Showing 1–50 of 67 results for author: Hron, J

An impressionist view of V Hydrae. When MATISSE paints Asymmetric Giant Blobs

Authors: L. Planquart, C. Paladini, A. Jorissen, A. Escorza, E. Pantin, J. Drevon, B. Aringer, F. Baron, A. Chiavassa, P. Cruzalèbes, W. Danchi, E. De Beck, M. A. T. Groenewegen, S. Höfner, J. Hron, T. Khouri, B. Lopez, F. Lykou, M. Montarges, N. Nardetto, K. Ohnaka, H. Olofsson, G. Rau, A. Rosales-Guzmán, J. Sanchez-Bermudez , et al. (7 additional authors not shown)

Abstract: Our purpose is to study the effect of binary companions located within the first 10 stellar radii from the primary AGB star. In this work, we target the mass-losing carbon star V Hydrae (V Hya), looking for signatures of its companion in the dust forming region of the atmosphere. The star was observed in the L- and N-bands with the VLTI/MATISSE instrument at low spectral resolution. We reconstruct… ▽ More Our purpose is to study the effect of binary companions located within the first 10 stellar radii from the primary AGB star. In this work, we target the mass-losing carbon star V Hydrae (V Hya), looking for signatures of its companion in the dust forming region of the atmosphere. The star was observed in the L- and N-bands with the VLTI/MATISSE instrument at low spectral resolution. We reconstructed images of V Hya's photosphere and surroundings using the two bands and compared our interferometric observables with VLTI/MIDI and VISIR archival data. To constrain the dust properties, we used DUSTY to model the spectral energy distribution. The star is dominated by dust emission in the L- and N- bands. The VISIR image confirms the presence of a large-scale dusty circumstellar envelope surrounding V Hya. The MATISSE reconstructed images show asymmetric and elongated structures in both infrared bands. In the L-band, we detected an elongated shape of approximately 15 mas, likely to be of photospheric origin. In the N-band, we found a 20 mas extension North-East from the star, and perpendicular to the L-band elongated axis. The position angle and the size of the N-band extension match the prediction of the companion position at MATISSE epoch. By comparing MATISSE N-band with MIDI data, we deduce that the elongation axis in the N-band has rotated since the previous interferometric measurements 13 years ago, supporting the idea that the particle enhancement is related to the dusty clump moving along with the companion. The MATISSE images unveil the presence of a dust enhancement at the companion position, opening new doors for further analysis on the binary interaction with an AGB component. △ Less

Submitted 13 May, 2024; originally announced May 2024.

Comments: Accepted for publication in Astronomy & Astrophysics

Determination of Navier's slip parameter using data assimilation

Authors: Alena Jarolímová, Jaroslav Hron

Abstract: One of the crucial aspects of patient-specific blood flow simulations is to specify material parameters and boundary conditions. The choice of boundary conditions can have a substantial impact on the character of the flow. While no-slip is the most popular wall boundary condition, some amount of slip, which determines how much fluid is allowed to flow along the wall, might be beneficial for better… ▽ More One of the crucial aspects of patient-specific blood flow simulations is to specify material parameters and boundary conditions. The choice of boundary conditions can have a substantial impact on the character of the flow. While no-slip is the most popular wall boundary condition, some amount of slip, which determines how much fluid is allowed to flow along the wall, might be beneficial for better agreement with flow patterns in medical images. However, even if one assumes the simple Navier's boundary conditions on the wall, in which the relationship between tangential components of the normal traction and the velocity is linear, the determination of the specific value of the slip parameter is often difficult. In this work, we present and test an optimal control method to estimate Navier's slip parameter on the wall and the velocity profile at the inlet using artificially generated flow domain and flow data. The results show that it is possible to recover the flow patterns and Navier's slip parameter by using sufficiently accurate discretization even from data containing a substantial amount of noise. △ Less

Submitted 7 February, 2024; originally announced February 2024.

Images of Betelgeuse with VLTI/MATISSE across the Great Dimming

Authors: J. Drevon, F. Millour, P. Cruzalèbes, C. Paladini, P. Scicluna, A. Matter, A. Chiavassa, M. Montargès, E. Cannon, F. Allouche, K. -H. Hofmann, S. Lagarde, B. Lopez, A. Meilland, R. Petrov, S. Robbe-Dubois, D. Schertl, G. Zins P. Abraham, P. Berio, Th. Henningm J. Hron, J. W. Isbell, W. Jaffe, L. Labadie, J. Varga, G. Weigelt , et al. (9 additional authors not shown)

Abstract: From Nov. 2019 to May 2020, the red supergiant star Betelgeuse experienced an unprecedented drop of brightness in the visible domain called the great dimming event. Large atmospheric dust clouds and large photospheric convective features are suspected to be responsible for it. To better understand the dimming event, we used mid-infrared long-baseline spectro-interferometric measurements of Betelge… ▽ More From Nov. 2019 to May 2020, the red supergiant star Betelgeuse experienced an unprecedented drop of brightness in the visible domain called the great dimming event. Large atmospheric dust clouds and large photospheric convective features are suspected to be responsible for it. To better understand the dimming event, we used mid-infrared long-baseline spectro-interferometric measurements of Betelgeuse taken with the VLTI/MATISSE instrument before (Dec. 2018), during (Feb. 2020), and after (Dec. 2020) the GDE. We present data in the 3.98 to 4.15\,$μみゅー$m range to cover SiO spectral features molecules as well as adjacent continuum. We have employed geometrical models, image reconstruction, as well as radiative transfer models to monitor the spatial distribution of SiO over the stellar surface. We find a strongly in-homogeneous spatial distribution of SiO that appears to be looking very different between our observing epochs, indicative of a vigorous activity in the stellar atmosphere. The contrast of our images is small in the pseudo-continuum for all epochs, implying that our MATISSE observations support both cold spot and dust cloud model. △ Less

Submitted 22 January, 2024; originally announced January 2024.

Beyond Human Data: Scaling Self-Training for Problem-Solving with Language Models

Authors: Avi Singh, John D. Co-Reyes, Rishabh Agarwal, Ankesh Anand, Piyush Patil, Xavier Garcia, Peter J. Liu, James Harrison, Jaehoon Lee, Kelvin Xu, Aaron Parisi, Abhishek Kumar, Alex Alemi, Alex Rizkowsky, Azade Nova, Ben Adlam, Bernd Bohnet, Gamaleldin Elsayed, Hanie Sedghi, Igor Mordatch, Isabelle Simpson, Izzeddin Gur, Jasper Snoek, Jeffrey Pennington, Jiri Hron , et al. (16 additional authors not shown)

Abstract: Fine-tuning language models~(LMs) on human-generated data remains a prevalent practice. However, the performance of such models is often limited by the quantity and diversity of high-quality human data. In this paper, we explore whether we can go beyond human data on tasks where we have access to scalar feedback, for example, on math problems where one can verify correctness. To do so, we investig… ▽ More Fine-tuning language models~(LMs) on human-generated data remains a prevalent practice. However, the performance of such models is often limited by the quantity and diversity of high-quality human data. In this paper, we explore whether we can go beyond human data on tasks where we have access to scalar feedback, for example, on math problems where one can verify correctness. To do so, we investigate a simple self-training method based on expectation-maximization, which we call ReST$^{EM}$, where we (1) generate samples from the model and filter them using binary feedback, (2) fine-tune the model on these samples, and (3) repeat this process a few times. Testing on advanced MATH reasoning and APPS coding benchmarks using PaLM-2 models, we find that ReST$^{EM}$ scales favorably with model size and significantly surpasses fine-tuning only on human data. Overall, our findings suggest self-training with feedback can substantially reduce dependence on human-generated data. △ Less

Submitted 17 April, 2024; v1 submitted 11 December, 2023; originally announced December 2023.

Comments: Accepted to TMLR. Camera-ready version. First three authors contributed equally

The Progression of Disparities within the Criminal Justice System: Differential Enforcement and Risk Assessment Instruments

Authors: Miri Zilka, Riccardo Fogliato, Jiri Hron, Bradley Butcher, Carolyn Ashurst, Adrian Weller

Abstract: Algorithmic risk assessment instruments (RAIs) increasingly inform decision-making in criminal justice. RAIs largely rely on arrest records as a proxy for underlying crime. Problematically, the extent to which arrests reflect overall offending can vary with the person's characteristics. We examine how the disconnect between crime and arrest rates impacts RAIs and their evaluation. Our main contrib… ▽ More Algorithmic risk assessment instruments (RAIs) increasingly inform decision-making in criminal justice. RAIs largely rely on arrest records as a proxy for underlying crime. Problematically, the extent to which arrests reflect overall offending can vary with the person's characteristics. We examine how the disconnect between crime and arrest rates impacts RAIs and their evaluation. Our main contribution is a method for quantifying this bias via estimation of the amount of unobserved offenses associated with particular demographics. These unobserved offenses are then used to augment real-world arrest records to create part real, part synthetic crime records. Using this data, we estimate that four currently deployed RAIs assign 0.5--2.8 percentage points higher risk scores to Black individuals than to White individuals with a similar \emph{arrest} record, but the gap grows to 4.5--11.0 percentage points when we match on the semi-synthetic \emph{crime} record. We conclude by discussing the potential risks around the use of RAIs, highlighting how they may exacerbate existing inequalities if the underlying disparities of the criminal justice system are not taken into account. In light of our findings, we provide recommendations to improve the development and evaluation of such tools. △ Less

Submitted 12 May, 2023; originally announced May 2023.

Comments: Accepted to FAccT '23

Optimising Human-Machine Collaboration for Efficient High-Precision Information Extraction from Text Documents

Authors: Bradley Butcher, Miri Zilka, Darren Cook, Jiri Hron, Adrian Weller

Abstract: While humans can extract information from unstructured text with high precision and recall, this is often too time-consuming to be practical. Automated approaches, on the other hand, produce nearly-immediate results, but may not be reliable enough for high-stakes applications where precision is essential. In this work, we consider the benefits and drawbacks of various human-only, human-machine, an… ▽ More While humans can extract information from unstructured text with high precision and recall, this is often too time-consuming to be practical. Automated approaches, on the other hand, produce nearly-immediate results, but may not be reliable enough for high-stakes applications where precision is essential. In this work, we consider the benefits and drawbacks of various human-only, human-machine, and machine-only information extraction approaches. We argue for the utility of a human-in-the-loop approach in applications where high precision is required, but purely manual extraction is infeasible. We present a framework and an accompanying tool for information extraction using weak-supervision labelling with human validation. We demonstrate our approach on three criminal justice datasets. We find that the combination of computer speed and human understanding yields precision comparable to manual annotation while requiring only a fraction of time, and significantly outperforms fully automated baselines in terms of precision. △ Less

Submitted 18 February, 2023; originally announced February 2023.

Locating dust and molecules in the inner circumstellar environment of R~Sculptoris with MATISSE

Authors: Julien Drevon, Florentin Millour, Pierre Cruzalèbes, Claudia Paladini, Josef Hron, A. Meilland, F. Allouche, K. -H. Hofmann, S. Lagarde, B. Lopez, A. Matter, R. Petrov, S. Robbe-Dubois, D. Schertl, M. Wittkowski, G. Zins, P. Ábrahám, P. Antonelli, U. Beckmann, P. Berio, F. Bettonvil, A. Glindemann, U. Graser, M. Heininger, Thomas Henning , et al. (27 additional authors not shown)

Abstract: AGB stars are one of the main sources of dust production in the Galaxy. However, it is not clear what this process looks like and where the dust is condensing in the circumstellar environment. By characterizing the location of the dust and the molecules in the close environment of an AGB star, we aim to achieve a better understanding the history of the dust formation process. We observed the carbo… ▽ More AGB stars are one of the main sources of dust production in the Galaxy. However, it is not clear what this process looks like and where the dust is condensing in the circumstellar environment. By characterizing the location of the dust and the molecules in the close environment of an AGB star, we aim to achieve a better understanding the history of the dust formation process. We observed the carbon star R Scl with the VLTI-MATISSE instrument in L- and N-bands. The high angular resolution of the VLTI observations, combined with a large uv-plane coverage allowed us to use image reconstruction methods. To constrain the dust and molecules' location, we used two different methods: MIRA image reconstruction and the 1D code RHAPSODY. We found evidence of C2H2 and HCN molecules between 1 and 3.4 Rstar which is much closer to the star than the location of the dust (between 3.8 and 17.0 Rstar). We also estimated a mass-loss rate of 1.2+-0.4x10-6 Msun per yr. In the meantime, we confirmed the previously published characteristics of a thin dust shell, composed of amorphous carbon (amC) and silicon carbide (SiC). However, no clear SiC feature has been detected in the MATISSE visibilities. This might be caused by molecular absorption that can affect the shape of the SiC band at 11.3 micron. The appearance of the molecular shells is in good agreement with predictions from dynamical atmosphere models. For the first time, we co-located dust and molecules in the environment of an AGB star. We confirm that the molecules are located closer to the star than the dust. The MIRA images unveil the presence of a clumpy environment in the fuzzy emission region beyond 4.0 Rstar. Furthermore, with the available dynamic range and angular resolution, we did not detect the presence of a binary companion. Additional observations combining MATISSE and SAM-VISIR instrument should enable this detection in future studies. △ Less

Submitted 23 August, 2022; originally announced August 2022.

