Showing 1–49 of 49 results for author: Ashley, R

Scaling Value Iteration Networks to 5000 Layers for Extreme Long-Term Planning

Authors: Yuhui Wang, Qingyuan Wu, Weida Li, Dylan R. Ashley, Francesco Faccio, Chao Huang, Jürgen Schmidhuber

Abstract: The Value Iteration Network (VIN) is an end-to-end differentiable architecture that performs value iteration on a latent MDP for planning in reinforcement learning (RL). However, VINs struggle to scale to long-term and large-scale planning tasks, such as navigating a $100\times 100$ maze -- a task which typically requires thousands of planning steps to solve. We observe that this deficiency is due… ▽ More The Value Iteration Network (VIN) is an end-to-end differentiable architecture that performs value iteration on a latent MDP for planning in reinforcement learning (RL). However, VINs struggle to scale to long-term and large-scale planning tasks, such as navigating a $100\times 100$ maze -- a task which typically requires thousands of planning steps to solve. We observe that this deficiency is due to two issues: the representation capacity of the latent MDP and the planning module's depth. We address these by augmenting the latent MDP with a dynamic transition kernel, dramatically improving its representational capacity, and, to mitigate the vanishing gradient problem, introducing an "adaptive highway loss" that constructs skip connections to improve gradient flow. We evaluate our method on both 2D maze navigation environments and the ViZDoom 3D navigation benchmark. We find that our new method, named Dynamic Transition VIN (DT-VIN), easily scales to 5000 layers and casually solves challenging versions of the above tasks. Altogether, we believe that DT-VIN represents a concrete step forward in performing long-term large-scale planning in RL environments. △ Less

Submitted 12 June, 2024; originally announced June 2024.

ACM Class: I.2.6

Towards a Robust Soft Baby Robot With Rich Interaction Ability for Advanced Machine Learning Algorithms

Authors: Mohannad Alhakami, Dylan R. Ashley, Joel Dunham, Francesco Faccio, Eric Feron, Jürgen Schmidhuber

Abstract: Artificial intelligence has made great strides in many areas lately, yet it has had comparatively little success in general-use robotics. We believe one of the reasons for this is the disconnect between traditional robotic design and the properties needed for open-ended, creativity-based AI systems. To that end, we, taking selective inspiration from nature, build a robust, partially soft robotic l… ▽ More Artificial intelligence has made great strides in many areas lately, yet it has had comparatively little success in general-use robotics. We believe one of the reasons for this is the disconnect between traditional robotic design and the properties needed for open-ended, creativity-based AI systems. To that end, we, taking selective inspiration from nature, build a robust, partially soft robotic limb with a large action space, rich sensory data stream from multiple cameras, and the ability to connect with others to enhance the action space and data stream. As a proof of concept, we train two contemporary machine learning algorithms to perform a simple target-finding task. Altogether, we believe that this design serves as a first step to building a robot tailor-made for achieving artificial general intelligence. △ Less

Submitted 11 April, 2024; originally announced April 2024.

Comments: 5 pages in main text + 1 page of references, 7 figures in main text; source code available at https://github.com/dylanashley/robot-limb-testai

ACM Class: I.2.9; I.2.6

TIC 378898110: A Bright, Short-Period AM CVn Binary in TESS

Authors: Matthew J. Green, J. J. Hermes, Brad N. Barlow, T. R. Marsh, Ingrid Pelisoli, Boris T. Gänsicke, Ben C. Kaiser, Alejandra Romero, Larissa Antunes Amaral, Kyle Corcoran, Dirk Grupe, Mark R. Kennedy, S. O. Kepler, James Munday, R. P. Ashley, Andrzej S. Baran, Elmé Breedt, Alex J. Brown, V. S. Dhillon, Martin J. Dyer, Paul Kerry, George W. King, S. P. Littlefair, Steven G. Parsons, David I. Sahman

Abstract: AM CVn-type systems are ultracompact, helium-accreting binary systems which are evolutionarily linked to the progenitors of thermonuclear supernovae and are expected to be strong Galactic sources of gravitational waves detectable to upcoming space-based interferometers. AM CVn binaries with orbital periods $\lesssim$ 20--23 min exist in a constant high state with a permanently ionised accretion di… ▽ More AM CVn-type systems are ultracompact, helium-accreting binary systems which are evolutionarily linked to the progenitors of thermonuclear supernovae and are expected to be strong Galactic sources of gravitational waves detectable to upcoming space-based interferometers. AM CVn binaries with orbital periods $\lesssim$ 20--23 min exist in a constant high state with a permanently ionised accretion disc. We present the discovery of TIC 378898110, a bright ($G=14.3$ mag), nearby ($309.3 \pm 1.8$ pc), high-state AM CVn binary discovered in TESS two-minute-cadence photometry. At optical wavelengths this is the third-brightest AM CVn binary known. The photometry of the system shows a 23.07172(6) min periodicity, which is likely to be the `superhump' period and implies an orbital period in the range 22--23 min. There is no detectable spectroscopic variability. The system underwent an unusual, year-long brightening event during which the dominant photometric period changed to a shorter period (constrained to $20.5 \pm 2.0$ min), which we suggest may be evidence for the onset of disc-edge eclipses. The estimated mass transfer rate, $\log (\dot{M} / \mathrm{M_\odot} \mathrm{yr}^{-1}) = -6.8 \pm 1.0$, is unusually high and may suggest a high-mass or thermally inflated donor. The binary is detected as an X-ray source, with a flux of $9.2 ^{+4.2}_{-1.8} \times 10^{-13}$ erg cm$^{-2}$ s$^{-1}$ in the 0.3--10 keV range. TIC 378898110 is the shortest-period binary system discovered with TESS, and its large predicted gravitational-wave amplitude makes it a compelling verification binary for future space-based gravitational wave detectors. △ Less

Submitted 2 November, 2023; originally announced November 2023.

