007/s11784-022-00945-w

MSC Class: 53D12; 53D35; 53D42

Journal ref: Journal of Fixed Point Theory and Applications volume 24, Article number: 47 (2022) in Symplectic geometry - A Festschrift in honour of Claude Viterbo's 60th birthday

MutaGAN: A Seq2seq GAN Framework to Predict Mutations of Evolving Protein Populations

Authors: Daniel S. Berman, Craig Howser, Thomas Mehoke, Jared D. Evans

Abstract: The ability to predict the evolution of a pathogen would significantly improve the ability to control, prevent, and treat disease. Despite significant progress in other problem spaces, deep learning has yet to contribute to the issue of predicting mutations of evolving populations. To address this gap, we developed a novel machine learning framework using generative adversarial networks (GANs) wit… ▽ More The ability to predict the evolution of a pathogen would significantly improve the ability to control, prevent, and treat disease. Despite significant progress in other problem spaces, deep learning has yet to contribute to the issue of predicting mutations of evolving populations. To address this gap, we developed a novel machine learning framework using generative adversarial networks (GANs) with recurrent neural networks (RNNs) to accurately predict genetic mutations and evolution of future biological populations. Using a generalized time-reversible phylogenetic model of protein evolution with bootstrapped maximum likelihood tree estimation, we trained a sequence-to-sequence generator within an adversarial framework, named MutaGAN, to generate complete protein sequences augmented with possible mutations of future virus populations. Influenza virus sequences were identified as an ideal test case for this deep learning framework because it is a significant human pathogen with new strains emerging annually and global surveillance efforts have generated a large amount of publicly available data from the National Center for Biotechnology Information's (NCBI) Influenza Virus Resource (IVR). MutaGAN generated "child" sequences from a given "parent" protein sequence with a median Levenshtein distance of 2.00 amino acids. Additionally, the generator was able to augment the majority of parent proteins with at least one mutation identified within the global influenza virus population. These results demonstrate the power of the MutaGAN framework to aid in pathogen forecasting with implications for broad utility in evolutionary prediction for any protein population. △ Less

Submitted 26 August, 2020; originally announced August 2020.

Comments: 28 pages, 9 figures, 2 tables, Daniel S. Berman and Craig Howser contributed equally to this work. This paper was submitted to Artificial Intelligence

A white dwarf bound to the transiting planetary system WASP-98

Authors: John Southworth, P. -E. Tremblay, B. T. Gaensicke, D. F. Evans, T. Mocnik

Abstract: WASP-98 is a planetary system containing a hot Jupiter transiting a late-G dwarf. A fainter star 12 arcsec distant has previously been identified as a white dwarf, with a distance and proper motion consistent with a physical association with the planetary system. We present spectroscopy of the white dwarf, with the aim of determining its mass, radius and temperature and hence the age of the system… ▽ More WASP-98 is a planetary system containing a hot Jupiter transiting a late-G dwarf. A fainter star 12 arcsec distant has previously been identified as a white dwarf, with a distance and proper motion consistent with a physical association with the planetary system. We present spectroscopy of the white dwarf, with the aim of determining its mass, radius and temperature and hence the age of the system. However, the spectra show the featureless continuum and lack of spectral lines characteristic of the DC class of white dwarfs. We therefore fitted theoretical white dwarf spectra to the ugriz apparent magnitudes and Gaia DR2 parallax of this object in order to determine its physical properties and the age of the system. We find that the system is old, with a lower limit of 3.6 Gyr, but theoretical uncertainties preclude a precise determination of its age. Its kinematics are consistent with membership of the thick disc, but do not allow us to rule out the thin-disc alternative. The old age and low metallicity of the system suggest it is subject to an age-metallicity relation, but analysis of the most metal-rich and metal-poor transiting planetary systems yields only insubstantial evidence of this. We conclude that the study of bound white dwarfs can yield independent ages to planetary systems, but such analysis may be better-suited to DA and DB rather than DC white dwarfs. △ Less

Submitted 6 August, 2020; originally announced August 2020.