Comments: 19 pages, published in A&A

Modeling Content Creator Incentives on Algorithm-Curated Platforms

Authors: Jiri Hron, Karl Krauth, Michael I. Jordan, Niki Kilbertus, Sarah Dean

Abstract: Content creators compete for user attention. Their reach crucially depends on algorithmic choices made by developers on online platforms. To maximize exposure, many creators adapt strategically, as evidenced by examples like the sprawling search engine optimization industry. This begets competition for the finite user attention pool. We formalize these dynamics in what we call an exposure game, a… ▽ More Content creators compete for user attention. Their reach crucially depends on algorithmic choices made by developers on online platforms. To maximize exposure, many creators adapt strategically, as evidenced by examples like the sprawling search engine optimization industry. This begets competition for the finite user attention pool. We formalize these dynamics in what we call an exposure game, a model of incentives induced by algorithms, including modern factorization and (deep) two-tower architectures. We prove that seemingly innocuous algorithmic choices, e.g., non-negative vs. unconstrained factorization, significantly affect the existence and character of (Nash) equilibria in exposure games. We proffer use of creator behavior models, like exposure games, for an (ex-ante) pre-deployment audit. Such an audit can identify misalignment between desirable and incentivized content, and thus complement post-hoc measures like content filtering and moderation. To this end, we propose tools for numerically finding equilibria in exposure games, and illustrate results of an audit on the MovieLens and LastFM datasets. Among else, we find that the strategically produced content exhibits strong dependence between algorithmic exploration and content diversity, and between model expressivity and bias towards gender-based user and creator groups. △ Less

Submitted 6 July, 2023; v1 submitted 27 June, 2022; originally announced June 2022.

Comments: presented at ICLR 2023 (top 5%)

Wide Bayesian neural networks have a simple weight posterior: theory and accelerated sampling

Authors: Jiri Hron, Roman Novak, Jeffrey Pennington, Jascha Sohl-Dickstein

Abstract: We introduce repriorisation, a data-dependent reparameterisation which transforms a Bayesian neural network (BNN) posterior to a distribution whose KL divergence to the BNN prior vanishes as layer widths grow. The repriorisation map acts directly on parameters, and its analytic simplicity complements the known neural network Gaussian process (NNGP) behaviour of wide BNNs in function space. Exploit… ▽ More We introduce repriorisation, a data-dependent reparameterisation which transforms a Bayesian neural network (BNN) posterior to a distribution whose KL divergence to the BNN prior vanishes as layer widths grow. The repriorisation map acts directly on parameters, and its analytic simplicity complements the known neural network Gaussian process (NNGP) behaviour of wide BNNs in function space. Exploiting the repriorisation, we develop a Markov chain Monte Carlo (MCMC) posterior sampling algorithm which mixes faster the wider the BNN. This contrasts with the typically poor performance of MCMC in high dimensions. We observe up to 50x higher effective sample size relative to no reparametrisation for both fully-connected and residual networks. Improvements are achieved at all widths, with the margin between reparametrised and standard BNNs growing with layer width. △ Less

Submitted 15 June, 2022; originally announced June 2022.

Comments: ICML 2022

Improving the diameters of interferometric calibrators with MATISSE

Authors: S. Robbe-Dubois, P. Cruzalèbes, Ph. Berio, A. Meilland, R. -G. Petrov, F. Allouche, D. Salabert, C. Paladini, A. Matter, F. Millour, S. Lagarde, B. Lopez, L. Burtscher, W. Jaffe, J. Hron, I. Percheron, R. van Boekel, G. Weigelt, Ph. Stee

Abstract: A good knowledge of the angular diameters of stars used to calibrate the observables in stellar interferometry is fundamental. As the available precision for giant stars is worse than the required per cent level, we aim to improve the knowledge of many diameters using MATISSE (Multiple AperTure mid-Infrared SpectroScopic Experiment) data in its different instrumental configurations. Using the squa… ▽ More A good knowledge of the angular diameters of stars used to calibrate the observables in stellar interferometry is fundamental. As the available precision for giant stars is worse than the required per cent level, we aim to improve the knowledge of many diameters using MATISSE (Multiple AperTure mid-Infrared SpectroScopic Experiment) data in its different instrumental configurations. Using the squared visibility MATISSE observable, we compute the angular diameter value, which ensures the best-fitting curves, assuming an intensity distribution of a uniform disc. We take into account that the transfer function varies over the wavelength and is different from one instrumental configuration to another. The uncertainties on the diameters are estimated using the residual bootstrap method. Using the low spectral resolution mode in the L band, we observed a set of 35 potential calibrators selected in the Mid-infrared stellar Diameter and Flux Compilation Catalogue with diameters ranging from about 1 to 3 mas. We reach a precision on the diameter estimates in the range 0.6 per cent to 4.1 per cent. The study of the stability of the transfer function in visibility over two nights makes us confident in our results. In addition, we identify one star, 75 Vir initially present in the calibrator lists, for which our method does not converge, and prove to be a binary star. This leads us to the conclusion that our method is actually necessary to improve the quality of the astrophysical results obtained with MATISSE, and that it can be used as a useful tool for 'bad calibrator' detection. △ Less

Submitted 3 January, 2022; originally announced January 2022.

Journal ref: Monthly Notices of the Royal Astronomical Society, Volume 510, Issue 1, p 82-94, February 2022

The extended atmosphere and circumstellar environment of the cool evolved star VX Sagittarii as seen by MATISSE

Authors: A. Chiavassa, K. Kravchenko, M. Montargès, F. Millour, A. Matter, B. Freytag, M. Wittkowski, V. Hocdé, P. Cruzalèbes, F. Allouche, B. Lopez, S. Lagarde, R. G. Petrov, A. Meilland, S. Robbe-Dubois, K. -H. Hofmann, G. Weigelt, P. Berio, P. Bendjoya, F. Bettonvil, A. Domiciano de Souza, M. Heininger, Th. Henning, J. W. Isbell, W. Jaffe , et al. (28 additional authors not shown)

Abstract: Context. VX Sgr is a cool, evolved, and luminous red star whose stellar parameters are difficult to determine, which affects its classification. Aims. We aim to spatially resolve the photospheric extent as well as the circumstellar environment. Methods. We used interferometric observations obtained with the MATISSE instrument in the L (3 to 4 μみゅーm), M (4.5 to 5 μみゅーm), and N (8 to 13 μみゅーm) bands. We reco… ▽ More Context. VX Sgr is a cool, evolved, and luminous red star whose stellar parameters are difficult to determine, which affects its classification. Aims. We aim to spatially resolve the photospheric extent as well as the circumstellar environment. Methods. We used interferometric observations obtained with the MATISSE instrument in the L (3 to 4 μみゅーm), M (4.5 to 5 μみゅーm), and N (8 to 13 μみゅーm) bands. We reconstructed monochromatic images using the MIRA software. We used 3D radiation-hydrodynamics (RHD) simulations carried out with CO5BOLD and a uniform disc model to estimate the apparent diameter and interpret the stellar surface structures. Moreover, we employed the radiative transfer codes Optim3D and Radmc3D to compute the spectral energy distribution for the L, M, and N bands, respectively. Results. MATISSE observations unveil, for the first time, the morphology of VX Sgr across the L, M, and N bands. The reconstructed images show a complex morphology with brighter areas whose characteristics depend on the wavelength probed. We measured the angular diameter as a function of the wavelength and showed that the photospheric extent in the L and M bands depends on the opacity through the atmosphere. In addition to this, we also concluded that the observed photospheric inhomogeneities can be interpreted as convection-related surface structures. The comparison in the N band yielded a qualitative agreement between the N band spectrum and simple dust radiative transfer simulations. However, it is not possible to firmly conclude on the interpretation of the current data because of the difficulty in constraing the model parameters using the limited accuracy of our absolute flux calibration. Conclusions. MATISSE observations and the derived reconstructed images unveil the appearance of the stellar surface and circumstellar environment across a very large spectral domain for the first time. △ Less

Submitted 20 December, 2021; originally announced December 2021.

Comments: Accepted for publication on Astronomy and Astrophysics

Journal ref: A&A 658, A185 (2022)

VLTI-MATISSE L- and N-band aperture-synthesis imaging of the unclassified B[e] star FS Canis Majoris

Authors: K. -H. Hofmann, A. Bensberg, D. Schertl, G. Weigelt, S. Wolf, A. Meilland, F. Millour, L. B. F. M. Waters, S. Kraus, K. Ohnaka, B. Lopez, R. G. Petrov, S. Lagarde, Ph. Berio, F. Allouche, S. Robbe-Dubois, W. Jaffe, Th. Henning, C. Paladini, M. Schöller, A. Mérand, A. Glindemann, U. Beckmann, M. Heininger, F. Bettonvil , et al. (36 additional authors not shown)

Abstract: Context: FS Canis Majoris (FS CMa, HD 45677) is an unclassified B[e] star surrounded by an inclined dust disk. The evolutionary stage of FS CMa is still debated. Perpendicular to the circumstellar disk, a bipolar outflow was detected. Infrared aperture-synthesis imaging provides us with a unique opportunity to study the disk structure. Aims: Our aim is to study the intensity distribution of the di… ▽ More Context: FS Canis Majoris (FS CMa, HD 45677) is an unclassified B[e] star surrounded by an inclined dust disk. The evolutionary stage of FS CMa is still debated. Perpendicular to the circumstellar disk, a bipolar outflow was detected. Infrared aperture-synthesis imaging provides us with a unique opportunity to study the disk structure. Aims: Our aim is to study the intensity distribution of the disk of FS CMa in the mid-infrared L and N bands. Methods: We performed aperture-synthesis imaging of FS CMa with the MATISSE instrument (Multi AperTure mid-Infrared SpectroScopic Experiment) in the low spectral resolution mode to obtain images in the L and N bands. We computed radiative transfer models that reproduce the L- and N-band intensity distributions of the resolved disks. Results: We present L- and N-band aperture-synthesis images of FS CMa reconstructed in the wavelength bands of 3.4-3.8 and 8.6-9.0 micrometer. In the L-band image, the inner rim region of an inclined circumstellar disk and the central object can be seen with a spatial resolution of 2.7 milliarcsec (mas). An inner disk cavity with an angular diameter of 6x12mas is resolved. The L-band disk consists of a bright northwestern (NW) disk region and a much fainter southeastern (SE) region. The images suggest that we are looking at the bright inner wall of the NW disk rim, which is on the far side of the disk. In the N band, only the bright NW disk region is seen. In addition to deriving the inclination and the inner disk radius, fitting the reconstructed brightness distributions via radiative transfer modeling allows one to constrain the innermost disk structure, in particular the shape of the inner disk rim. △ Less

Submitted 24 November, 2021; originally announced November 2021.