Comments: 15 pages, 14 figures. Accepted to MNRAS

Mindstorms in Natural Language-Based Societies of Mind

Authors: Mingchen Zhuge, Haozhe Liu, Francesco Faccio, Dylan R. Ashley, Róbert Csordás, Anand Gopalakrishnan, Abdullah Hamdi, Hasan Abed Al Kader Hammoud, Vincent Herrmann, Kazuki Irie, Louis Kirsch, Bing Li, Guohao Li, Shuming Liu, **jie Mai, Piotr Piękos, Aditya Ramesh, Imanol Schlag, Weimin Shi, Aleksandar Stanić, Wenyi Wang, Yuhui Wang, Mengmeng Xu, Deng-** Fan, Bernard Ghanem , et al. (1 additional authors not shown)

Abstract: Both Minsky's "society of mind" and Schmidhuber's "learning to think" inspire diverse societies of large multimodal neural networks (NNs) that solve problems by interviewing each other in a "mindstorm." Recent implementations of NN-based societies of minds consist of large language models (LLMs) and other NN-based experts communicating through a natural language interface. In doing so, they overco… ▽ More Both Minsky's "society of mind" and Schmidhuber's "learning to think" inspire diverse societies of large multimodal neural networks (NNs) that solve problems by interviewing each other in a "mindstorm." Recent implementations of NN-based societies of minds consist of large language models (LLMs) and other NN-based experts communicating through a natural language interface. In doing so, they overcome the limitations of single LLMs, improving multimodal zero-shot reasoning. In these natural language-based societies of mind (NLSOMs), new agents -- all communicating through the same universal symbolic language -- are easily added in a modular fashion. To demonstrate the power of NLSOMs, we assemble and experiment with several of them (having up to 129 members), leveraging mindstorms in them to solve some practical AI tasks: visual question answering, image captioning, text-to-image synthesis, 3D generation, egocentric retrieval, embodied AI, and general language-based task solving. We view this as a starting point towards much larger NLSOMs with billions of agents-some of which may be humans. And with this emergence of great societies of heterogeneous minds, many new research questions have suddenly become paramount to the future of artificial intelligence. What should be the social structure of an NLSOM? What would be the (dis)advantages of having a monarchical rather than a democratic structure? How can principles of NN economies be used to maximize the total reward of a reinforcement learning NLSOM? In this work, we identify, discuss, and try to answer some of these questions. △ Less

Submitted 26 May, 2023; originally announced May 2023.

Comments: 9 pages in main text + 7 pages of references + 38 pages of appendices, 14 figures in main text + 13 in appendices, 7 tables in appendices

MSC Class: 68T07 ACM Class: I.2.6; I.2.11

The wide-field, multiplexed, spectroscopic facility WEAVE: Survey design, overview, and simulated implementation

Authors: Shoko **, Scott C. Trager, Gavin B. Dalton, J. Alfonso L. Aguerri, J. E. Drew, Jesús Falcón-Barroso, Boris T. Gänsicke, Vanessa Hill, Angela Iovino, Matthew M. Pieri, Bianca M. Poggianti, D. J. B. Smith, Antonella Vallenari, Don Carlos Abrams, David S. Aguado, Teresa Antoja, Alfonso Aragón-Salamanca, Yago Ascasibar, Carine Babusiaux, Marc Balcells, R. Barrena, Giuseppina Battaglia, Vasily Belokurov, Thomas Bensby, Piercarlo Bonifacio , et al. (190 additional authors not shown)

Abstract: WEAVE, the new wide-field, massively multiplexed spectroscopic survey facility for the William Herschel Telescope, will see first light in late 2022. WEAVE comprises a new 2-degree field-of-view prime-focus corrector system, a nearly 1000-multiplex fibre positioner, 20 individually deployable 'mini' integral field units (IFUs), and a single large IFU. These fibre systems feed a dual-beam spectrogr… ▽ More WEAVE, the new wide-field, massively multiplexed spectroscopic survey facility for the William Herschel Telescope, will see first light in late 2022. WEAVE comprises a new 2-degree field-of-view prime-focus corrector system, a nearly 1000-multiplex fibre positioner, 20 individually deployable 'mini' integral field units (IFUs), and a single large IFU. These fibre systems feed a dual-beam spectrograph covering the wavelength range 366$-$959\,nm at $R\sim5000$, or two shorter ranges at $R\sim20\,000$. After summarising the design and implementation of WEAVE and its data systems, we present the organisation, science drivers and design of a five- to seven-year programme of eight individual surveys to: (i) study our Galaxy's origins by completing Gaia's phase-space information, providing metallicities to its limiting magnitude for $\sim$3 million stars and detailed abundances for $\sim1.5$ million brighter field and open-cluster stars; (ii) survey $\sim0.4$ million Galactic-plane OBA stars, young stellar objects and nearby gas to understand the evolution of young stars and their environments; (iii) perform an extensive spectral survey of white dwarfs; (iv) survey $\sim400$ neutral-hydrogen-selected galaxies with the IFUs; (v) study properties and kinematics of stellar populations and ionised gas in $z<0.5$ cluster galaxies; (vi) survey stellar populations and kinematics in $\sim25\,000$ field galaxies at $0.3\lesssim z \lesssim 0.7$; (vii) study the cosmic evolution of accretion and star formation using $>1$ million spectra of LOFAR-selected radio sources; (viii) trace structures using intergalactic/circumgalactic gas at $z>2$. Finally, we describe the WEAVE Operational Rehearsals using the WEAVE Simulator. △ Less

Submitted 31 October, 2023; v1 submitted 7 December, 2022; originally announced December 2022.