Comments: Accepted for publication in MNRAS. 7 pages, 1 table, 5 colour figures

Model-Targeted Poisoning Attacks with Provable Convergence

Authors: Fnu Suya, Saeed Mahloujifar, Anshuman Suri, David Evans, Yuan Tian

Abstract: In a poisoning attack, an adversary with control over a small fraction of the training data attempts to select that data in a way that induces a corrupted model that misbehaves in favor of the adversary. We consider poisoning attacks against convex machine learning models and propose an efficient poisoning attack designed to induce a specified model. Unlike previous model-targeted poisoning attack… ▽ More In a poisoning attack, an adversary with control over a small fraction of the training data attempts to select that data in a way that induces a corrupted model that misbehaves in favor of the adversary. We consider poisoning attacks against convex machine learning models and propose an efficient poisoning attack designed to induce a specified model. Unlike previous model-targeted poisoning attacks, our attack comes with provable convergence to {\it any} attainable target classifier. The distance from the induced classifier to the target classifier is inversely proportional to the square root of the number of poisoning points. We also provide a lower bound on the minimum number of poisoning points needed to achieve a given target classifier. Our method uses online convex optimization, so finds poisoning points incrementally. This provides more flexibility than previous attacks which require a priori assumption about the number of poisoning points. Our attack is the first model-targeted poisoning attack that provides provable convergence for convex models, and in our experiments, it either exceeds or matches state-of-the-art attacks in terms of attack success rate and distance to the target model. △ Less

Submitted 21 April, 2021; v1 submitted 29 June, 2020; originally announced June 2020.

Comments: 32 pages, code available at: https://github.com/suyeecav/model-targeted-poisoning

Controlling local resistance via electric-field induced dislocations

Authors: D. M. Evans, D. R. Småbråten, T. S. Holstad, P. E. Vullum, A. B. Mosberg, Z. Yan, E. Bourret, A. T. J. Van Helvoort, S. M. Selbach, D. Meier

Abstract: Dislocations are one-dimensional (1D) topological line defects where the lattice deviates from the perfect crystal structure. The presence of dislocations transcends condensed matter research and gives rise to a diverse range of emergent phenomena [1-6], ranging from geological effects [7] to light emission from diodes [8]. Despite their ubiquity, to date, the controlled formation of dislocations… ▽ More Dislocations are one-dimensional (1D) topological line defects where the lattice deviates from the perfect crystal structure. The presence of dislocations transcends condensed matter research and gives rise to a diverse range of emergent phenomena [1-6], ranging from geological effects [7] to light emission from diodes [8]. Despite their ubiquity, to date, the controlled formation of dislocations is usually achieved via strain fields, applied either during growth [9,10] or retrospectively via deformation, e.g., (nano [11-14])-indentation [15]. Here we show how partial dislocations can be induced using local electric fields, altering the structure and electronic response of the material where the field is applied. By combining high-resolution imaging techniques and density functional theory calculations, we directly image these dislocations in the ferroelectric hexagonal manganite Er(Ti,Mn)O3 and study their impact on the local electric transport behaviour. The use of an electric field to induce partial dislocations is a conceptually new approach to the burgeoning field of emergent defect-driven phenomena and enables local property control without the need of external macroscopic strain fields. This control is an important step towards integrating and functionalising dislocations in practical devices for future oxide electronics. △ Less

Submitted 26 June, 2020; originally announced June 2020.

Comments: 20 pages, 7 figures

Pointwise Paraphrase Appraisal is Potentially Problematic

Authors: Hannah Chen, Yangfeng Ji, David Evans

Abstract: The prevailing approach for training and evaluating paraphrase identification models is constructed as a binary classification problem: the model is given a pair of sentences, and is judged by how accurately it classifies pairs as either paraphrases or non-paraphrases. This pointwise-based evaluation method does not match well the objective of most real world applications, so the goal of our work… ▽ More The prevailing approach for training and evaluating paraphrase identification models is constructed as a binary classification problem: the model is given a pair of sentences, and is judged by how accurately it classifies pairs as either paraphrases or non-paraphrases. This pointwise-based evaluation method does not match well the objective of most real world applications, so the goal of our work is to understand how models which perform well under pointwise evaluation may fail in practice and find better methods for evaluating paraphrase identification models. As a first step towards that goal, we show that although the standard way of fine-tuning BERT for paraphrase identification by pairing two sentences as one sequence results in a model with state-of-the-art performance, that model may perform poorly on simple tasks like identifying pairs with two identical sentences. Moreover, we show that these models may even predict a pair of randomly-selected sentences with higher paraphrase score than a pair of identical ones. △ Less

Submitted 4 June, 2020; v1 submitted 25 May, 2020; originally announced May 2020.