Comments: accepted for publication in A&A

MATISSE, the VLTI mid-infrared imaging spectro-interferometer

Authors: B. Lopez, S. Lagarde, R. G. Petrov, W. Jaffe, P. Antonelli, F. Allouche, P. Berio, A. Matter, A. Meilland, F. Millour, S. Robbe-Dubois, Th. Henning, G. Weigelt, A. Glindemann, T. Agocs, Ch. Bailet, U. Beckmann, F. Bettonvil, R. van Boekel, P. Bourget, Y. Bresson, P. Bristow, P. Cruzalèbes, E. Eldswijk, Y. Fanteï Caujolle , et al. (128 additional authors not shown)

Abstract: Context:Optical interferometry is at a key development stage. ESO's VLTI has established a stable, robust infrastructure for long-baseline interferometry for general astronomical observers. The present second-generation instruments offer a wide wavelength coverage and improved performance. Their sensitivity and measurement accuracy lead to data and images of high reliability. Aims:We have develope… ▽ More Context:Optical interferometry is at a key development stage. ESO's VLTI has established a stable, robust infrastructure for long-baseline interferometry for general astronomical observers. The present second-generation instruments offer a wide wavelength coverage and improved performance. Their sensitivity and measurement accuracy lead to data and images of high reliability. Aims:We have developed MATISSE, the Multi AperTure mid-Infrared SpectroScopic Experiment, to access high resolution imaging in a wide spectral domain and explore topics such: stellar activity and mass loss; planet formation and evolution in the gas and dust disks around young stars; accretion processes around super massive black holes in AGN. Methods:The instrument is a spectro-interferometric imager covering three atmospheric bands (L,M,N) from 2.8 to 13.0 mu, combining four optical beams from the VLTI's telscopes. Its concept, related observing procedure, data reduction and calibration approach are the product of 30 years of instrumental research. The instrument utilizes a multi-axial beam combination that delivers spectrally dispersed fringes. The signal provides the following quantities at several spectral resolutions: photometric flux, coherent fluxes, visibilities, closure phases, wavelength differential visibilities and phases, and aperture-synthesis imaging. Results:We provide an overview of the physical principle of the instrument and its functionalities, the characteristics of the delivered signal, a description of the observing modes and of their performance limits. An ensemble of data and reconstructed images are illustrating the first acquired key observations. Conclusion:The instrument has been in operation at Cerro Paranal, ESO, Chile since 2018, and has been open for science use by the international community since April 2019. The first scientific results are being published now. △ Less

Submitted 2 March, 2022; v1 submitted 29 October, 2021; originally announced October 2021.

Comments: 24 pages, 26 figures, submitted to Astronomy & Astrophysics

Journal ref: A&A 659, A192 (2022)

On the determination of vortex ring vorticity using Lagrangian particles

Authors: O. Outrata, M. Pavelka, J. Hron, M. La Mantia, J. I. Polanco, G. Krstulovic

Abstract: Particles are a widespread tool for obtaining information from fluid flows. When Eulerian data are unavailable, they may be employed to estimate flow fields or to identify coherent flow structures. Here we numerically examine the possibility of using particles to capture the dynamics of isolated vortex rings propagating in a quiescent fluid. The analysis is performed starting from numerical simula… ▽ More Particles are a widespread tool for obtaining information from fluid flows. When Eulerian data are unavailable, they may be employed to estimate flow fields or to identify coherent flow structures. Here we numerically examine the possibility of using particles to capture the dynamics of isolated vortex rings propagating in a quiescent fluid. The analysis is performed starting from numerical simulations of the Navier--Stokes and the Hall--Vinen--Bekarevich--Khalatnikov equations, respectively describing the dynamics of a Newtonian fluid and a finite-temperature superfluid. The flow-induced positions and velocities of particles suspended in the fluid are specifically used to compute the Lagrangian pseudovorticity field, a proxy for the Eulerian vorticity field recently employed in the context of superfluid $^4$He experiments. We show that, when calculated from ideal Lagrangian tracers or from particles with low inertia, the pseudovorticity field can be accurately used to estimate the propagation velocity and the growth of isolated vortex rings, although the quantitative reconstruction of the corresponding vorticity fields remains challenging. On the other hand, particles with high inertia tend to preferentially sample specific flow regions, resulting in biased pseudovorticity fields that pollute the estimation of the vortex ring properties. Overall, this work neatly demonstrates that the Lagrangian pseudovorticity is a valuable tool for estimating the strength of macroscopic vortical structures in the absence of Eulerian data, which is, for example, the case of superfluid $^4$He experiments. △ Less

Submitted 16 July, 2021; originally announced July 2021.

Journal ref: J. Fluid Mech. 924 (2021) A44

On component interactions in two-stage recommender systems

Authors: Jiri Hron, Karl Krauth, Michael I. Jordan, Niki Kilbertus

Abstract: Thanks to their scalability, two-stage recommenders are used by many of today's largest online platforms, including YouTube, LinkedIn, and Pinterest. These systems produce recommendations in two steps: (i) multiple nominators, tuned for low prediction latency, preselect a small subset of candidates from the whole item pool; (ii) a slower but more accurate ranker further narrows down the nominated… ▽ More Thanks to their scalability, two-stage recommenders are used by many of today's largest online platforms, including YouTube, LinkedIn, and Pinterest. These systems produce recommendations in two steps: (i) multiple nominators, tuned for low prediction latency, preselect a small subset of candidates from the whole item pool; (ii) a slower but more accurate ranker further narrows down the nominated items, and serves to the user. Despite their popularity, the literature on two-stage recommenders is relatively scarce, and the algorithms are often treated as mere sums of their parts. Such treatment presupposes that the two-stage performance is explained by the behavior of the individual components in isolation. This is not the case: using synthetic and real-world data, we demonstrate that interactions between the ranker and the nominators substantially affect the overall performance. Motivated by these findings, we derive a generalization lower bound which shows that independent nominator training can lead to performance on par with uniformly random recommendations. We find that careful design of item pools, each assigned to a different nominator, alleviates these issues. As manual search for a good pool allocation is difficult, we propose to learn one instead using a Mixture-of-Experts based approach. This significantly improves both precision and recall at K. △ Less

Submitted 12 January, 2022; v1 submitted 28 June, 2021; originally announced June 2021.

Comments: Appears in the proceedings of the NeurIPS 2021 conference

First MATISSE L-band observations of HD 179218. Is the inner 10 au region rich in carbon dust particles?

Authors: E. Kokoulina, A. Matter, B. Lopez, E. Pantin, N. Ysard, G. Weigelt, E. Habart, J. Varga, A. Jones, A. Meilland, E. Dartois, L. Klarmann, J. -C. Augereau, R. van Boekel, M. Hogerheijde, G. Yoffe, L. B. F. M. Waters, C. Dominik, W. Jaffe, F. Millour, Th. Henning, K. -H. Hofmann, D. Schertl, S. Lagarde, R. G. Petrov , et al. (36 additional authors not shown)

Abstract: Carbon is one of the most abundant components in the Universe. While silicates have been the main focus of solid phase studies in protoplanetary discs (PPDs), little is known about the solid carbon content especially in the planet-forming regions ($\sim $0.1 to 10 au). Fortunately, several refractory carbonaceous species present C-H bonds (such as hydrogenated nano-diamond and amorphous carbon as… ▽ More Carbon is one of the most abundant components in the Universe. While silicates have been the main focus of solid phase studies in protoplanetary discs (PPDs), little is known about the solid carbon content especially in the planet-forming regions ($\sim $0.1 to 10 au). Fortunately, several refractory carbonaceous species present C-H bonds (such as hydrogenated nano-diamond and amorphous carbon as well as polycyclic aromatic hydrocarbons (PAHs)), which generate infrared (IR) features that can be used to trace the solid carbon reservoirs. The new mid-IR instrument MATISSE, installed at the Very Large Telescope Interferometer (VLTI), can spatially resolve the inner regions ($\sim$ 1 to 10 au) of PPDs and locate, down to the au-scale, the emission coming from carbon grains. Our aim is to provide a consistent view on the radial structure, down to the au-scale, as well as basic physical properties and the nature of the material responsible for the IR continuum emission in the inner disk region around HD 179218. We implemented a temperature-gradient model to interpret the disk IR continuum emission, based on a multiwavelength dataset comprising a broadband spectral energy distribution (SED) and VLTI H-, L-, and N-bands interferometric data obtained in low spectral resolution. Then, we added a ring-like component, representing the carbonaceous L-band features-emitting region, to assess its detectability in future higher spectral resolution observations employing mid-IR interferometry. △ Less

Submitted 29 July, 2021; v1 submitted 24 June, 2021; originally announced June 2021.

Journal ref: A&A 652, A61 (2021)

Mid-infrared circumstellar emission of the long-period Cepheid l Carinae resolved with VLTI/MATISSE

Authors: V. Hocdé, N. Nardetto, A. Matter, E. Lagadec, A. Mérand, P. Cruzalèbes, A. Meilland, F. Millour, B. Lopez, P. Berio, G. Weigelt, R. Petrov, J. W. Isbell, W. Jaffe, P. Kervella, A. Glindemann, M. Schöller, F. Allouche, A. Gallenne, A. Domiciano de Souza, G. Niccolini, E. Kokoulina, J. Varga, S. Lagarde, J. -C. Augereau , et al. (129 additional authors not shown)

Abstract: The nature of circumstellar envelopes (CSE) around Cepheids is still a matter of debate. The physical origin of their infrared (IR) excess could be either a shell of ionized gas, or a dust envelope, or both. This study aims at constraining the geometry and the IR excess of the environment of the long-period Cepheid $\ell$ Car (P=35.5 days) at mid-IR wavelengths to understand its physical nature. W… ▽ More The nature of circumstellar envelopes (CSE) around Cepheids is still a matter of debate. The physical origin of their infrared (IR) excess could be either a shell of ionized gas, or a dust envelope, or both. This study aims at constraining the geometry and the IR excess of the environment of the long-period Cepheid $\ell$ Car (P=35.5 days) at mid-IR wavelengths to understand its physical nature. We first use photometric observations in various bands and Spitzer Space Telescope spectroscopy to constrain the IR excess of $\ell$ Car. Then, we analyze the VLTI/MATISSE measurements at a specific phase of observation, in order to determine the flux contribution, the size and shape of the environment of the star in the L band. We finally test the hypothesis of a shell of ionized gas in order to model the IR excess. We report the first detection in the L band of a centro-symmetric extended emission around l Car, of about 1.7$R_\star$ in FWHM, producing an excess of about 7.0\% in this band. In the N band, there is no clear evidence for dust emission from VLTI/MATISSE correlated flux and Spitzer data. On the other side, the modeled shell of ionized gas implies a more compact CSE ($1.13\pm0.02\,R_\star$) and fainter (IR excess of 1\% in the L band). We provide new evidences for a compact CSE of $\ell$ Car and we demonstrate the capabilities of VLTI/MATISSE for determining common properties of CSEs. While the compact CSE of $\ell$ Car is probably of gaseous nature, the tested model of a shell of ionized gas is not able to simultaneously reproduce the IR excess and the interferometric observations. Further Galactic Cepheids observations with VLTI/MATISSE are necessary for determining the properties of CSEs, which may also depend on both the pulsation period and the evolutionary state of the stars. △ Less

Submitted 31 March, 2021; originally announced March 2021.

Comments: 13 pages, 8 figures, accepted in Astronomy and Astrophysics

The asymmetric inner disk of the Herbig Ae star HD 163296 in the eyes of VLTI/MATISSE: evidence for a vortex?