Comments: 41 pages, 27 figures, accepted for publication by MNRAS; updated version including information on individual grants in a revised Acknowledgements section, corrections to the affiliation list, and an updated references list

On Narrative Information and the Distillation of Stories

Authors: Dylan R. Ashley, Vincent Herrmann, Zachary Friggstad, Jürgen Schmidhuber

Abstract: The act of telling stories is a fundamental part of what it means to be human. This work introduces the concept of narrative information, which we define to be the overlap in information space between a story and the items that compose the story. Using contrastive learning methods, we show how modern artificial neural networks can be leveraged to distill stories and extract a representation of the… ▽ More The act of telling stories is a fundamental part of what it means to be human. This work introduces the concept of narrative information, which we define to be the overlap in information space between a story and the items that compose the story. Using contrastive learning methods, we show how modern artificial neural networks can be leveraged to distill stories and extract a representation of the narrative information. We then demonstrate how evolutionary algorithms can leverage this to extract a set of narrative templates and how these templates -- in tandem with a novel curve-fitting algorithm we introduce -- can reorder music albums to automatically induce stories in them. In the process of doing so, we give strong statistical evidence that these narrative information templates are present in existing albums. While we experiment only with music albums here, the premises of our work extend to any form of (largely) independent media. △ Less

Submitted 13 February, 2023; v1 submitted 22 November, 2022; originally announced November 2022.

Comments: presented in the Information-Theoretic Principles in Cognitive Systems Workshop at the 36th Conference on Neural Information Processing Systems; 4 pages in main text + 2 pages of references + 8 pages of appendices, 2 figures in main text + 3 in appendices, 1 table in main text, 2 algorithms in appendices; source code available at https://github.com/dylanashley/story-distiller

MSC Class: 68T07 (Primary) 68P30; 68W50; 94A15 (Secondary) ACM Class: H.1.1; H.5.5; I.2.6; I.5.1; J.5

Long-term photometric monitoring and spectroscopy of the white dwarf pulsar AR Scorpii

Authors: Ingrid Pelisoli, T. R. Marsh, S. G. Parsons, A. Aungwerojwit, R. P. Ashley, E. Breedt, A. J. Brown, V. S. Dhillon, M. J. Dyer, M. J. Green, P. Kerry, S. P. Littlefair, D. I. Sahman, T. Shahbaz, J. F. Wild, A. Chakpor, R. Lakhom

Abstract: AR Scorpii (AR Sco) is the only radio-pulsing white dwarf known to date. It shows a broad-band spectrum extending from radio to X-rays whose luminosity cannot be explained by thermal emission from the system components alone, and is instead explained through synchrotron emission powered by the spin-down of the white dwarf. We analysed NTT/ULTRACAM, TNT/ULTRASPEC, and GTC/HiPERCAM high-speed photom… ▽ More AR Scorpii (AR Sco) is the only radio-pulsing white dwarf known to date. It shows a broad-band spectrum extending from radio to X-rays whose luminosity cannot be explained by thermal emission from the system components alone, and is instead explained through synchrotron emission powered by the spin-down of the white dwarf. We analysed NTT/ULTRACAM, TNT/ULTRASPEC, and GTC/HiPERCAM high-speed photometric data for AR Sco spanning almost seven years and obtained a precise estimate of the spin frequency derivative, now confirmed with 50-sigma significance. Using archival photometry, we show that the spin down rate of P/Pdot = 5.6e6 years has remained constant since 2005. As well as employing the method of pulse-arrival time fitting used for previous estimates, we also found a consistent value via traditional Fourier analysis for the first time. In addition, we obtained optical time-resolved spectra with WHT/ISIS and VLT/X-shooter. We performed modulated Doppler tomography for the first time for the system, finding evidence of emission modulated on the orbital period. We have also estimated the projected rotational velocity of the M-dwarf as a function of orbital period and found that it must be close to Roche lobe filling. Our findings provide further constraints for modelling this unique system. △ Less

Submitted 17 August, 2022; originally announced August 2022.

Comments: 15 pages, 18 figures. Accepted for publication in MNRAS

Upside-Down Reinforcement Learning Can Diverge in Stochastic Environments With Episodic Resets

Authors: Miroslav Štrupl, Francesco Faccio, Dylan R. Ashley, Jürgen Schmidhuber, Rupesh Kumar Srivastava

Abstract: Upside-Down Reinforcement Learning (UDRL) is an approach for solving RL problems that does not require value functions and uses only supervised learning, where the targets for given inputs in a dataset do not change over time. Ghosh et al. proved that Goal-Conditional Supervised Learning (GCSL) -- which can be viewed as a simplified version of UDRL -- optimizes a lower bound on goal-reaching perfo… ▽ More Upside-Down Reinforcement Learning (UDRL) is an approach for solving RL problems that does not require value functions and uses only supervised learning, where the targets for given inputs in a dataset do not change over time. Ghosh et al. proved that Goal-Conditional Supervised Learning (GCSL) -- which can be viewed as a simplified version of UDRL -- optimizes a lower bound on goal-reaching performance. This raises expectations that such algorithms may enjoy guaranteed convergence to the optimal policy in arbitrary environments, similar to certain well-known traditional RL algorithms. Here we show that for a specific episodic UDRL algorithm (eUDRL, including GCSL), this is not the case, and give the causes of this limitation. To do so, we first introduce a helpful rewrite of eUDRL as a recursive policy update. This formulation helps to disprove its convergence to the optimal policy for a wide class of stochastic environments. Finally, we provide a concrete example of a very simple environment where eUDRL diverges. Since the primary aim of this paper is to present a negative result, and the best counterexamples are the simplest ones, we restrict all discussions to finite (discrete) environments, ignoring issues of function approximation and limited sample size. △ Less

Submitted 13 May, 2022; originally announced May 2022.