Comments: ACL 2020 Student Research Workshop

Journal ref: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop (2020) 150-155

Revisiting Membership Inference Under Realistic Assumptions

Authors: Bargav Jayaraman, Lingxiao Wang, Katherine Knipmeyer, Quanquan Gu, David Evans

Abstract: We study membership inference in settings where some of the assumptions typically used in previous research are relaxed. First, we consider skewed priors, to cover cases such as when only a small fraction of the candidate pool targeted by the adversary are actually members and develop a PPV-based metric suitable for this setting. This setting is more realistic than the balanced prior setting typic… ▽ More We study membership inference in settings where some of the assumptions typically used in previous research are relaxed. First, we consider skewed priors, to cover cases such as when only a small fraction of the candidate pool targeted by the adversary are actually members and develop a PPV-based metric suitable for this setting. This setting is more realistic than the balanced prior setting typically considered by researchers. Second, we consider adversaries that select inference thresholds according to their attack goals and develop a threshold selection procedure that improves inference attacks. Since previous inference attacks fail in imbalanced prior setting, we develop a new inference attack based on the intuition that inputs corresponding to training set members will be near a local minimum in the loss function, and show that an attack that combines this with thresholds on the per-instance loss can achieve high PPV even in settings where other attacks appear to be ineffective. Code for our experiments can be found here: https://github.com/bargavj/EvaluatingDPML. △ Less

Submitted 13 January, 2021; v1 submitted 21 May, 2020; originally announced May 2020.

Certifying Joint Adversarial Robustness for Model Ensembles

Authors: Mainuddin Ahmad Jonas, David Evans

Abstract: Deep Neural Networks (DNNs) are often vulnerable to adversarial examples.Several proposed defenses deploy an ensemble of models with the hope that, although the individual models may be vulnerable, an adversary will not be able to find an adversarial example that succeeds against the ensemble. Depending on how the ensemble is used, an attacker may need to find a single adversarial example that suc… ▽ More Deep Neural Networks (DNNs) are often vulnerable to adversarial examples.Several proposed defenses deploy an ensemble of models with the hope that, although the individual models may be vulnerable, an adversary will not be able to find an adversarial example that succeeds against the ensemble. Depending on how the ensemble is used, an attacker may need to find a single adversarial example that succeeds against all, or a majority, of the models in the ensemble. The effectiveness of ensemble defenses against strong adversaries depends on the vulnerability spaces of models in the ensemble being disjoint. We consider the joint vulnerability of an ensemble of models, and propose a novel technique for certifying the joint robustness of ensembles, building upon prior works on single-model robustness certification. We evaluate the robustness of various models ensembles, including models trained using cost-sensitive robustness to be diverse, to improve understanding of the potential effectiveness of ensemble models as a defense against adversarial examples. △ Less

Submitted 21 April, 2020; originally announced April 2020.

Comments: Open source code for our implementation and for reproducing our experiments is available at https://github.com/jonas-maj/ensemble-adversarial-robustness

OGLE-2017-BLG-0406: ${\it Spitzer}$ Microlens Parallax Reveals Saturn-mass Planet orbiting M-dwarf Host in the Inner Galactic Disk

Authors: Yuki Hirao, David P. Bennett, Yoon-Hyun Ryu, Naoki Koshimoto, Andrzej Udalski, Jennifer C. Yee, Takahiro Sumi, Ian A. Bond, Yossi Shvartzvald, Fumio Abe, Richard K. Barry, Aparna Bhattacharya, Martin Donachie, Akihiko Fukui, Yoshitaka Itow, Iona Kondo, Man Cheung Alex Li, Yutaka Matsubara, Taro Matsuo, Shota Miyazaki, Yasushi Muraki, Masayuki Nagakane, Clement Ranc, Nicholas J. Rattenbury, Haruno Suematsu , et al. (71 additional authors not shown)