Authors: J. Varga, M. Hogerheijde, R. van Boekel, L. Klarmann, R. Petrov, L. B. F. M. Waters, S. Lagarde, E. Pantin, Ph. Berio, G. Weigelt, S. Robbe-Dubois, B. Lopez, F. Millour, J. -C. Augereau, H. Meheut, A. Meilland, Th. Henning, W. Jaffe, F. Bettonvil, P. Bristow, K. -H. Hofmann, A. Matter, G. Zins, S. Wolf, F. Allouche , et al. (111 additional authors not shown)

Abstract: Context. The inner few au region of planet-forming disks is a complex environment. High angular resolution observations have a key role in understanding the disk structure and the dynamical processes at work. Aims. In this study we aim to characterize the mid-infrared brightness distribution of the inner disk of the young intermediate-mass star HD 163296, from VLTI/MATISSE observations. Methods. W… ▽ More Context. The inner few au region of planet-forming disks is a complex environment. High angular resolution observations have a key role in understanding the disk structure and the dynamical processes at work. Aims. In this study we aim to characterize the mid-infrared brightness distribution of the inner disk of the young intermediate-mass star HD 163296, from VLTI/MATISSE observations. Methods. We use geometric models to fit the data. Our models include a smoothed ring, a flat disk with inner cavity, and a 2D Gaussian. The models can account for disk inclination and for azimuthal asymmetries as well. We also perform numerical hydro-dynamical simulations of the inner edge of the disk. Results. Our modeling reveals a significant brightness asymmetry in the L-band disk emission. The brightness maximum of the asymmetry is located at the NW part of the disk image, nearly at the position angle of the semimajor axis. The surface brightness ratio in the azimuthal variation is $3.5 \pm 0.2$. Comparing our result on the location of the asymmetry with other interferometric measurements, we confirm that the morphology of the $r<0.3$ au disk region is time-variable. We propose that this asymmetric structure, located in or near the inner rim of the dusty disk, orbits the star. For the physical origin of the asymmetry, we tested a hypothesis where a vortex is created by Rossby wave instability, and we find that a unique large scale vortex may be compatible with our data. The half-light radius of the L-band emitting region is $0.33\pm 0.01$ au, the inclination is ${52^\circ}^{+5^\circ}_{-7^\circ}$, and the position angle is $143^\circ \pm 3^\circ$. Our models predict that a non-negligible fraction of the L-band disk emission originates inside the dust sublimation radius for $μみゅー$m-sized grains. Refractory grains or large ($\gtrsim 10\ μみゅー$m-sized) grains could be the origin for this emission. △ Less

Submitted 10 December, 2020; originally announced December 2020.

Comments: accepted for publication in A&A

Journal ref: A&A 647, A56 (2021)

Phase-field modeling of multivariant martensitic transformation at finite-strain: computational aspects and large-scale finite-element simulations

Authors: K. Tůma, M. Rezaee-Hajidehi, J. Hron, P. E. Farrell, S. Stupkiewicz

Abstract: Large-scale 3D martensitic microstructure evolution problems are studied using a finite-element discretization of a finite-strain phase-field model. The model admits an arbitrary crystallography of transformation and arbitrary elastic anisotropy of the phases, and incorporates Hencky-type elasticity, a penalty-regularized double-obstacle potential, and viscous dissipation. The finite-element discr… ▽ More Large-scale 3D martensitic microstructure evolution problems are studied using a finite-element discretization of a finite-strain phase-field model. The model admits an arbitrary crystallography of transformation and arbitrary elastic anisotropy of the phases, and incorporates Hencky-type elasticity, a penalty-regularized double-obstacle potential, and viscous dissipation. The finite-element discretization of the model is performed in Firedrake and relies on the PETSc solver library. The large systems of linear equations arising are efficiently solved using GMRES and a geometric multigrid preconditioner with a carefully chosen relaxation. The modeling capabilities are illustrated through a 3D simulation of the microstructure evolution in a pseudoelastic CuAlNi single crystal during nano-indentation, with all six orthorhombic martensite variants taken into account. Robustness and a good parallel scaling performance have been demonstrated, with the problem size reaching 150 million degrees of freedom. △ Less

Submitted 18 February, 2021; v1 submitted 17 November, 2020; originally announced November 2020.

Comments: Version accepted at Computer Methods in Applied Mechanics and Engineering. It includes Supplementary Material

Journal ref: Computer Methods in Applied Mechanics and Engineering 377, 113705 (2021)

Exploration in two-stage recommender systems

Authors: Jiri Hron, Karl Krauth, Michael I. Jordan, Niki Kilbertus

Abstract: Two-stage recommender systems are widely adopted in industry due to their scalability and maintainability. These systems produce recommendations in two steps: (i) multiple nominators preselect a small number of items from a large pool using cheap-to-compute item embeddings; (ii) with a richer set of features, a ranker rearranges the nominated items and serves them to the user. A key challenge of t… ▽ More Two-stage recommender systems are widely adopted in industry due to their scalability and maintainability. These systems produce recommendations in two steps: (i) multiple nominators preselect a small number of items from a large pool using cheap-to-compute item embeddings; (ii) with a richer set of features, a ranker rearranges the nominated items and serves them to the user. A key challenge of this setup is that optimal performance of each stage in isolation does not imply optimal global performance. In response to this issue, Ma et al. (2020) proposed a nominator training objective importance weighted by the ranker's probability of recommending each item. In this work, we focus on the complementary issue of exploration. Modeled as a contextual bandit problem, we find LinUCB (a near optimal exploration strategy for single-stage systems) may lead to linear regret when deployed in two-stage recommenders. We therefore propose a method of synchronising the exploration strategies between the ranker and the nominators. Our algorithm only relies on quantities already computed by standard LinUCB at each stage and can be implemented in three lines of additional code. We end by demonstrating the effectiveness of our algorithm experimentally. △ Less

Submitted 1 September, 2020; originally announced September 2020.

Comments: Published at the REVEAL 2020 workshop (RecSys 2020)

Origin of large meteoritic SiC stardust grains in metal-rich AGB stars

Authors: Maria Lugaro, Borbála Cseh, Blanka Világos, Amanda I. Karakas, Paolo Ventura, Flavia Dell'Agli, Reto Trappitsch, Melanie Hampel, Valentina D'Orazi, Claudio B. Pereira, Giuseppe Tagliente, Gyula M. Szabó, Marco Pignatari, Umberto Battino, Ashley Tattersall, Mattias Ek, Maria Schönbächler, Josef Hron, Larry R. Nittler

Abstract: Stardust grains that originated in ancient stars and supernovae are recovered from meteorites and carry the detailed composition of their astronomical sites of origin. We present evidence that the majority of large ($μみゅー$m-sized) meteoritic silicon carbide (SiC) grains formed in C-rich asymptotic giant branch (AGB) stars that were more metal-rich than the Sun. In the framework of the slow neutron-ca… ▽ More Stardust grains that originated in ancient stars and supernovae are recovered from meteorites and carry the detailed composition of their astronomical sites of origin. We present evidence that the majority of large ($μみゅー$m-sized) meteoritic silicon carbide (SiC) grains formed in C-rich asymptotic giant branch (AGB) stars that were more metal-rich than the Sun. In the framework of the slow neutron-captures (the s process) that occurs in AGB stars the lower-than-solar 88Sr/86Sr isotopic ratios measured in the large SiC grains can only be accompanied by Ce/Y elemental ratios that are also lower than solar, and predominately observed in metal-rich barium stars - the binary companions of AGB stars. Such an origin suggests that these large grains represent the material from high-metallicity AGB stars needed to explain the s-process nucleosynthesis variations observed in bulk meteorites (Ek et al. 2020). In the outflows of metal-rich, C-rich AGB stars SiC grains are predicted to be small ($\simeq$ 0.2 $μみゅー$m-sized); large ($\simeq$ $μみゅー$m-sized) SiC grains can grow if the number of dust seeds is two to three orders of magnitude lower than the standard value of $10^{-13}$ times the number of H atoms. We therefore predict that with increasing metallicity the number of dust seeds might decrease, resulting in the production of larger SiC grains. △ Less

Submitted 19 June, 2020; originally announced June 2020.

Comments: accepted for publication in The Astrophysical Journal

Exact posterior distributions of wide Bayesian neural networks

Authors: Jiri Hron, Yasaman Bahri, Roman Novak, Jeffrey Pennington, Jascha Sohl-Dickstein

Abstract: Recent work has shown that the prior over functions induced by a deep Bayesian neural network (BNN) behaves as a Gaussian process (GP) as the width of all layers becomes large. However, many BNN applications are concerned with the BNN function space posterior. While some empirical evidence of the posterior convergence was provided in the original works of Neal (1996) and Matthews et al. (2018), it… ▽ More Recent work has shown that the prior over functions induced by a deep Bayesian neural network (BNN) behaves as a Gaussian process (GP) as the width of all layers becomes large. However, many BNN applications are concerned with the BNN function space posterior. While some empirical evidence of the posterior convergence was provided in the original works of Neal (1996) and Matthews et al. (2018), it is limited to small datasets or architectures due to the notorious difficulty of obtaining and verifying exactness of BNN posterior approximations. We provide the missing theoretical proof that the exact BNN posterior converges (weakly) to the one induced by the GP limit of the prior. For empirical validation, we show how to generate exact samples from a finite BNN on a small dataset via rejection sampling. △ Less

Submitted 26 November, 2020; v1 submitted 18 June, 2020; originally announced June 2020.

Infinite attention: NNGP and NTK for deep attention networks

Authors: Jiri Hron, Yasaman Bahri, Jascha Sohl-Dickstein, Roman Novak

Abstract: There is a growing amount of literature on the relationship between wide neural networks (NNs) and Gaussian processes (GPs), identifying an equivalence between the two for a variety of NN architectures. This equivalence enables, for instance, accurate approximation of the behaviour of wide Bayesian NNs without MCMC or variational approximations, or characterisation of the distribution of randomly… ▽ More There is a growing amount of literature on the relationship between wide neural networks (NNs) and Gaussian processes (GPs), identifying an equivalence between the two for a variety of NN architectures. This equivalence enables, for instance, accurate approximation of the behaviour of wide Bayesian NNs without MCMC or variational approximations, or characterisation of the distribution of randomly initialised wide NNs optimised by gradient descent without ever running an optimiser. We provide a rigorous extension of these results to NNs involving attention layers, showing that unlike single-head attention, which induces non-Gaussian behaviour, multi-head attention architectures behave as GPs as the number of heads tends to infinity. We further discuss the effects of positional encodings and layer normalisation, and propose modifications of the attention mechanism which lead to improved results for both finite and infinitely wide NNs. We evaluate attention kernels empirically, leading to a moderate improvement upon the previous state-of-the-art on CIFAR-10 for GPs without trainable kernels and advanced data preprocessing. Finally, we introduce new features to the Neural Tangents library (Novak et al., 2020) allowing applications of NNGP/NTK models, with and without attention, to variable-length sequences, with an example on the IMDb reviews dataset. △ Less

Submitted 18 June, 2020; originally announced June 2020.

Comments: ICML 2020

Neural Tangents: Fast and Easy Infinite Neural Networks in Python

Authors: Roman Novak, Lechao Xiao, Jiri Hron, Jaehoon Lee, Alexander A. Alemi, Jascha Sohl-Dickstein, Samuel S. Schoenholz

Abstract: Neural Tangents is a library designed to enable research into infinite-width neural networks. It provides a high-level API for specifying complex and hierarchical neural network architectures. These networks can then be trained and evaluated either at finite-width as usual or in their infinite-width limit. Infinite-width networks can be trained analytically using exact Bayesian inference or using… ▽ More Neural Tangents is a library designed to enable research into infinite-width neural networks. It provides a high-level API for specifying complex and hierarchical neural network architectures. These networks can then be trained and evaluated either at finite-width as usual or in their infinite-width limit. Infinite-width networks can be trained analytically using exact Bayesian inference or using gradient descent via the Neural Tangent Kernel. Additionally, Neural Tangents provides tools to study gradient descent training dynamics of wide but finite networks in either function space or weight space. The entire library runs out-of-the-box on CPU, GPU, or TPU. All computations can be automatically distributed over multiple accelerators with near-linear scaling in the number of devices. Neural Tangents is available at www.github.com/google/neural-tangents. We also provide an accompanying interactive Colab notebook. △ Less

Submitted 5 December, 2019; originally announced December 2019.