Comments: presented at the 5th Multidisciplinary Conference on Reinforcement Learning and Decision Making; 5 pages in main text + 1 page of references + 3 pages of appendices, 1 figure in main text; source code available at https://github.com/struplm/UDRL-GCSL-counterexample.git

MSC Class: 68T05 ACM Class: I.2.6

Characterising eclipsing white dwarf M dwarf binaries from multi-band eclipse photometry

Authors: Alex J. Brown, Steven G. Parsons, Stuart P. Littlefair, James F. Wild, Richard P. Ashley, Elme Breedt, Vik S. Dhillon, Martin J. Dyer, Matthew J. Green, Paul Kerry, Tom R. Marsh, Ingrid Pelisoli, Dave I. Sahman

Abstract: With the prevalence of wide-field, time-domain photometric sky surveys, the number of eclipsing white dwarf systems being discovered is increasing dramatically. An efficient method to follow these up will be key to determining any population trends and finding any particularly interesting examples. We demonstrate that multi-band eclipse photometry of binaries containing a white dwarf and an M~dwar… ▽ More With the prevalence of wide-field, time-domain photometric sky surveys, the number of eclipsing white dwarf systems being discovered is increasing dramatically. An efficient method to follow these up will be key to determining any population trends and finding any particularly interesting examples. We demonstrate that multi-band eclipse photometry of binaries containing a white dwarf and an M~dwarf can be used to determine the masses and temperatures of the white dwarfs to better than 5 per cent. For the M~dwarfs we measure their parameters to a precision of better than 6 per cent with the uncertainty dominated by the intrinsic scatter of the M~dwarf mass-radius relationship. This precision is better than what can typically be achieved with low-resolution spectroscopy. The nature of this method means that it will be applicable to LSST data in the future, enabling direct characterisation without follow-up spectroscopy. Additionally, we characterise three new post-common-envelope binaries from their eclipse photometry, finding two systems containing hot helium-core white dwarfs with low-mass companions (one near the brown dwarf transition regime) and a possible detached cataclysmic variable at the lower edge of the period gap. △ Less

Submitted 11 April, 2022; originally announced April 2022.

Comments: 10 pages (and appendix), 5 figures, accepted to MNRAS

Learning Relative Return Policies With Upside-Down Reinforcement Learning

Authors: Dylan R. Ashley, Kai Arulkumaran, Jürgen Schmidhuber, Rupesh Kumar Srivastava

Abstract: Lately, there has been a resurgence of interest in using supervised learning to solve reinforcement learning problems. Recent work in this area has largely focused on learning command-conditioned policies. We investigate the potential of one such method -- upside-down reinforcement learning -- to work with commands that specify a desired relationship between some scalar value and the observed retu… ▽ More Lately, there has been a resurgence of interest in using supervised learning to solve reinforcement learning problems. Recent work in this area has largely focused on learning command-conditioned policies. We investigate the potential of one such method -- upside-down reinforcement learning -- to work with commands that specify a desired relationship between some scalar value and the observed return. We show that upside-down reinforcement learning can learn to carry out such commands online in a tabular bandit setting and in CartPole with non-linear function approximation. By doing so, we demonstrate the power of this family of methods and open the way for their practical use under more complicated command structures. △ Less

Submitted 10 May, 2022; v1 submitted 23 February, 2022; originally announced February 2022.

Comments: presented at the 5th Multidisciplinary Conference on Reinforcement Learning and Decision Making; 5 pages in main text, 2 figures in main text

ACM Class: I.2.6

All You Need Is Supervised Learning: From Imitation Learning to Meta-RL With Upside Down RL

Authors: Kai Arulkumaran, Dylan R. Ashley, Jürgen Schmidhuber, Rupesh K. Srivastava

Abstract: Upside down reinforcement learning (UDRL) flips the conventional use of the return in the objective function in RL upside down, by taking returns as input and predicting actions. UDRL is based purely on supervised learning, and bypasses some prominent issues in RL: bootstrap**, off-policy corrections, and discount factors. While previous work with UDRL demonstrated it in a traditional online RL… ▽ More Upside down reinforcement learning (UDRL) flips the conventional use of the return in the objective function in RL upside down, by taking returns as input and predicting actions. UDRL is based purely on supervised learning, and bypasses some prominent issues in RL: bootstrap**, off-policy corrections, and discount factors. While previous work with UDRL demonstrated it in a traditional online RL setting, here we show that this single algorithm can also work in the imitation learning and offline RL settings, be extended to the goal-conditioned RL setting, and even the meta-RL setting. With a general agent architecture, a single UDRL agent can learn across all paradigms. △ Less

Submitted 24 February, 2022; originally announced February 2022.

Automatic Embedding of Stories Into Collections of Independent Media

Authors: Dylan R. Ashley, Vincent Herrmann, Zachary Friggstad, Kory W. Mathewson, Jürgen Schmidhuber

Abstract: We look at how machine learning techniques that derive properties of items in a collection of independent media can be used to automatically embed stories into such collections. To do so, we use models that extract the tempo of songs to make a music playlist follow a narrative arc. Our work specifies an open-source tool that uses pre-trained neural network models to extract the global tempo of a s… ▽ More We look at how machine learning techniques that derive properties of items in a collection of independent media can be used to automatically embed stories into such collections. To do so, we use models that extract the tempo of songs to make a music playlist follow a narrative arc. Our work specifies an open-source tool that uses pre-trained neural network models to extract the global tempo of a set of raw audio files and applies these measures to create a narrative-following playlist. This tool is available at https://github.com/dylanashley/playlist-story-builder/releases/tag/v1.0.0 △ Less

Submitted 3 November, 2021; originally announced November 2021.