Abstract: We report the discovery and analysis of the planetary microlensing event OGLE-2017-BLG-0406, which was observed both from the ground and by the ${\it Spitzer}$ satellite in a solar orbit. At high magnification, the anomaly in the light curve was densely observed by ground-based-survey and follow-up groups, and it was found to be explained by a planetary lens with a planet/host mass ratio of… ▽ More We report the discovery and analysis of the planetary microlensing event OGLE-2017-BLG-0406, which was observed both from the ground and by the ${\it Spitzer}$ satellite in a solar orbit. At high magnification, the anomaly in the light curve was densely observed by ground-based-survey and follow-up groups, and it was found to be explained by a planetary lens with a planet/host mass ratio of $q=7.0 \times 10^{-4}$ from the light-curve modeling. The ground-only and ${\it Spitzer}$-"only" data each provide very strong one-dimensional (1-D) constraints on the 2-D microlens parallax vector $\bf{π_{\rm E}}$. When combined, these yield a precise measurement of $\bf{π_{\rm E}}$, and so of the masses of the host $M_{\rm host}=0.56\pm0.07\,M_\odot$ and planet $M_{\rm planet} = 0.41 \pm 0.05\,M_{\rm Jup}$. The system lies at a distance $D_{\rm L}=5.2 \pm 0.5 \ {\rm kpc}$ from the Sun toward the Galactic bulge, and the host is more likely to be a disk population star according to the kinematics of the lens. The projected separation of the planet from the host is $a_{\perp} = 3.5 \pm 0.3 \ {\rm au}$, i.e., just over twice the snow line. The Galactic-disk kinematics are established in part from a precise measurement of the source proper motion based on OGLE-IV data. By contrast, the ${\it Gaia}$ proper-motion measurement of the source suffers from a catastrophic $10\,σ$ error. △ Less

Submitted 6 June, 2020; v1 submitted 20 April, 2020; originally announced April 2020.

Comments: 40 pages, 12 figures, 10 tables, accepted for publication in The Astronomical Journal

Electric Field Control of Three-Dimensional Vortex States in Core-Shell Ferroelectric Nanoparticles

Authors: Anna N. Morozovska, Eugene A. Eliseev, Riccardo Hertel, Yevhen M. Fomichov, Viktoriia Tulaidan, Victor Yu. Reshetnyak, Dean R. Evans

Abstract: In the framework of the Landau-Ginzburg-Devonshire approach coupled with electrostatic equations, we performed finite element modeling of the polarization behavior in a ferroelectric barium titanate core covered with a tunable paraelectric strontium titanate shell placed in a polymer or liquid medium. A stable two-dimensional vortex is formed in the core after a zero-field relaxation of an initial… ▽ More In the framework of the Landau-Ginzburg-Devonshire approach coupled with electrostatic equations, we performed finite element modeling of the polarization behavior in a ferroelectric barium titanate core covered with a tunable paraelectric strontium titanate shell placed in a polymer or liquid medium. A stable two-dimensional vortex is formed in the core after a zero-field relaxation of an initial random or poly-domain distribution of the polarization, where the vortex axis is directed along one of the core crystallographic axes. Subsequently, sinusoidal pulses of a homogeneous electric field with variable period, strength, and direction are applied. The field-induced changes of the vortex structure consist in the appearance of an axial kernel in the form of a prolate nanodomain, the kernel growth, an increasing orientation of the polarization along the field, and the onset of a single-domain state. We introduced the term "kernel" to name the prolate nanodomain developed near the vortex axis and polarized perpendicular to the vortex plane. In ferromagnetism, this region is generally known as the vortex core. The vortex states with a kernel possess a manifold degeneracy, appearing from three equiprobable directions of vortex axis, clockwise and counterclockwise directions of polarization rotation along the vortex axis, and two polarization directions in the kernel. This multitude of the vortex states in a single core are promising for applications of core-shell nanoparticles and their ensembles as multi-bit memory and related logic units. The rotation of a vortex kernel over a sphere, possible for the core-shell nanoparticles in a soft matter medium with controllable viscosity, may be used to imitate qubit features. △ Less

Submitted 12 March, 2020; originally announced April 2020.

Comments: 54 pages, 12 figures, 4 appendices

Journal ref: Acta Mater. 200, 256-273 (2020)

Tambara-Yamagami, loop groups, bundles and KK-theory

Authors: David E Evans, Terry Gannon

Abstract: This paper is part of a sequence interpreting quantities of conformal field theories K-theoretically. Here we give geometric constructions of the associated module categories (modular invariants, nimreps, etc). In particular, we give a KK-theory interpretation of all modular invariants for the loop groups of tori, as well as most known modular invariants of loop groups. In addition, we find unexpe… ▽ More This paper is part of a sequence interpreting quantities of conformal field theories K-theoretically. Here we give geometric constructions of the associated module categories (modular invariants, nimreps, etc). In particular, we give a KK-theory interpretation of all modular invariants for the loop groups of tori, as well as most known modular invariants of loop groups. In addition, we find unexpectedly that the Tambara-Yamagami fusion category has an elegant description as bundles over a groupoid, and use that to interpret its module categories as KK-elements. We establish reconstruction for the doubles of all Tambara-Yamagami categories, generalizing work of Bischoff to even-order groups. We conclude by relating the modular group representations coming from finite groups and loop groups to the Chern character and to the Fourier-Mukai transform △ Less

Submitted 21 March, 2020; originally announced March 2020.