A catalogue of stellar diameters and fluxes for mid-infrared interferometry

Authors: P. Cruzalèbes, R. G. Petrov, S. Robbe-Dubois, J. Varga, L. Burtscher, F. Allouche, P. Berio, K. -H. Hofmann, J. Hron, W. Jaffe, S. Lagarde, B. Lopez, A. Matter, A. Meilland, K. Meisenheimer, F. Millour, D. Schertl

Abstract: We present the Mid-infrared stellar Diameters and Fluxes compilation Catalogue (MDFC) dedicated to long-baseline interferometry at mid-infrared wavelengths (3-13 mum). It gathers data for half a million stars, i.e. nearly all the stars of the Hipparcos-Tycho catalogue whose spectral type is reported in the SIMBAD database. We cross-match 26 databases to provide basic information, binarity elements… ▽ More We present the Mid-infrared stellar Diameters and Fluxes compilation Catalogue (MDFC) dedicated to long-baseline interferometry at mid-infrared wavelengths (3-13 mum). It gathers data for half a million stars, i.e. nearly all the stars of the Hipparcos-Tycho catalogue whose spectral type is reported in the SIMBAD database. We cross-match 26 databases to provide basic information, binarity elements, angular diameter, magnitude and flux in the near and mid-infrared, as well as flags that allow us to identify the potential calibrators. The catalogue covers the entire sky with 465 857 stars, mainly dwarfs and giants from B to M spectral types closer than 18 kpc. The smallest reported values reach 0.16 muJy in L and 0.1 muJy in N for the flux, and 2 microarcsec for the angular diameter. We build 4 lists of calibrator candidates for the L- and N-bands suitable with the Very Large Telescope Interferometer (VLTI) sub- and main arrays using the MATISSE instrument. We identify 1 621 candidates for L and 44 candidates for N with the Auxiliary Telescopes (ATs), 375 candidates for both bands with the ATs, and 259 candidates for both bands with the Unit Telescopes (UTs). Predominantly cool giants, these sources are small and bright enough to belong to the primary lists of calibrator candidates. In the near future, we plan to measure their angular diameter with 1% accuracy. △ Less

Submitted 1 October, 2019; originally announced October 2019.

Comments: Accepted for publication in MNRAS (2019, oct. 1st)

On multiple solutions to the steady flow of incompressible fluids subject to do-nothing or constant traction boundary conditions on artificial boundaries

Authors: M. Lanzendörfer, J. Hron

Abstract: The boundary conditions prescribing the constant traction or the so-called do-nothing conditions are frequently taken on artificial boundaries in the numerical simulations of steady flow of incompressible fluids, despite the fact that they do not facilitate a well-posed problem, not allowing to establish the standard energy estimates. In a pursuit to understand better the possible consequences of… ▽ More The boundary conditions prescribing the constant traction or the so-called do-nothing conditions are frequently taken on artificial boundaries in the numerical simulations of steady flow of incompressible fluids, despite the fact that they do not facilitate a well-posed problem, not allowing to establish the standard energy estimates. In a pursuit to understand better the possible consequences of using these conditions, we present a particular set of examples of flow problems, where we find none or two analytical or numerical solutions. Namely, we consider problems where the flow connects two such artificial boundaries. In the simple case of the isotropic radial flows where both steady and unsteady analytical solutions are derived easily, we demonstrate that while for some (large) boundary data all unsteady solutions blow up in finite time, for other data (including small or trivial) the unsteady flows either converge asymptotically to one of two steady solutions, or blow up in finite time, depending on the initial state. We then document the very same behavior of the numerical solutions for planar flow in a diverging channel. Finally, we provide an illustrative example of two steady numerical solutions to the flow in a three-dimensional bifurcating tube, where the inflow velocity is prescribed at the inlet, while the two outlets are treated by the do-nothing boundary condition. △ Less

Submitted 24 February, 2020; v1 submitted 9 April, 2019; originally announced April 2019.

Comments: 15 pages

Journal ref: Journal of Mathematical Fluid Mechanics 22(11), 2020

Orthogonal Estimation of Wasserstein Distances

Authors: Mark Rowland, Jiri Hron, Yunhao Tang, Krzysztof Choromanski, Tamas Sarlos, Adrian Weller

Abstract: Wasserstein distances are increasingly used in a wide variety of applications in machine learning. Sliced Wasserstein distances form an important subclass which may be estimated efficiently through one-dimensional sorting operations. In this paper, we propose a new variant of sliced Wasserstein distance, study the use of orthogonal coupling in Monte Carlo estimation of Wasserstein distances and dr… ▽ More Wasserstein distances are increasingly used in a wide variety of applications in machine learning. Sliced Wasserstein distances form an important subclass which may be estimated efficiently through one-dimensional sorting operations. In this paper, we propose a new variant of sliced Wasserstein distance, study the use of orthogonal coupling in Monte Carlo estimation of Wasserstein distances and draw connections with stratified sampling, and evaluate our approaches experimentally in a range of large-scale experiments in generative modelling and reinforcement learning. △ Less

Submitted 5 April, 2019; v1 submitted 9 March, 2019; originally announced March 2019.

Comments: Published at AISTATS 2019

Successor Uncertainties: Exploration and Uncertainty in Temporal Difference Learning

Authors: David Janz, Jiri Hron, Przemysław Mazur, Katja Hofmann, José Miguel Hernández-Lobato, Sebastian Tschiatschek

Abstract: Posterior sampling for reinforcement learning (PSRL) is an effective method for balancing exploration and exploitation in reinforcement learning. Randomised value functions (RVF) can be viewed as a promising approach to scaling PSRL. However, we show that most contemporary algorithms combining RVF with neural network function approximation do not possess the properties which make PSRL effective, a… ▽ More Posterior sampling for reinforcement learning (PSRL) is an effective method for balancing exploration and exploitation in reinforcement learning. Randomised value functions (RVF) can be viewed as a promising approach to scaling PSRL. However, we show that most contemporary algorithms combining RVF with neural network function approximation do not possess the properties which make PSRL effective, and provably fail in sparse reward problems. Moreover, we find that propagation of uncertainty, a property of PSRL previously thought important for exploration, does not preclude this failure. We use these insights to design Successor Uncertainties (SU), a cheap and easy to implement RVF algorithm that retains key properties of PSRL. SU is highly effective on hard tabular exploration benchmarks. Furthermore, on the Atari 2600 domain, it surpasses human performance on 38 of 49 games tested (achieving a median human normalised score of 2.09), and outperforms its closest RVF competitor, Bootstrapped DQN, on 36 of those. △ Less

Submitted 3 December, 2019; v1 submitted 15 October, 2018; originally announced October 2018.

Comments: Camera ready version, NeurIPS 2019

Bayesian Deep Convolutional Networks with Many Channels are Gaussian Processes

Authors: Roman Novak, Lechao Xiao, Jaehoon Lee, Yasaman Bahri, Greg Yang, Jiri Hron, Daniel A. Abolafia, Jeffrey Pennington, Jascha Sohl-Dickstein

Abstract: There is a previously identified equivalence between wide fully connected neural networks (FCNs) and Gaussian processes (GPs). This equivalence enables, for instance, test set predictions that would have resulted from a fully Bayesian, infinitely wide trained FCN to be computed without ever instantiating the FCN, but by instead evaluating the corresponding GP. In this work, we derive an analogous… ▽ More There is a previously identified equivalence between wide fully connected neural networks (FCNs) and Gaussian processes (GPs). This equivalence enables, for instance, test set predictions that would have resulted from a fully Bayesian, infinitely wide trained FCN to be computed without ever instantiating the FCN, but by instead evaluating the corresponding GP. In this work, we derive an analogous equivalence for multi-layer convolutional neural networks (CNNs) both with and without pooling layers, and achieve state of the art results on CIFAR10 for GPs without trainable kernels. We also introduce a Monte Carlo method to estimate the GP corresponding to a given neural network architecture, even in cases where the analytic form has too many terms to be computationally feasible. Surprisingly, in the absence of pooling layers, the GPs corresponding to CNNs with and without weight sharing are identical. As a consequence, translation equivariance, beneficial in finite channel CNNs trained with stochastic gradient descent (SGD), is guaranteed to play no role in the Bayesian treatment of the infinite channel limit - a qualitative difference between the two regimes that is not present in the FCN case. We confirm experimentally, that while in some scenarios the performance of SGD-trained finite CNNs approaches that of the corresponding GPs as the channel count increases, with careful tuning SGD-trained CNNs can significantly outperform their corresponding GPs, suggesting advantages from SGD training compared to fully Bayesian parameter estimation. △ Less

Submitted 21 August, 2020; v1 submitted 11 October, 2018; originally announced October 2018.

Comments: Published as a conference paper at ICLR 2019

Variational Bayesian dropout: pitfalls and fixes

Authors: Jiri Hron, Alexander G. de G. Matthews, Zoubin Ghahramani

Abstract: Dropout, a stochastic regularisation technique for training of neural networks, has recently been reinterpreted as a specific type of approximate inference algorithm for Bayesian neural networks. The main contribution of the reinterpretation is in providing a theoretical framework useful for analysing and extending the algorithm. We show that the proposed framework suffers from several issues; fro… ▽ More Dropout, a stochastic regularisation technique for training of neural networks, has recently been reinterpreted as a specific type of approximate inference algorithm for Bayesian neural networks. The main contribution of the reinterpretation is in providing a theoretical framework useful for analysing and extending the algorithm. We show that the proposed framework suffers from several issues; from undefined or pathological behaviour of the true posterior related to use of improper priors, to an ill-defined variational objective due to singularity of the approximating distribution relative to the true posterior. Our analysis of the improper log uniform prior used in variational Gaussian dropout suggests the pathologies are generally irredeemable, and that the algorithm still works only because the variational formulation annuls some of the pathologies. To address the singularity issue, we proffer Quasi-KL (QKL) divergence, a new approximate inference objective for approximation of high-dimensional distributions. We show that motivations for variational Bernoulli dropout based on discretisation and noise have QKL as a limit. Properties of QKL are studied both theoretically and on a simple practical example which shows that the QKL-optimal approximation of a full rank Gaussian with a degenerate one naturally leads to the Principal Component Analysis solution. △ Less

Submitted 5 July, 2018; originally announced July 2018.

Comments: Extended version of the paper accepted to ICML 2018: more details in the proofs, few minor modifications

Gaussian Process Behaviour in Wide Deep Neural Networks

Authors: Alexander G. de G. Matthews, Mark Rowland, Jiri Hron, Richard E. Turner, Zoubin Ghahramani

Abstract: Whilst deep neural networks have shown great empirical success, there is still much work to be done to understand their theoretical properties. In this paper, we study the relationship between random, wide, fully connected, feedforward networks with more than one hidden layer and Gaussian processes with a recursive kernel definition. We show that, under broad conditions, as we make the architectur… ▽ More Whilst deep neural networks have shown great empirical success, there is still much work to be done to understand their theoretical properties. In this paper, we study the relationship between random, wide, fully connected, feedforward networks with more than one hidden layer and Gaussian processes with a recursive kernel definition. We show that, under broad conditions, as we make the architecture increasingly wide, the implied random function converges in distribution to a Gaussian process, formalising and extending existing results by Neal (1996) to deep networks. To evaluate convergence rates empirically, we use maximum mean discrepancy. We then compare finite Bayesian deep networks from the literature to Gaussian processes in terms of the key predictive quantities of interest, finding that in some cases the agreement can be very close. We discuss the desirability of Gaussian process behaviour and review non-Gaussian alternative models from the literature. △ Less

Submitted 16 August, 2018; v1 submitted 30 April, 2018; originally announced April 2018.

Comments: This work substantially extends the work of Matthews et al. (2018) published at the International Conference on Learning Representations (ICLR) 2018

Variational Gaussian Dropout is not Bayesian

Authors: Jiri Hron, Alexander G. de G. Matthews, Zoubin Ghahramani

Abstract: Gaussian multiplicative noise is commonly used as a stochastic regularisation technique in training of deterministic neural networks. A recent paper reinterpreted the technique as a specific algorithm for approximate inference in Bayesian neural networks; several extensions ensued. We show that the log-uniform prior used in all the above publications does not generally induce a proper posterior, a… ▽ More Gaussian multiplicative noise is commonly used as a stochastic regularisation technique in training of deterministic neural networks. A recent paper reinterpreted the technique as a specific algorithm for approximate inference in Bayesian neural networks; several extensions ensued. We show that the log-uniform prior used in all the above publications does not generally induce a proper posterior, and thus Bayesian inference in such models is ill-posed. Independent of the log-uniform prior, the correlated weight noise approximation has further issues leading to either infinite objective or high risk of overfitting. The above implies that the reported sparsity of obtained solutions cannot be explained by Bayesian or the related minimum description length arguments. We thus study the objective from a non-Bayesian perspective, provide its previously unknown analytical form which allows exact gradient evaluation, and show that the later proposed additive reparametrisation introduces minima not present in the original multiplicative parametrisation. Implications and future research directions are discussed. △ Less

Submitted 8 November, 2017; originally announced November 2017.