Comments: 2 pages in main text + 1 page of references + 6 pages of appendices, 2 figures in main text + 3 figures in appendices, 1 algorithm in appendices; source code available at https://gist.github.com/dylanashley/1387a99deb85bfc0bce11286810cd98b

ACM Class: H.5.5; I.2.6; J.5

The SNIa Runaway LP 398-9: Detection of Circumstellar Material and Surface Rotation

Authors: Vedant Chandra, Hsiang-Chih Hwang, Nadia L. Zakamska, Simon Blouin, Andrew Swan, Thomas R. Marsh, Ken J. Shen, Boris T. Gänsicke, J. J. Hermes, Odelia Putterman, Evan B. Bauer, Evan Petrosky, Vikram S. Dhillon, Stuart P. Littlefair, Richard P. Ashley

Abstract: A promising progenitor scenario for Type Ia supernovae (SNeIa) is the thermonuclear detonation of a white dwarf in a close binary system with another white dwarf. After the primary star explodes, the surviving donor can be spontaneously released as a hypervelocity runaway. One such runaway donor candidate is LP 398-9, whose orbital trajectory traces back $\approx 10^5$ years to a known supernova r… ▽ More A promising progenitor scenario for Type Ia supernovae (SNeIa) is the thermonuclear detonation of a white dwarf in a close binary system with another white dwarf. After the primary star explodes, the surviving donor can be spontaneously released as a hypervelocity runaway. One such runaway donor candidate is LP 398-9, whose orbital trajectory traces back $\approx 10^5$ years to a known supernova remnant. Here we report the discovery of carbon-rich circumstellar material around LP 398-9, revealed by a strong infrared excess and analyzed with follow-up spectroscopy. The circumstellar material is most plausibly composed of inflated layers from the star itself, mechanically and radioactively heated by the past companion's supernova. We also detect a 15.4 hr periodic signal in the UV and optical light curves of LP 398-9, which we interpret as surface rotation. The rotation rate is consistent with theoretical predictions from this supernova mechanism, and the brightness variations could originate from surface inhomogeneity deposited by the supernova itself. Our observations strengthen the case for this double-degenerate SNIa progenitor channel, and motivate the search for more runaway SNIa donors. △ Less

Submitted 13 April, 2022; v1 submitted 13 October, 2021; originally announced October 2021.

Comments: Accepted to MNRAS

ASAS J071404+7004.3 -- a close, bright nova-like cataclysmic variable with gusty winds

Authors: keith Inight, Boris Gaensicke, Dominic Blondel, David Boyd, Richard Ashley, Christian Knigge, Knox Long, Tom Marsh, Jack McCleery, Simone Scaringi, Danny Steeghs, John Thorstensen, Tonny Vanmunster, Peter Wheatley

Abstract: Despite being bright ($V=12$) and nearby ($d=212$ pc) ASAS J071404+7004.3 has only recently been identified as a nova-like cataclysmic variable. We present time-resolved optical spectroscopy obtained at the Isaac Newton Telescope together with $\textit{Swift}$ X-ray and ultraviolet observations. We combined these with $\textit{TESS}$ photometry and find a period of 3.28h and a mass transfer rate o… ▽ More Despite being bright ($V=12$) and nearby ($d=212$ pc) ASAS J071404+7004.3 has only recently been identified as a nova-like cataclysmic variable. We present time-resolved optical spectroscopy obtained at the Isaac Newton Telescope together with $\textit{Swift}$ X-ray and ultraviolet observations. We combined these with $\textit{TESS}$ photometry and find a period of 3.28h and a mass transfer rate of $4-9 \times 10^{-9} M_{sun}/yr$. Historical photometry shows at least one low state establishing the system as a VY Scl star. Our high-cadence spectroscopy also revealed rapidly changing winds emanating from the accretion disc. We have modelled these using the Monte Carlo PYTHON code and shown that all the emission lines could emanate from the wind - which would explain the lack of double-peaked lines in such systems. In passing,we discuss the effect of variability on the position of cataclysmic variables in the $\textit{Gaia}$ Hertzsprung-Russell diagram. △ Less

Submitted 13 December, 2021; v1 submitted 29 September, 2021; originally announced September 2021.

Comments: Accepted for publication by MNRAS. 20 pages, 14 figures

Optical detection of the rapidly spinning white dwarf in V1460 Her

Authors: Ingrid Pelisoli, T. R. Marsh, R. P. Ashley, Pasi Hakala, A. Aungwerojwit, K. Burdge, E. Breedt, A. J. Brown, K. Chanthorn, V. S. Dhillon, M. J. Dyer, M. J. Green, P. Kerry, S. P. Littlefair, S. G. Parsons, D. I. Sahman, J. F. Wild, S. Yotthanathong

Abstract: Accreting magnetic white dwarfs offer an opportunity to understand the interplay between spin-up and spin-down torques in binary systems. Monitoring of the white dwarf spin may reveal whether the white dwarf spin is currently in a state of near-equilibrium, or of uni-directional evolution towards longer or shorter periods, reflecting the recent history of the system and providing constraints for e… ▽ More Accreting magnetic white dwarfs offer an opportunity to understand the interplay between spin-up and spin-down torques in binary systems. Monitoring of the white dwarf spin may reveal whether the white dwarf spin is currently in a state of near-equilibrium, or of uni-directional evolution towards longer or shorter periods, reflecting the recent history of the system and providing constraints for evolutionary models. This makes the monitoring of the spin history of magnetic white dwarfs of high interest. In this paper we report the results of a campaign of follow-up optical photometry to detect and track the 39 sec white dwarf spin pulses recently discovered in Hubble Space Telescope data of the cataclysmic variable V1460 Her. We find the spin pulsations to be present in g-band photometry at a typical amplitude of 0.4%. Under favourable observing conditions, the spin signal is detectable using 2-meter class telescopes. We measured pulse-arrival times for all our observations, which allowed us to derive a precise ephemeris for the white dwarf spin. We have also derived an orbital modulation correction that can be applied to the measurements. With our limited baseline of just over four years, we detect no evidence yet for spin-up or spin-down of the white dwarf, obtaining a lower limit of |P/Pdot|> 4e7 years, which is already 4 to 8 times longer than the timescales measured in two other cataclysmic variable systems containing rapidly rotating white dwarfs, AE Aqr and AR Sco. △ Less

Submitted 1 September, 2021; originally announced September 2021.