Comments: 60 pages

One Neuron to Fool Them All

Authors: Anshuman Suri, David Evans

Abstract: Despite vast research in adversarial examples, the root causes of model susceptibility are not well understood. Instead of looking at attack-specific robustness, we propose a notion that evaluates the sensitivity of individual neurons in terms of how robust the model's output is to direct perturbations of that neuron's output. Analyzing models from this perspective reveals distinctive characterist… ▽ More Despite vast research in adversarial examples, the root causes of model susceptibility are not well understood. Instead of looking at attack-specific robustness, we propose a notion that evaluates the sensitivity of individual neurons in terms of how robust the model's output is to direct perturbations of that neuron's output. Analyzing models from this perspective reveals distinctive characteristics of standard as well as adversarially-trained robust models, and leads to several curious results. In our experiments on CIFAR-10 and ImageNet, we find that attacks using a loss function that targets just a single sensitive neuron find adversarial examples nearly as effectively as ones that target the full model. We analyze the properties of these sensitive neurons to propose a regularization term that can help a model achieve robustness to a variety of different perturbation constraints while maintaining accuracy on natural data distributions. Code for all our experiments is available at https://github.com/iamgroot42/sauron . △ Less

Submitted 9 June, 2020; v1 submitted 20 March, 2020; originally announced March 2020.

Comments: Updated 'PGD' columns of Table 1: numbers reported earlier for this column were (100 - accuracy) instead of attack success rates. Observations and conclusions remain unchanged

Understanding the Intrinsic Robustness of Image Distributions using Conditional Generative Models

Authors: Xiao Zhang, **ghui Chen, Quanquan Gu, David Evans

Abstract: Starting with Gilmer et al. (2018), several works have demonstrated the inevitability of adversarial examples based on different assumptions about the underlying input probability space. It remains unclear, however, whether these results apply to natural image distributions. In this work, we assume the underlying data distribution is captured by some conditional generative model, and prove intrins… ▽ More Starting with Gilmer et al. (2018), several works have demonstrated the inevitability of adversarial examples based on different assumptions about the underlying input probability space. It remains unclear, however, whether these results apply to natural image distributions. In this work, we assume the underlying data distribution is captured by some conditional generative model, and prove intrinsic robustness bounds for a general class of classifiers, which solves an open problem in Fawzi et al. (2018). Building upon the state-of-the-art conditional generative models, we study the intrinsic robustness of two common image benchmarks under $\ell_2$ perturbations, and show the existence of a large gap between the robustness limits implied by our theory and the adversarial robustness achieved by current state-of-the-art robust models. Code for all our experiments is available at https://github.com/xiaozhanguva/Intrinsic-Rob. △ Less

Submitted 29 February, 2020; originally announced March 2020.

Comments: 14 pages, 2 figures, 5 tables, AISTATS final paper reformatted for readability

Learning Adversarially Robust Representations via Worst-Case Mutual Information Maximization

Authors: Sicheng Zhu, Xiao Zhang, David Evans

Abstract: Training machine learning models that are robust against adversarial inputs poses seemingly insurmountable challenges. To better understand adversarial robustness, we consider the underlying problem of learning robust representations. We develop a notion of representation vulnerability that captures the maximum change of mutual information between the input and output distributions, under the wors… ▽ More Training machine learning models that are robust against adversarial inputs poses seemingly insurmountable challenges. To better understand adversarial robustness, we consider the underlying problem of learning robust representations. We develop a notion of representation vulnerability that captures the maximum change of mutual information between the input and output distributions, under the worst-case input perturbation. Then, we prove a theorem that establishes a lower bound on the minimum adversarial risk that can be achieved for any downstream classifier based on its representation vulnerability. We propose an unsupervised learning method for obtaining intrinsically robust representations by maximizing the worst-case mutual information between the input and output distributions. Experiments on downstream classification tasks support the robustness of the representations found using unsupervised learning with our training principle. △ Less

Submitted 5 July, 2020; v1 submitted 26 February, 2020; originally announced February 2020.