Concrete Dropout

Authors: Yarin Gal, Jiri Hron, Alex Kendall

Abstract: Dropout is used as a practical tool to obtain uncertainty estimates in large vision models and reinforcement learning (RL) tasks. But to obtain well-calibrated uncertainty estimates, a grid-search over the dropout probabilities is necessary - a prohibitive operation with large models, and an impossible one with RL. We propose a new dropout variant which gives improved performance and better calibr… ▽ More Dropout is used as a practical tool to obtain uncertainty estimates in large vision models and reinforcement learning (RL) tasks. But to obtain well-calibrated uncertainty estimates, a grid-search over the dropout probabilities is necessary - a prohibitive operation with large models, and an impossible one with RL. We propose a new dropout variant which gives improved performance and better calibrated uncertainties. Relying on recent developments in Bayesian deep learning, we use a continuous relaxation of dropout's discrete masks. Together with a principled optimisation objective, this allows for automatic tuning of the dropout probability in large models, and as a result faster experimentation cycles. In RL this allows the agent to adapt its uncertainty dynamically as more data is observed. We analyse the proposed variant extensively on a range of tasks, and give insights into common practice in the field where larger dropout probabilities are often used in deeper model layers. △ Less

Submitted 22 May, 2017; originally announced May 2017.

Aperture synthesis imaging of the carbon AGB star R Sculptoris: Detection of a complex structure and a dominating spot on the stellar disk

Authors: M. Wittkowski, K. -H. Hofmann, S. Höfner, J. B. Le Bouquin, W. Nowotny, C. Paladini, J. Young, J. -P. Berger, M. Brunner, I. de Gregorio-Monsalvo, K. Eriksson, J. Hron, E. M. L. Humphreys, M. Lindqvist, M. Maercker, S. Mohamed, H. Olofsson, S. Ramstedt, G. Weigelt

Abstract: We present near-infrared interferometry of the carbon-rich asymptotic giant branch (AGB) star R Sculptoris. The visibility data indicate a broadly circular resolved stellar disk with a complex substructure. The observed AMBER squared visibility values show drops at the positions of CO and CN bands, indicating that these lines form in extended layers above the photosphere. The AMBER visibility va… ▽ More We present near-infrared interferometry of the carbon-rich asymptotic giant branch (AGB) star R Sculptoris. The visibility data indicate a broadly circular resolved stellar disk with a complex substructure. The observed AMBER squared visibility values show drops at the positions of CO and CN bands, indicating that these lines form in extended layers above the photosphere. The AMBER visibility values are best fit by a model without a wind. The PIONIER data are consistent with the same model. We obtain a Rosseland angular diameter of 8.9+-0.3 mas, corresponding to a Rosseland radius of 355+-55 Rsun, an effective temperature of 2640+-80 K, and a luminosity of log L/Lsun=3.74+-0.18. These parameters match evolutionary tracks of initial mass 1.5+-0.5 Msun and current mass 1.3+-0.7 Msun. The reconstructed PIONIER images exhibit a complex structure within the stellar disk including a dominant bright spot located at the western part of the stellar disk. The spot has an H-band peak intensity of 40% to 60% above the average intensity of the limb-darkening-corrected stellar disk. The contrast between the minimum and maximum intensity on the stellar disk is about 1:2.5. Our observations are broadly consistent with predictions by dynamic atmosphere and wind models, although models with wind appear to have a circumstellar envelope that is too extended compared to our observations. The detected complex structure within the stellar disk is most likely caused by giant convection cells, resulting in large-scale shock fronts, and their effects on clumpy molecule and dust formation seen against the photosphere at distances of 2-3 stellar radii. △ Less

Submitted 8 February, 2017; originally announced February 2017.

Comments: 13 pages, 11 figures, 4 pages appendix, accepted for publication in Astronomy and Astrophysics (A&A)

Journal ref: A&A 601, A3 (2017)

The VLTI/MIDI view on the inner mass loss of evolved stars from the Herschel MESS sample

Authors: C. Paladini, D. Klotz, S. Sacuto, E. Lagadec, M. Wittkowski, A. Richichi, J. Hron, A. Jorissen, M. A. T. Groenewegen, F. Kerschbaum, T. Verhoelst, G. Rau, H. Olofsson, R. Zhao-Geisler, A. Matter

Abstract: The mass-loss process from evolved stars is a key ingredient for our understanding of many fields of astrophysics, including stellar evolution and the chemical enrichment of the interstellar medium via stellar yields. One the main unsolved questions is the geometry of the mass-loss process. Taking advantage of the results from the Herschel Mass loss of Evolved StarS (MESS) programme, we initiated… ▽ More The mass-loss process from evolved stars is a key ingredient for our understanding of many fields of astrophysics, including stellar evolution and the chemical enrichment of the interstellar medium via stellar yields. One the main unsolved questions is the geometry of the mass-loss process. Taking advantage of the results from the Herschel Mass loss of Evolved StarS (MESS) programme, we initiated a coordinated effort to characterise the geometry of mass loss from evolved red giants at various spatial scales. For this purpose we used the MID-infrared interferometric Instrument (MIDI) to resolve the inner envelope of 14 asymptotic giant branch stars (AGBs) in the MESS sample. In this contribution we present an overview of the interferometric data collected within the frame of our Large Programme, and we also add archive data for completeness. We studied the geometry of the inner atmosphere by comparing the observations with predictions from different geometric models. Asymmetries are detected for five O-rich and S-type, suggesting that asymmetries in the N band are more common among stars with such chemistry. We speculate that this fact is related to the characteristics of the dust grains. Except for one star, no interferometric variability is detected, i.e. the changes in size of the shells of non-mira stars correspond to changes of the visibility of less than 10%. The observed spectral variability confirms previous findings from the literature. The detection of dust in our sample follows the location of the AGBs in the IRAS colour-colour diagram: more dust is detected around oxygen-rich stars in region II and in the carbon stars in region VII. The SiC dust feature does not appear in the visibility spectrum of UAnt and SSct, which are two carbon stars with detached shells. This finding has implications for the theory of SiC dust formation. △ Less

Submitted 26 January, 2017; v1 submitted 19 January, 2017; originally announced January 2017.

Comments: 43 pages, 31 figures; accepted for publication in Astronomy & Astrophysics. Abstract shortened for compilation reasons. Metadata corrected

Journal ref: A&A 600, A136 (2017)

The adventure of carbon stars - Observations and modelling of a set of C-rich AGB stars

Authors: Gioia Rau, Josef Hron, Claudia Paladini, Benard Aringer, Kjell Eriksson, Paola Marigo, Walter Nowotny, Rebekka Grellmann

Abstract: Modeling stellar atmospheres is a complex, intriguing task in modern astronomy. A systematic comparison of models with multi-technique observations is the only efficient way to constrain them. Aims: We performed a self-consistent modeling of the atmospheres of six C-rich AGB stars: R Lep, R Vol, Y Pav, AQ Sgr, U Hya and X TrA, with the aim of enlarging the knowledge of the dynamic processes occurr… ▽ More Modeling stellar atmospheres is a complex, intriguing task in modern astronomy. A systematic comparison of models with multi-technique observations is the only efficient way to constrain them. Aims: We performed a self-consistent modeling of the atmospheres of six C-rich AGB stars: R Lep, R Vol, Y Pav, AQ Sgr, U Hya and X TrA, with the aim of enlarging the knowledge of the dynamic processes occurring in their atmospheres. Methods. We used VLTI/MIDI interferometric observations, in combination with spectro-photometric data, and compared them with self-consistent dynamic models atmospheres. Results: The models can reproduce SED data well at wavelengths longwards of 1μみゅーm, and the interferometric observations between 8μみゅーm and 10μみゅーm. Discrepancies could be due to a combination of data- and model-related effects. The models best fitting the Miras are significantly extended, with a prominent shell-like structure, while the models best fitting the non-Miras are more compact, showing lower average mass-loss. The mass loss is of episodic or multi-periodic nature, but causes the visual amplitudes to be notably larger than the observed ones. Stellar parameters were derived from the model fitting: T_eff, L_bol, M, C/O, $\dot{\text{M}}$. Our findings agree well with literature values within the uncertainties. T_eff and L_bol are also in good agreement with the T derived from the angular diameter θしーた_{(V-K)} and L_bol from the SED fitting, except for AQ Sgr. Finally, θしーた_Ross and θしーた_(V-K) agree with each other better for the Miras than for the non-Miras,which is probably connected to the episodic nature of the latter models. We also located the stars in the H-R diagram, comparing them with evolutionary tracks. We found that the main derived properties from the model fitting are in good agreement with TP-AGB evolutionary calculations for carbon stars (COLIBRI code). △ Less

Submitted 16 January, 2017; originally announced January 2017.

Comments: 22 pages. Accepted for publication in A&A

Journal ref: A&A 600, A92 (2017)

On thermodynamics of viscoelastic rate type fluids with temperature dependent material coefficients

Authors: Jaroslav Hron, Vojtěch Miloš, Vít Průša, Ondřej Souček, Karel Tůma

Abstract: We derive a class of thermodynamically consistent variants of Maxwell/Oldroyd-B type models for viscoelastic fluids. In particular, we study the models that allow one to consider temperature dependent material coefficients. This naturally calls for the formulation of a temperature evolution equation that would accompany the evolution equations for the mechanical quantities. The evolution equation… ▽ More We derive a class of thermodynamically consistent variants of Maxwell/Oldroyd-B type models for viscoelastic fluids. In particular, we study the models that allow one to consider temperature dependent material coefficients. This naturally calls for the formulation of a temperature evolution equation that would accompany the evolution equations for the mechanical quantities. The evolution equation for the temperature is explicitly formulated, and it is shown to be consistent with the laws of thermodynamics and the evolution equations for the mechanical quantities. The temperature evolution equation contains terms that are ignored or even not thought of in most of the works dealing with this class of fluids. The impact of the additional terms in the temperature evolution equation on the flow dynamics is documented by the solution of simple initial/boundary value problems. △ Less

Submitted 13 December, 2016; v1 submitted 6 December, 2016; originally announced December 2016.

Comments: Introduction, Page 2 and Page 3: We have been referred to relevant papers by I. J. Rao, K. Kannan and K. R. Rajagopal, who have considered viscoelastic rate type fluids with temperature dependent material coefficients. The comment on the relation between their work and our work has been added to the introduction

MSC Class: 76A05; 35Q79

Science with MATISSE

Authors: S. Wolf, B. Lopez, J. -Ch. Augereau, M. Delbo, C. Dominik, Th. Henning, K. -H. Hofmann, M. Hogerheijde, J. Hron, W. Jaffe, Th. Lanz, K. Meisenheimer, F. Millour, E. Pantin, R. Petrov, D. Schertl, R. van Boekel, G. Weigelt, A. Chiavassa, A. Juhasz, A. Matter, A. Meilland, N. Nardetto, C. Paladini

Abstract: We present an overview of the scientific potential of MATISSE, the Multi Aperture mid-Infrared SpectroScopic Experiment for the Very Large Telescope Interferometer. For this purpose we outline selected case studies from various areas, such as star and planet formation, active galactic nuclei, evolved stars, extrasolar planets, and solar system minor bodies and discuss strategies for the planning a… ▽ More We present an overview of the scientific potential of MATISSE, the Multi Aperture mid-Infrared SpectroScopic Experiment for the Very Large Telescope Interferometer. For this purpose we outline selected case studies from various areas, such as star and planet formation, active galactic nuclei, evolved stars, extrasolar planets, and solar system minor bodies and discuss strategies for the planning and analysis of future MATISSE observations. Moreover, the importance of MATISSE observations in the context of complementary high-angular resolution observations at near-infrared and submillimeter/millimeter wavelengths is highlighted. △ Less

Submitted 20 September, 2016; originally announced September 2016.