Comments: 8 pages, 6 figures. Accepted for publication in MNRAS

High resolution H-alpha imaging of the Northern Galactic Plane, and the IGAPS images database

Authors: R. Greimel, J. E. Drew, M. Monguió, R. P. Ashley, G. Barentsen, J. Eislöffel, A. Mampaso, R. A. H. Morris, T. Naylor, C. Roe, L. Sabin, B. Stecklum, N. J. Wright, P. J. Groot, M. J. Irwin, M. J. Barlow, C. Fariña, A. Fernández-Martín, Q. A. Parker, S. Phillipps, S. Scaringi, A. A. Zijlstra

Abstract: The INT Galactic Plane Survey (IGAPS) is the merger of the optical photometric surveys, IPHAS and UVEX, based on data from the Isaac Newton Telescope (INT) obtained between 2003 and 2018. These capture the entire northern Galactic plane within the Galactic coordinate range, -5<b<+5 deg. and 30<l<215 deg. From the beginning, the incorporation of narrowband H-alpha imaging has been a unique and dist… ▽ More The INT Galactic Plane Survey (IGAPS) is the merger of the optical photometric surveys, IPHAS and UVEX, based on data from the Isaac Newton Telescope (INT) obtained between 2003 and 2018. These capture the entire northern Galactic plane within the Galactic coordinate range, -5<b<+5 deg. and 30<l<215 deg. From the beginning, the incorporation of narrowband H-alpha imaging has been a unique and distinctive feature of this effort. Alongside a focused discussion of the nature and application of the H-alpha data, we present the IGAPS world-accessible database of images for all 5 survey filters, i, r, g, U-RGO and narrowband H-alpha, observed on a pixel scale of 0.33 arcsec and at an effective (median) angular resolution of 1.1 to 1.3 arcsec. The background, noise, and sensitivity characteristics of the narrowband H-alpha filter images are outlined. Typical noise levels in this band correspond to a surface brightness at full one-arcsec resolution of around 2e-16 erg/cm2/s/arcsec2. Illustrative applications of the H-alpha data to planetary nebulae and Herbig-Haro objects are outlined and, as part of a discussion of mosaicking technique, we present a very large background-subtracted narrowband mosaic of the supernova remnant, Simeis 147. Finally we lay out a method that exploits the database via an automated selection of bright ionized diffuse interstellar emission targets for the coming generation of wide-field massive-multiplex spectrographs. Two examples of the diffuse H-alpha maps output from this selection process are presented and compared with previously published data. △ Less

Submitted 27 July, 2021; originally announced July 2021.

Comments: 21 pages, 14 main-text figures, 11 appendix figures. Images database and other supplementary items mentioned in the paper are available from http://www.igapsimages.org

Journal ref: A&A 655, A49 (2021)

HiPERCAM: a quintuple-beam, high-speed optical imager on the 10.4-m Gran Telescopio Canarias

Authors: V. S. Dhillon, N. Bezawada, M. Black, S. D. Dixon, T. Gamble, X. Gao, D. M. Henry, P. Kerry, S. P. Littlefair, D. W. Lunney, T. R. Marsh, C. Miller, S. G. Parsons, R. P. Ashley, E. Breedt, A. Brown, M. J. Dyer, M. J. Green, I. Pelisoli, D. I. Sahman, J. Wild, D. J. Ives, L. Mehrgan, J. Stegmeier, C. M. Dubbeldam , et al. (14 additional authors not shown)

Abstract: HiPERCAM is a portable, quintuple-beam optical imager that saw first light on the 10.4-m Gran Telescopio Canarias (GTC) in 2018. The instrument uses re-imaging optics and 4 dichroic beamsplitters to record $u_s g_s r_s i_s z_s$ ($320-1060$ nm) images simultaneously on its five CCD cameras, each of 3.1 arcmin (diagonal) field of view. The detectors in HiPERCAM are frame-transfer devices cooled ther… ▽ More HiPERCAM is a portable, quintuple-beam optical imager that saw first light on the 10.4-m Gran Telescopio Canarias (GTC) in 2018. The instrument uses re-imaging optics and 4 dichroic beamsplitters to record $u_s g_s r_s i_s z_s$ ($320-1060$ nm) images simultaneously on its five CCD cameras, each of 3.1 arcmin (diagonal) field of view. The detectors in HiPERCAM are frame-transfer devices cooled thermo-electrically to 183 K, thereby allowing both long-exposure, deep imaging of faint targets, as well as high-speed (over 1000 windowed frames per second) imaging of rapidly varying targets. A comparison-star pick-off system in the telescope focal plane increases the effective field of view to 6.7 arcmin for differential photometry. Combining HiPERCAM with the world's largest optical telescope enables the detection of astronomical sources to $g_s \sim 23$ in 1 s and $g_s \sim 28$ in 1 h. In this paper we describe the scientific motivation behind HiPERCAM, present its design, report on its measured performance, and outline some planned enhancements. △ Less

Submitted 21 July, 2021; originally announced July 2021.

Comments: 18 pages, 12 figures, 1 table, accepted for publication in Monthly Notices of the Royal Astronomical Society

Reward-Weighted Regression Converges to a Global Optimum

Authors: Miroslav Štrupl, Francesco Faccio, Dylan R. Ashley, Rupesh Kumar Srivastava, Jürgen Schmidhuber

Abstract: Reward-Weighted Regression (RWR) belongs to a family of widely known iterative Reinforcement Learning algorithms based on the Expectation-Maximization framework. In this family, learning at each iteration consists of sampling a batch of trajectories using the current policy and fitting a new policy to maximize a return-weighted log-likelihood of actions. Although RWR is known to yield monotonic im… ▽ More Reward-Weighted Regression (RWR) belongs to a family of widely known iterative Reinforcement Learning algorithms based on the Expectation-Maximization framework. In this family, learning at each iteration consists of sampling a batch of trajectories using the current policy and fitting a new policy to maximize a return-weighted log-likelihood of actions. Although RWR is known to yield monotonic improvement of the policy under certain circumstances, whether and under which conditions RWR converges to the optimal policy have remained open questions. In this paper, we provide for the first time a proof that RWR converges to a global optimum when no function approximation is used, in a general compact setting. Furthermore, for the simpler case with finite state and action spaces we prove R-linear convergence of the state-value function to the optimum. △ Less

Submitted 23 February, 2022; v1 submitted 19 July, 2021; originally announced July 2021.