Comments: ICML 2020

Electromagnetic counterparts to gravitational wave events from Gaia

Authors: Z. Kostrzewa-Rutkowska, P. G. Jonker, S. T. Hodgkin, D. Eappachen, D. L. Harrison, S. E. Koposov, G. Rixon, L. Wyrzykowski, A. Yoldas, E. Breedt, A. Delgado, M. van Leeuwen, T. Wevers, P. W. Burgess, F. De Angeli, D. W. Evans, P. J. Osborne, M. Riello

Abstract: The recent discoveries of gravitational wave events and in one case also its electromagnetic (EM) counterpart allow us to study the Universe in a novel way. The increased sensitivity of the LIGO and Virgo detectors has opened the possibility for regular detections of EM transient events from mergers of stellar remnants. Gravitational wave sources are expected to have sky localisation up to a few h… ▽ More The recent discoveries of gravitational wave events and in one case also its electromagnetic (EM) counterpart allow us to study the Universe in a novel way. The increased sensitivity of the LIGO and Virgo detectors has opened the possibility for regular detections of EM transient events from mergers of stellar remnants. Gravitational wave sources are expected to have sky localisation up to a few hundred square degrees, thus Gaia as an all-sky multi-epoch photometric survey has the potential to be a good tool to search for the EM counterparts. In this paper we study the possibility of detecting EM counterparts to gravitational wave sources using the Gaia Science Alerts system. We develop an extension to current used algorithms to find transients and test its capabilities in discovering candidate transients on a sample of events from the observation periods O1 and O2 of LIGO and Virgo. For the gravitational wave events from the current run O3 we expect that about 16 (25) per cent should fall in sky regions observed by Gaia 7 (10) days after gravitational wave. The new algorithm will provide about 10 candidates per day from the whole sky. △ Less

Submitted 23 April, 2020; v1 submitted 12 February, 2020; originally announced February 2020.

Comments: MNRAS accepted, revised version with corrections from the erratum

A multiplicity study of transiting exoplanet host stars. I. High-contrast imaging with VLT/SPHERE

Authors: A. J. Bohn, J. Southworth, C. Ginski, M. A. Kenworthy, P. F. L. Maxted, D. F. Evans

Abstract: We study the multiplicity of host stars to known transiting extra-solar planets to test competing theories on the formation mechanisms of hot Jupiters. We observed 45 exoplanet host stars using VLT/SPHERE/IRDIS to search for potential companions. For each identified candidate companion we determined the probability that it is gravitationally bound to its host by performing common proper motion che… ▽ More We study the multiplicity of host stars to known transiting extra-solar planets to test competing theories on the formation mechanisms of hot Jupiters. We observed 45 exoplanet host stars using VLT/SPHERE/IRDIS to search for potential companions. For each identified candidate companion we determined the probability that it is gravitationally bound to its host by performing common proper motion checks and modelling of synthetic stellar populations around the host. We detected new candidate companions around K2-38, WASP-72, WASP-80, WASP-87, WASP-88, WASP-108, WASP-118, WASP-120, WASP-122, WASP123, WASP-130, WASP-131 and WASP-137. The closest candidates were detected at separations of $0.124''\pm0.007''$ and $0.189''\pm0.003''$ around WASP-108 and WASP-131; the measured $K$ band contrasts indicate that these are stellar companions of $0.35\pm0.02\,M_{\odot}$ and $0.62^{+0.05}_{-0.04}\,M_{\odot}$, respectively. Including the re-detection and confirmation of previously known companions in 13 other systems we derived a multiplicity fraction of $55.4^{+5.9}_{-9.4}\,\%$. For the representative sub-sample of 40 hot Jupiter host stars among our targets, the derived multiplicity rate is $54.8^{+6.3}_{-9.9}\,\%$. Our data do not confirm any trend that systems with eccentric planetary companions are preferably part of multiple systems. On average, we reached a magnitude contrast of $8.5\pm0.9$ mag at an angular separation of 0.5''. This allows to exclude additional stellar companions with masses larger than $0.08$ M$_\odot$ for almost all observed systems; around the closest and youngest systems this sensitivity is achieved at physical separations as small as 10 au. The presented study shows that SPHERE is an ideal instrument to detect and characterize close companions to exoplanetary host stars. △ Less

Submitted 22 January, 2020; originally announced January 2020.