Comments: SPIE Astronomical Telescopes and Instrumentation conference, June 2016, 20 pages, 6 Figures

An overview of the mid-infrared spectro-interferometer MATISSE: science, concept, and current status

Authors: A. Matter, B. Lopez, P. Antonelli, M. Lehmitz, F. Bettonvil, U. Beckmann, S. Lagarde, W. Jaffe, R. G. Petrov, P. Berio, F. Millour, S. Robbe-Dubois, A. Glindemann, P. Bristow, M. Schoeller, T. Lanz, T. Henning, G. Weigelt, M. Heininger, S. Morel, P. Cruzalebes, K. Meisenheimer, R. Hofferbert, S. Wolf, Y. Bresson , et al. (82 additional authors not shown)

Abstract: MATISSE is the second-generation mid-infrared spectrograph and imager for the Very Large Telescope Interferometer (VLTI) at Paranal. This new interferometric instrument will allow significant advances by opening new avenues in various fundamental research fields: studying the planet-forming region of disks around young stellar objects, understanding the surface structures and mass loss phenomena a… ▽ More MATISSE is the second-generation mid-infrared spectrograph and imager for the Very Large Telescope Interferometer (VLTI) at Paranal. This new interferometric instrument will allow significant advances by opening new avenues in various fundamental research fields: studying the planet-forming region of disks around young stellar objects, understanding the surface structures and mass loss phenomena affecting evolved stars, and probing the environments of black holes in active galactic nuclei. As a first breakthrough, MATISSE will enlarge the spectral domain of current optical interferometers by offering the L and M bands in addition to the N band. This will open a wide wavelength domain, ranging from 2.8 to 13 um, exploring angular scales as small as 3 mas (L band) / 10 mas (N band). As a second breakthrough, MATISSE will allow mid-infrared imaging - closure-phase aperture-synthesis imaging - with up to four Unit Telescopes (UT) or Auxiliary Telescopes (AT) of the VLTI. Moreover, MATISSE will offer a spectral resolution range from R ~ 30 to R ~ 5000. Here, we present one of the main science objectives, the study of protoplanetary disks, that has driven the instrument design and motivated several VLTI upgrades (GRA4MAT and NAOMI). We introduce the physical concept of MATISSE including a description of the signal on the detectors and an evaluation of the expected performances. We also discuss the current status of the MATISSE instrument, which is entering its testing phase, and the foreseen schedule for the next two years that will lead to the first light at Paranal. △ Less

Submitted 8 August, 2016; originally announced August 2016.

Comments: SPIE Astronomical Telescopes and Instrumentation conference, June 2016, 11 pages, 6 Figures

Grown-up stars physics with MATISSE

Authors: Florentin Millour, J Hron, A Chiavassa, G Weigelt, A Soulain, Z Khorrami, A Meilland, N Nardetto, C Paladini, A Domiciano de Souza, G Niccolini, K. -H Hofmann, D Schertl, P Stee, P Bendjoya, F Thévenin, F Vakili, P Berio, T Lanz, A Matter, P Cruzalèbes, R Petrov, B Lopez

Abstract: MATISSE represents a great opportunity to image the environment around massive and evolved stars. This will allow one to put constraints on the circumstellar structure, on the mass ejection of dust and its reorganization , and on the dust-nature and formation processes. MATISSE measurements will often be pivotal for the understanding of large multiwavelength datasets on the same targets collected… ▽ More MATISSE represents a great opportunity to image the environment around massive and evolved stars. This will allow one to put constraints on the circumstellar structure, on the mass ejection of dust and its reorganization , and on the dust-nature and formation processes. MATISSE measurements will often be pivotal for the understanding of large multiwavelength datasets on the same targets collected through many high-angular resolution facilities at ESO like sub-millimeter interferometry (ALMA), near-infrared adaptive optics (NACO, SPHERE), interferometry (PIONIER, GRAVITY), spectroscopy (CRIRES), and mid-infrared imaging (VISIR). Among main sequence and evolved stars, several cases of interest have been identified that we describe in this paper. △ Less

Submitted 8 July, 2016; originally announced July 2016.

Comments: SPIE, Jun 2016, Edimbourgh, France

The complex environment of the bright carbon star TX Psc as probed by spectro-astrometry

Authors: J. Hron, S. Uttenthaler, B. Aringer, D. Klotz, T. Lebzelter, C. Paladini, G. Wiedemann

Abstract: Context: Stars on the asymptotic giant branch (AGB) show broad evidence of inhomogeneous atmospheres and circumstellar envelopes. These have been studied by a variety of methods on various angular scales. In this paper we explore the envelope of the well-studied carbon star TX Psc by the technique of spectro-astrometry. Aims: We explore the potential of this method for detecting asymmetries around… ▽ More Context: Stars on the asymptotic giant branch (AGB) show broad evidence of inhomogeneous atmospheres and circumstellar envelopes. These have been studied by a variety of methods on various angular scales. In this paper we explore the envelope of the well-studied carbon star TX Psc by the technique of spectro-astrometry. Aims: We explore the potential of this method for detecting asymmetries around AGB stars. Methods:We obtained CRIRES observations of several CO $Δでるた$v=1 lines near 4.6 $μみゅー$m and HCN lines near 3 $μみゅー$m in 2010 and 2013. These were then searched for spectro-astrometric signatures. For the interpretation of the results, we used simple simulated observations. Results: Several lines show significant photocentre shifts with a clear dependence on position angle. In all cases, tilde-shaped signatures are found where the positive and negative shifts (at PA 0deg) are associated with blue and weaker red components of the lines. The shifts can be modelled with a bright blob 70 mas to 210 mas south of the star with a flux of several percent of the photospheric flux. We estimate a lower limit of the blob temperature of 1000 K. The blob may be related to a mass ejection as found for AGB stars or red supergiants. We also consider the scenario of a companion object. Conclusions: Although there is clear spectro-astrometric evidence of a rather prominent structure near TX Psc, it does not seem to relate to the other evidence of asymmetries, so no definite explanation can be given. Our data thus underline the very complex structure of the environment of this star, but further observations that sample the angular scales out to a few hundred milli-arcseconds are needed to get a clearer picture. △ Less

Submitted 9 October, 2015; originally announced October 2015.

Journal ref: A&A 584, A27 (2015)

Modelling the atmosphere of the carbon-rich Mira RU Vir

Authors: G. Rau, C. Paladini, J. Hron, B. Aringer, M. A. T. Groenewegen, W. Nowotny

Abstract: Context. We study the atmosphere of the carbon-rich Mira RU Vir using the mid-infrared high spatial resolution interferometric observations from VLTI/MIDI. Aims. The aim of this work is to analyse the atmosphere of the carbon-rich Mira RU Vir, with state of the art models, in this way deepening the knowledge of the dynamic processes at work in carbon-rich Miras. Methods. We compare spectro-photome… ▽ More Context. We study the atmosphere of the carbon-rich Mira RU Vir using the mid-infrared high spatial resolution interferometric observations from VLTI/MIDI. Aims. The aim of this work is to analyse the atmosphere of the carbon-rich Mira RU Vir, with state of the art models, in this way deepening the knowledge of the dynamic processes at work in carbon-rich Miras. Methods. We compare spectro-photometric and interferometric measurements of this carbon-rich Mira AGB star, with the predictions of different kinds of modelling approaches (hydrostatic model atmospheres plus MOD-More Of Dusty, self-consistent dynamic model atmospheres). A geometric model fitting tool is used for a first interpretation of the interferometric data. Results. The results show that a joint use of different kind of observations (photometry, spectroscopy, interferometry) is essential to shed light on the structure of the atmosphere of a carbon-rich Mira. The dynamic model atmospheres fit well the ISO spectrum in the wavelength range λらむだ = [2.9, 25.0] μみゅーm. Nevertheless, a discrepancy is noticeable both in the SED (visible), and in the visibilities (shape and level). A possible explanation are intra-/inter-cycle variations in the dynamic model atmospheres as well as in the observations. The presence of a companion star and/or a disk or a decrease of mass loss within the last few hundred years cannot be excluded but are considered unlikely. △ Less

Submitted 31 August, 2015; v1 submitted 12 June, 2015; originally announced June 2015.

Comments: 15 pages. Accepted in A&A

Journal ref: A&A 583, A106 (2015)

Dissecting the AGB star L2 Puppis: a torus in the making

Authors: F. Lykou, D. Klotz, C. Paladini, J. Hron, A. A. Zijlstra, J. Kluska, B. R. M. Norris, P. G. Tuthill, S. Ramstedt, E. Lagadec, M. Wittkowski, M. Maercker, J. Meisner, A. Mayer

Abstract: The circumstellar environment of L2 Pup, an oxygen-rich semiregular variable, was observed to understand the evolution of mass loss and the shaping of ejecta in the late stages of stellar evolution. High-angular resolution observations from a single 8 m telescope were obtained using aperture masking in the near-infrared (1.64, 2.30 and 3.74 $\rmμみゅーm$) on the NACO/VLT, both in imaging and polarimetri… ▽ More The circumstellar environment of L2 Pup, an oxygen-rich semiregular variable, was observed to understand the evolution of mass loss and the shaping of ejecta in the late stages of stellar evolution. High-angular resolution observations from a single 8 m telescope were obtained using aperture masking in the near-infrared (1.64, 2.30 and 3.74 $\rmμみゅーm$) on the NACO/VLT, both in imaging and polarimetric modes. The aperture-masking images of L2 Pup at 2.30 $\rmμみゅーm$ show a resolved structure that resembles a toroidal structure with a major axis of ~140 milliarcseconds (mas) and an east-west orientation. Two clumps can be seen on either side of the star, ~65 mas from the star, beyond the edge of the circumstellar envelope (estimated diameter is ~27 mas), while a faint, hook-like structure appear toward the northeast. The patterns are visible both in the imaging and polarimetric mode, although the latter was only used to measure the total intensity (Stokes I). The overall shape of the structure is similar at the 3.74 $\rmμみゅーm$ pseudo-continuum (dust emission), where the clumps appear to be embedded within a dark, dusty lane. The faint, hook-like patterns are also seen at this wavelength, extending northeast and southwest with the central, dark lane being an apparent axis of symmetry. We interpret the structure as a circumstellar torus with inner radius of 4.2 au. With a rotation velocity of 10 km s$^{-1}$ as suggested by the SiO maser profile, we estimate a stellar mass of 0.7 M$_\odot$.} △ Less

Submitted 17 March, 2015; originally announced March 2015.

Comments: 9 pages, 7 figures, to be published in A&A

Journal ref: A&A 576, A46 (2015)

The PTI Carbon Star Angular Size Survey: Effective Temperatures and Non-Sphericity

Authors: Gerard T. van Belle, Claudia Paladini, Bernhard Aringer, Josef Hron, David R. Ciardi

Abstract: We report new interferometric angular diameter observations of 41 carbon stars observed with the Palomar Testbed Interferometer (PTI). Two of these stars are CH carbon stars and represent the first such measurements for this subtype. Of these, 39 have Yamashita (1972,1975) spectral classes and are of sufficiently high quality that we may determine the dependence of effective temperature on spectra… ▽ More We report new interferometric angular diameter observations of 41 carbon stars observed with the Palomar Testbed Interferometer (PTI). Two of these stars are CH carbon stars and represent the first such measurements for this subtype. Of these, 39 have Yamashita (1972,1975) spectral classes and are of sufficiently high quality that we may determine the dependence of effective temperature on spectral type. We find that there is a tendency for the effective temperature to increase with increasing temperature index by ~120K per step, starting at T_EFF ~= 2500K for C3,y, although there is a large amount of scatter about this relationship. Overall, the median effective temperature for the carbon star sample is found to be 2800 +- 270K, and the median linear radius is 360 +- 100 R_SUN. We also find agreement on average within 15K with the T_EFF determinations of Bergeat (2001,2002a,b), and a refinement of carbon star angular size prediction based on V & K magnitudes is presented that is good to an rms of 12%. A subsample of our stars have sufficient {u,v} coverage to permit non-spherical modeling of their photospheres, and a general tendency for detection of statistically significant departures from sphericity with increasing signal-to-noise of the interferometric data is seen. The implications of most - and potentially all - carbon stars being non-spherical is considered in the context of surface inhomogeneities and a rotation-mass loss connection. △ Less

Submitted 24 July, 2013; originally announced July 2013.