Comments: 7 pages in main text + 2 pages of references + 6 pages of appendices, 1 figure in main text + 1 figure in appendices; source code available at https://github.com/dylanashley/reward-weighted-regression

MSC Class: 68T05 ACM Class: I.2.6

System parameters of three short period cataclysmic variable stars

Authors: J. F. Wild, S. P. Littlefair, R. P. Ashley, E. Breedt, A. Brown, V. S. Dhillon, M. J. Dyer, M. J. Green, P. Kerry, T. R. Marsh, S. G. Parsons, D. I. Sahman

Abstract: Using photometric ULTRACAM observations of three new short period cataclysmic variables, we model the primary eclipse lightcurves to extract the orbital separation, masses, and radii of their component stars. We find donor masses of 0.060 +/- 0.008 solar masses, 0.042 +/- 0.001 solar masses, and 0.042 +/- 0.004 solar masses, two being very low-mass sub-stellar donors, and one within 2 sigma of the… ▽ More Using photometric ULTRACAM observations of three new short period cataclysmic variables, we model the primary eclipse lightcurves to extract the orbital separation, masses, and radii of their component stars. We find donor masses of 0.060 +/- 0.008 solar masses, 0.042 +/- 0.001 solar masses, and 0.042 +/- 0.004 solar masses, two being very low-mass sub-stellar donors, and one within 2 sigma of the hydrogen burning limit. All three of the new systems lie close to the modified, "optimal" model evolutionary sequence of Knigge et al. (2011). We briefly re-evaluate the long-standing discrepancy between observed donor mass and radius data, and theoretical CV evolutionary tracks. By looking at the difference in the observed period at each mass and the period predicted by the Knigge et al. (2011) evolutionary sequence, we qualitatively examine the form of excess angular momentum loss that is missing from the models below the period gap. We show indications that the excess angular momentum loss missing from CV models grows in importance relative to gravitational losses as the period decreases. Detailed CV evolutionary models are necessary to draw more quantitative conclusions in the future. △ Less

Submitted 10 November, 2021; v1 submitted 15 July, 2021; originally announced July 2021.

Comments: 17 pages, 14 figures, 5 tables

Does the Adam Optimizer Exacerbate Catastrophic Forgetting?

Authors: Dylan R. Ashley, Sina Ghiassian, Richard S. Sutton

Abstract: Catastrophic forgetting remains a severe hindrance to the broad application of artificial neural networks (ANNs), however, it continues to be a poorly understood phenomenon. Despite the extensive amount of work on catastrophic forgetting, we argue that it is still unclear how exactly the phenomenon should be quantified, and, moreover, to what degree all of the choices we make when designing learni… ▽ More Catastrophic forgetting remains a severe hindrance to the broad application of artificial neural networks (ANNs), however, it continues to be a poorly understood phenomenon. Despite the extensive amount of work on catastrophic forgetting, we argue that it is still unclear how exactly the phenomenon should be quantified, and, moreover, to what degree all of the choices we make when designing learning systems affect the amount of catastrophic forgetting. We use various testbeds from the reinforcement learning and supervised learning literature to (1) provide evidence that the choice of which modern gradient-based optimization algorithm is used to train an ANN has a significant impact on the amount of catastrophic forgetting and show that-surprisingly-in many instances classical algorithms such as vanilla SGD experience less catastrophic forgetting than the more modern algorithms such as Adam. We empirically compare four different existing metrics for quantifying catastrophic forgetting and (2) show that the degree to which the learning systems experience catastrophic forgetting is sufficiently sensitive to the metric used that a change from one principled metric to another is enough to change the conclusions of a study dramatically. Our results suggest that a much more rigorous experimental methodology is required when looking at catastrophic forgetting. Based on our results, we recommend inter-task forgetting in supervised learning must be measured with both retention and relearning metrics concurrently, and intra-task forgetting in reinforcement learning must-at the very least-be measured with pairwise interference. △ Less

Submitted 9 June, 2021; v1 submitted 15 February, 2021; originally announced February 2021.

Comments: 9 pages in main text + 3 pages of references + 16 pages of appendices, 6 figures in main text + 21 figures in appendices, 6 tables in appendices; source code available at https://github.com/dylanashley/catastrophic-forgetting/tree/arxiv

ACM Class: I.2.6

Magnetic white dwarfs in post-common-envelope binaries

Authors: S. G. Parsons, B. T. Gänsicke, M. R. Schreiber, T. R. Marsh, R. P. Ashley, E. Breedt, S. P. Littlefair, H. Meusinger