Comments: Accepted for publication in A&A (22 pages, 14 figures, 4 tables)

Journal ref: A&A 635, A73 (2020)

OGLE-2013-BLG-0911Lb: A Secondary on the Brown-Dwarf Planet Boundary around an M-dwarf

Authors: Shota Miyazaki, Takahiro Sumi, David P. Bennett, Andrzej Udalski, Yossi Shvartzvald, Rachel Street, Valerio Bozza, Jennifer C. Yee, Ian A. Bond, Nicholas Rattenbury, Naoki Koshimoto, Daisuke Suzuki, Akihiko Fukui, F. Abe, A. Bhattacharya, R. Barry, M. Donachie, H. Fujii, Y. Hirao, Y. Itow, Y. Kamei, I. Kondo, M. C. A. Li, C. H. Ling, Y. Matsubara , et al. (71 additional authors not shown)

Abstract: We present the analysis of the binary-lens microlensing event OGLE-2013-BLG-0911. The best-fit solutions indicate the binary mass ratio of q~0.03 which differs from that reported in Shvartzvald+2016. The event suffers from the well-known close/wide degeneracy, resulting in two groups of solutions for the projected separation normalized by the Einstein radius of s~0.15 or s~7. The finite source and… ▽ More We present the analysis of the binary-lens microlensing event OGLE-2013-BLG-0911. The best-fit solutions indicate the binary mass ratio of q~0.03 which differs from that reported in Shvartzvald+2016. The event suffers from the well-known close/wide degeneracy, resulting in two groups of solutions for the projected separation normalized by the Einstein radius of s~0.15 or s~7. The finite source and the parallax observations allow us to measure the lens physical parameters. The lens system is an M-dwarf orbited by a massive Jupiter companion at very close (M_{host}=0.30^{+0.08}_{-0.06} M_{Sun}, M_{comp}=10.1^{+2.9}_{-2.2} M_{Jup}, a_{exp}=0.40^{+0.05}_{-0.04} au) or wide (M_{host}=0.28^{+0.10}_{-0.08} M_{Sun}, M_{comp}=9.9^{+3.8}_{-3.5}M_{Jup}, a_{exp}=18.0^{+3.2}_{-3.2} au) separation. Although the mass ratio is slightly above the planet-brown dwarf (BD) mass-ratio boundary of q=0.03 which is generally used, the median physical mass of the companion is slightly below the planet-BD mass boundary of 13M_{Jup}. It is likely that the formation mechanisms for BDs and planets are different and the objects near the boundaries could have been formed by either mechanism. It is important to probe the distribution of such companions with masses of ~13M_{Jup} in order to statistically constrain the formation theories for both BDs and massive planets. In particular, the microlensing method is able to probe the distribution around low-mass M-dwarfs and even BDs which is challenging for other exoplanet detection methods. △ Less

Submitted 19 December, 2019; originally announced December 2019.

Comments: 22 pages, 10 figures, Accepted for publication in The Astronomical Journal

Gaia's revolution in stellar variability

Authors: L. Eyer, L. Rimoldini, L. Rohrbasser, B. Holl, M. Audard, D. W. Evans, P. Garcia-Lario, P. Gavras, G. Clementini, S. T. Hodgkin, G. Jevardat de Fombelle, A. C. Lanzafame, T. Lebzelter, I. Lecoeur-Taibi, N. Mowlavi, K. Nienartowicz, V. Ripepi, L. Wyrzykowski

Abstract: Stellar variability studies are now reaching a completely new level thanks to ESA's Gaia mission, which enables us to locate many variable stars in the Hertzsprung-Russell diagram and determine the various instability strips/bands. Furthermore, this mission also allows us to detect, characterise and classify many millions of new variable stars thanks to its very unique nearly simultaneous multi-ep… ▽ More Stellar variability studies are now reaching a completely new level thanks to ESA's Gaia mission, which enables us to locate many variable stars in the Hertzsprung-Russell diagram and determine the various instability strips/bands. Furthermore, this mission also allows us to detect, characterise and classify many millions of new variable stars thanks to its very unique nearly simultaneous multi-epoch survey with different instruments (photometer, spectro-photometer, radial velocity spectrometer). An overview of what can be found in literature in terms of mostly data products by the Gaia consortium is given. This concerns the various catalogues of variable stars derived from the Gaia time series and also the location and motion of variable stars in the observational Hertzsprung-Russell diagram. In addition, we provide a list of a few thousands of variable white dwarf candidates derived from the DR2 published data, among them probably many hundreds of new pulsating white dwarfs. On a very different topic, we also show how Gaia allows us to reveal the 3D structures of and around the Milky Way thanks to the RR Lyrae stars. △ Less

Submitted 16 December, 2019; originally announced December 2019.