Comments: 59 pages, 15 figures, 6 tables; accepted for publication in ApJ

Catching the fish - Constraining stellar parameters for TX Psc using spectro-interferometric observations

Authors: D. Klotz, C. Paladini, J. Hron, B. Aringer, S. Sacuto, P. Marigo, T. Verhoelst

Abstract: Stellar parameter determination is a challenging task when dealing with galactic giant stars. The combination of different investigation techniques has proven to be a promising approach. We analyse archive spectra obtained with the Short-Wavelength-Spectrometer (SWS) onboard of ISO, and new interferometric observations from the Very Large Telescope MID-infrared Interferometric instrument (VLTI/MID… ▽ More Stellar parameter determination is a challenging task when dealing with galactic giant stars. The combination of different investigation techniques has proven to be a promising approach. We analyse archive spectra obtained with the Short-Wavelength-Spectrometer (SWS) onboard of ISO, and new interferometric observations from the Very Large Telescope MID-infrared Interferometric instrument (VLTI/MIDI) of a very well studied carbon-rich giant: TX Psc. The aim of this work is to determine stellar parameters using spectroscopy and interferometry. The observations are used to constrain the model atmosphere, and eventually the stellar evolutionary model in the region where the tracks map the beginning of the carbon star sequence. Two different approaches are used to determine stellar parameters: (i) the 'classic' interferometric approach where the effective temperature is fixed by using the angular diameter in the N-band (from interferometry) and the apparent bolometric magnitude; (ii) parameters are obtained by fitting a grid of state-of-the-art hydrostatic models to spectroscopic and interferometric observations. We find a good agreement between the parameters of the two methods. The effective temperature and luminosity clearly place TX Psc in the carbon-rich AGB star domain in the H-R-diagram. Current evolutionary tracks suggest that TX Psc became a C-star just recently, which means that the star is still in a 'quiet' phase compared to the subsequent strong-wind regime. This is in agreement with the C/O ratio being only slightly larger than 1. △ Less

Submitted 3 January, 2013; originally announced January 2013.

Comments: 11 pages, 9 figures, 5 tables

Detection of an asymmetry in the envelope of the carbon Mira R Fornacis using VLTI/MIDI

Authors: C. Paladini, S. Sacuto, D. Klotz, K. Ohnaka, M. Wittkowski, W. Nowotny, A. Jorissen, J. Hron

Abstract: Aims. We present a study of the envelope morphology of the carbon Mira R For with VLTI/MIDI. This object is one of the few asymptotic giant branch (AGB) stars that underwent a dust-obscuration event. The cause of such events is still a matter of discussion. Several symmetric and asymmetric scenarios have been suggested in the literature. Methods. Mid-infrared interferometric observations were obta… ▽ More Aims. We present a study of the envelope morphology of the carbon Mira R For with VLTI/MIDI. This object is one of the few asymptotic giant branch (AGB) stars that underwent a dust-obscuration event. The cause of such events is still a matter of discussion. Several symmetric and asymmetric scenarios have been suggested in the literature. Methods. Mid-infrared interferometric observations were obtained separated by two years. The observations probe different depths of the atmosphere and cover different pulsation phases. The visibilities and the differential phases were interpreted using GEM-FIND, a tool for fitting spectrally dispersed interferometric observations with the help of wavelength-dependent geometric models. Results. We report the detection of an asymmetric structure revealed through the MIDI differential phase. This asymmetry is observed at the same baseline and position angle two years later. The observations are best simulated with a model that includes a uniform-disc plus a Gaussian envelope plus a point-source. The geometric model can reproduce both the visibilities and the differential phase signatures. Conclusions. Our MIDI data favour explanations of the R For obscuration event that are based on an asymmetric geometry. We clearly detect a photocentre shift between the star and the strongly resolved dust component. This might be caused by a dust clump or a substellar companion. However, the available observations do not allow us to distinguish between the two options. The finding has strong implications for future studies of the geometry of the envelope of AGB stars: if this is a binary, are all AGB stars that show an obscuration event binaries as well? Or are we looking at asymmetric mass-loss processes (i.e. dusty clumps) in the inner part of a carbon-rich Mira? △ Less

Submitted 19 July, 2012; v1 submitted 17 July, 2012; originally announced July 2012.

Comments: accepted for publication as A&A letter

Geometrical model fitting for interferometric data: GEM-FIND

Authors: D. Klotz, S. Sacuto, C. Paladini, J. Hron, G. Wachter

Abstract: We developed the tool GEM-FIND that allows to constrain the morphology and brightness distribution of objects. The software fits geometrical models to spectrally dispersed interferometric visibility measurements in the N-band using the Levenberg-Marquardt minimization method. Each geometrical model describes the brightness distribution of the object in the Fourier space using a set of wavelength-i… ▽ More We developed the tool GEM-FIND that allows to constrain the morphology and brightness distribution of objects. The software fits geometrical models to spectrally dispersed interferometric visibility measurements in the N-band using the Levenberg-Marquardt minimization method. Each geometrical model describes the brightness distribution of the object in the Fourier space using a set of wavelength-independent and/or wavelength-dependent parameters. In this contribution we numerically analyze the stability of our nonlinear fitting approach by applying it to sets of synthetic visibilities with statistically applied errors, answering the following questions: How stable is the parameter determination with respect to (i) the number of uv-points, (ii) the distribution of points in the uv-plane, (iii) the noise level of the observations? △ Less

Submitted 9 July, 2012; originally announced July 2012.

Comments: 14 pages, 12 figures; to appear in SPIE Astronomical Telescopes and Instrumentation, SPIE Conference Series 8445

The geometry of the close environment of SV Psc as probed by VLTI/MIDI

Authors: D. Klotz, S. Sacuto, F. Kerschbaum, C. Paladini, H. Olofsson, J. Hron

Abstract: Context. SV Psc is an asymptotic giant branch (AGB) star surrounded by an oxygen-rich dust envelope. The mm-CO line profile of the object's outflow shows a clear double-component structure. Because of the high angular resolution, mid-IR interferometry may give strong constraints on the origin of this composite profile. Aims. The aim of this work is to investigate the morphology of the environmen… ▽ More Context. SV Psc is an asymptotic giant branch (AGB) star surrounded by an oxygen-rich dust envelope. The mm-CO line profile of the object's outflow shows a clear double-component structure. Because of the high angular resolution, mid-IR interferometry may give strong constraints on the origin of this composite profile. Aims. The aim of this work is to investigate the morphology of the environment around SV Psc using high-angular resolution interferometry observations in the mid-IR with the Very Large Telescope MID-infrared Interferometric instrument (VLTI/MIDI). Methods. Interferometric data in the N-band taken at different baseline lengths (ranging from 32-64 m) and position angles (73- 142°) allow a study of the morphology of the circumstellar environment close to the star. The data are interpreted on the basis of 2-dimensional, chromatic geometrical models using the fitting software tool GEM-FIND developed for this purpose. Results. The results favor two scenarios: (i) the presence of a highly inclined, optically thin, dusty disk surrounding the central star; (ii) the presence of an unresolved binary companion at a separation of 13.7 AUえーゆー and a position angle of 121.8° NE. The derived orbital period of the binary is 38.1 yr. This detection is in good agreement with hydrodynamic simulations showing that a close companion could be responsible for the entrainment of the gas and dust into a circumbinary structure. △ Less

Submitted 23 April, 2012; originally announced April 2012.

Comments: 10 pages, 12 figures

Imaging the heart of astrophysical objects with optical long-baseline interferometry

Authors: J. -P. Berger, F. Malbet, F. Baron, A. Chiavassa, G. Duvert, M. Elitzur, B. Freytag, F. Gueth, S. Hönig, J. Hron, H. Jang-Condell, J. -B. Le Bouquin, J. -L Monin, J. D. Monnier, G. Perrin, B. Plez, T. Ratzka, S. Renard, S. Stefl, E. Thiébaut, K. Tristram, T. Verhoelst, S. Wolf, J. Young

Abstract: The number of publications of aperture-synthesis images based on optical long-baseline interferometry measurements has recently increased due to easier access to visible and infrared interferometers. The interferometry technique has now reached a technical maturity level that opens new avenues for numerous astrophysical topics requiring milli-arcsecond model-independent imaging. In writing this pa… ▽ More The number of publications of aperture-synthesis images based on optical long-baseline interferometry measurements has recently increased due to easier access to visible and infrared interferometers. The interferometry technique has now reached a technical maturity level that opens new avenues for numerous astrophysical topics requiring milli-arcsecond model-independent imaging. In writing this paper our motivation was twofold: 1) review and publicize emblematic excerpts of the impressive corpus accumulated in the field of optical interferometry image reconstruction; 2) discuss future prospects for this technique by selecting four representative astrophysical science cases in order to review the potential benefits of using optical long baseline interferometers. For this second goal we have simulated interferometric data from those selected astrophysical environments and used state-of-the-art codes to provide the reconstructed images that are reachable with current or soon-to-be facilities. The image reconstruction process was "blind" in the sense that reconstructors had no knowledge of the input brightness distributions. We discuss the impact of optical interferometry in those four astrophysical fields. We show that image reconstruction software successfully provides accurate morphological information on a variety of astrophysical topics and review the current strengths and weaknesses of such reconstructions. We investigate how to improve image reconstruction and the quality of the image possibly by upgrading the current facilities. We finally argue that optical interferometers and their corresponding instrumentation, existing or to come, with 6 to 10 telescopes, should be well suited to provide images of complex sceneries. △ Less

Submitted 19 April, 2012; originally announced April 2012.

Comments: Acccepted to Astronomy and Astrophysics Review

X Her and TX Psc: Two cases of ISM interaction with stellar winds observed by Herschel

Authors: A. Jorissen, A. Mayer, S. Van Eck, R. Ottensamer, F. Kerschbaum, T. Ueta, P. Bergman, J. A. D. L. Blommaert, L. Decin, M. A. T. Groenewegen, J. Hron, W. Nowotny, H. Olofsson, Th. Posch, L. O. Sjouwerman, B. Vandenbussche, C. Waelkens

Abstract: The asymptotic giant branch (AGB) stars X Her and TX Psc have been imaged at 70 and 160 microns with the PACS instrument onboard the Herschel satellite, as part of the large MESS (Mass loss of Evolved StarS) Guaranteed Time Key Program. The images reveal an axisymmetric extended structure with its axis oriented along the space motion of the stars. This extended structure is very likely to be shape… ▽ More The asymptotic giant branch (AGB) stars X Her and TX Psc have been imaged at 70 and 160 microns with the PACS instrument onboard the Herschel satellite, as part of the large MESS (Mass loss of Evolved StarS) Guaranteed Time Key Program. The images reveal an axisymmetric extended structure with its axis oriented along the space motion of the stars. This extended structure is very likely to be shaped by the interaction of the wind ejected by the AGB star with the surrounding interstellar medium (ISM). As predicted by numerical simulations, the detailed structure of the wind-ISM interface depends upon the relative velocity between star+wind and the ISM, which is large for these two stars (108 and 55 km/s for X Her and TX Psc, respectively). In both cases, there is a compact blob upstream whose origin is not fully elucidated, but that could be the signature of some instability in the wind-ISM shock. Deconvolved images of X Her and TX Psc reveal several discrete structures along the outermost filaments, which could be Kelvin-Helmholtz vortices. Finally, TX Psc is surrounded by an almost circular ring (the signature of the termination shock?) that contrasts with the outer, more structured filaments. A similar inner circular structure seems to be present in X Her as well, albeit less clearly. △ Less

Submitted 18 June, 2011; originally announced June 2011.

Comments: 11 pages, Astronomy & Astrophysics, in press