Comments: 7 pages, 4 figures, invited contribution at the conference "Stars and their variability observed from space - Celebrating the 5th anniversary of BRITE-Constellation", Vienna, Austria, August 19 - 23, 2019. Eds: C. Neiner, W. Weiss, D. Baade, E. Griffin, C. Lovekin, A. Moffat

Advances and Open Problems in Federated Learning

Authors: Peter Kairouz, H. Brendan McMahan, Brendan Avent, Aurélien Bellet, Mehdi Bennis, Arjun Nitin Bhagoji, Kallista Bonawitz, Zachary Charles, Graham Cormode, Rachel Cummings, Rafael G. L. D'Oliveira, Hubert Eichner, Salim El Rouayheb, David Evans, Josh Gardner, Zachary Garrett, Adrià Gascón, Badih Ghazi, Phillip B. Gibbons, Marco Gruteser, Zaid Harchaoui, Chaoyang He, Lie He, Zhouyuan Huo, Ben Hutchinson , et al. (34 additional authors not shown)

Abstract: Federated learning (FL) is a machine learning setting where many clients (e.g. mobile devices or whole organizations) collaboratively train a model under the orchestration of a central server (e.g. service provider), while kee** the training data decentralized. FL embodies the principles of focused data collection and minimization, and can mitigate many of the systemic privacy risks and costs re… ▽ More Federated learning (FL) is a machine learning setting where many clients (e.g. mobile devices or whole organizations) collaboratively train a model under the orchestration of a central server (e.g. service provider), while kee** the training data decentralized. FL embodies the principles of focused data collection and minimization, and can mitigate many of the systemic privacy risks and costs resulting from traditional, centralized machine learning and data science approaches. Motivated by the explosive growth in FL research, this paper discusses recent advances and presents an extensive collection of open problems and challenges. △ Less

Submitted 8 March, 2021; v1 submitted 10 December, 2019; originally announced December 2019.

Comments: Published in Foundations and Trends in Machine Learning Vol 4 Issue 1. See: https://www.nowpublishers.com/article/Details/MAL-083

The International Virtual Observatory Alliance in 2019

Authors: Mark A. Allen, Patrick Dowler, Janet D. Evans, Chenzhou Cui, Tim Jenness, Bruno Merin, G. Bruce Berriman, J. J. Kavelaars

Abstract: The International Virtual Observatory Alliance (IVOA) held its bi-annual Interoperability Meetings in May 2019, and in October 2019 following the ADASS XXIX conference. We provide a brief report on the status of the IVOA and the activities of the Interoperability Meetings. The International Virtual Observatory Alliance (IVOA) held its bi-annual Interoperability Meetings in May 2019, and in October 2019 following the ADASS XXIX conference. We provide a brief report on the status of the IVOA and the activities of the Interoperability Meetings. △ Less

Submitted 3 December, 2019; originally announced December 2019.

Comments: 4 pages, submitted to proceedings of ADASS XXIV to be published in ASP Conf. Series

GravityCam: Wide-field Imaging Surveys in the Visible from the Ground

Authors: C. Mackay, M. Dominik, I. A. Steele, C. Snodgrass, U. G. Jørgensen, J. Skottfelt, K. Stefanov, B. Carry, F. Braga-Ribas, A. Doressoundiram, V. D. Ivanov, P. Gandhi, D. F. Evans, M. Hundertmark, S. Serjeant, S. Ortolani

Abstract: GravityCam is a new concept of ground-based imaging instrument capable of delivering significantly sharper images from the ground than is normally possible without adaptive optics. Advances in optical and near infrared imaging technologies allow images to be acquired at high speed without significant noise penalty. Aligning these images before they are combined can yield a 3-5 fold improvement in… ▽ More GravityCam is a new concept of ground-based imaging instrument capable of delivering significantly sharper images from the ground than is normally possible without adaptive optics. Advances in optical and near infrared imaging technologies allow images to be acquired at high speed without significant noise penalty. Aligning these images before they are combined can yield a 3-5 fold improvement in image resolution. By using arrays of such detectors, survey fields may be as wide as the telescope optics allows. We describe the instrument and detail its application to accelerate greatly the rate of detection of Earth size planets by gravitational microlensing. GravityCam will improve substantially the quality of weak shear studies of dark matter distribution in distant clusters of galaxies. An extensive microlensing survey will also provide a vast dataset for asteroseismology studies, and GravityCam promises to generate a unique data set on the population of the Kuiper belt and possibly the Oort cloud. △ Less

Submitted 18 November, 2019; originally announced November 2019.

Comments: 13 pages, 9 figures. Paper presented at the Scientific Detector Workshop, Baltimore, 2017