Abstract: Magnitude-limited samples have shown that 20-25 per cent of cataclysmic variables contain white dwarfs with magnetic fields of Mega Gauss strength, in stark contrast to the approximately 5 per cent of single white dwarfs with similar magnetic field strengths. Moreover, the lack of identifiable progenitor systems for magnetic cataclysmic variables leads to considerable challenges when trying to und… ▽ More Magnitude-limited samples have shown that 20-25 per cent of cataclysmic variables contain white dwarfs with magnetic fields of Mega Gauss strength, in stark contrast to the approximately 5 per cent of single white dwarfs with similar magnetic field strengths. Moreover, the lack of identifiable progenitor systems for magnetic cataclysmic variables leads to considerable challenges when trying to understand how these systems form and evolve. Here we present a sample of six magnetic white dwarfs in detached binaries with low-mass stellar companions where we have constrained the stellar and binary parameters including, for the first time, reliable mass estimates for these magnetic white dwarfs. We find that they are systematically more massive than non-magnetic white dwarfs in detached binaries. These magnetic white dwarfs generally have cooling ages of more than 1 Gyr and reside in systems that are very close to Roche-lobe filling. Our findings are more consistent with these systems being temporarily detached cataclysmic variables, rather than pre-cataclysmic binaries, but we cannot rule out the latter possibility. We find that these systems can display unusual asymmetric light curves that may offer a way to identify them in larger numbers in future. Seven new candidate magnetic white dwarf systems are also presented, three of which have asymmetric light curves. Finally, we note that several newly identified magnetic systems have archival spectra where there is no clear evidence of magnetism, meaning that these binaries have been previously missed. Nevertheless, there remains a clear lack of younger detached magnetic white dwarf systems. △ Less

Submitted 29 January, 2021; v1 submitted 21 January, 2021; originally announced January 2021.

Comments: Accepted for publication in MNRAS

V1460 Her: A fast spinning white dwarf accreting from an evolved donor star

Authors: R. P. Ashley, T. R. Marsh, E. Breedt, B. T. Gaensicke, A. F. Pala, O. Toloza, P. Chote, John R. Thorstensen, M. R. Burleigh

Abstract: We present time-resolved optical and ultraviolet spectroscopy and photometry of V1460~Her, an eclipsing cataclysmic variable with a 4.99\,h orbital period and an overluminous K5-type donor star. The optical spectra show emission lines from an accretion disc along with absorption lines from the donor. We use these to measure radial velocities, which, together with constraints upon the orbital incli… ▽ More We present time-resolved optical and ultraviolet spectroscopy and photometry of V1460~Her, an eclipsing cataclysmic variable with a 4.99\,h orbital period and an overluminous K5-type donor star. The optical spectra show emission lines from an accretion disc along with absorption lines from the donor. We use these to measure radial velocities, which, together with constraints upon the orbital inclination from photometry, imply masses of $M_1=0.869\pm0.006\,\mathrm{M}_\odot$ and $M_2=0.295\pm0.004\,\mathrm{M}_\odot$ for the white dwarf and the donor. The radius of the donor, $R_2=0.43\pm0.002\,\mathrm{R}_\odot$, is $\approx 50$ per cent larger than expected given its mass, while its spectral type is much earlier than the M3.5 type that would be expected from a main sequence star with a similar mass. HST spectra show strong $\mathrm{N{\small V}}$ 1240 A emission but no $\mathrm{C{\small IV}}$ 1550 A emission, evidence for CNO-processed material. The donor is therefore a bloated, over-luminous remnant of a thermal-timescale stage of high mass transfer and has yet to re-establish thermal equilibrium. Remarkably, the HST ultraviolet data also show a strong 30 per cent peak-to-peak, $38.9\,$s pulsation that we explain as being due to the spin of the white dwarf, potentially putting V1460 Her in a similar category to the propeller system AE Aqr in terms of its spin frequency and evolutionary path. AE Aqr also features a post-thermal timescale mass donor, and V1460 Her may therefore be its weak magnetic field analogue since the accretion disc is still present, with the white dwarf spin-up a result of a recent high accretion rate. △ Less

Submitted 30 August, 2020; originally announced August 2020.

Spectroscopic and Photometric Periods of Six Ultracompact Accreting Binaries

Authors: Matthew J. Green, Thomas R. Marsh, Philip J. Carter, Danny Steeghs, Elmé Breedt, V. S. Dhillon, S. P. Littlefair, Steven G. Parsons, Paul Kerry, Nicola P. Gentile Fusillo, R. P. Ashley, Madelon C. P. Bours, Tim Cunningham, Martin J. Dyer, Boris T. Gänsicke, Paula Izquierdo, Anna F. Pala, Chuangwit Pattama, Sabrina Outmani, David I. Sahman, Boonchoo Sukaum, James Wild

Abstract: Ultracompact accreting binary systems each consist of a stellar remnant accreting helium-enriched material from a compact donor star. Such binaries include two related sub-classes, AM CVn-type binaries and helium cataclysmic variables, in both of which the central star is a white dwarf. We present a spectroscopic and photometric study of six accreting binaries with orbital periods in the range of… ▽ More Ultracompact accreting binary systems each consist of a stellar remnant accreting helium-enriched material from a compact donor star. Such binaries include two related sub-classes, AM CVn-type binaries and helium cataclysmic variables, in both of which the central star is a white dwarf. We present a spectroscopic and photometric study of six accreting binaries with orbital periods in the range of 40--70 min, including phase-resolved VLT spectroscopy and high-speed ULTRACAM photometry. Four of these are AM CVn systems and two are helium cataclysmic variables. For four of these binaries we are able to identify orbital periods (of which three are spectroscopic). SDSS J1505+0659 has an orbital period of 67.8 min, significantly longer than previously believed, and longer than any other known AM CVn binary. We identify a WISE infrared excess in SDSS J1505+0659 that we believe to be the first direct detection of an AM CVn donor star in a non-direct impacting binary. The mass ratio of SDSS J1505+0659 is consistent with a white dwarf donor. CRTS J1028-0819 has an orbital period of 52.1 min, the shortest period of any helium cataclysmic variable. MOA 2010-BLG-087 is co-aligned with a K-class star that dominates its spectrum. ASASSN-14ei and ASASSN-14mv both show a remarkable number of echo outbursts following superoutbursts (13 and 10 echo outbursts respectively). ASASSN-14ei shows an increased outburst rate over the years following its superoutburst, perhaps resulting from an increased accretion rate. △ Less

Submitted 26 May, 2020; originally announced May 2020.

Comments: 22 pages, 21 figures. Accepted for publication by MNRAS

Optical spectra of FO Aquarii during low and high accretion rates

Authors: M. R. Kennedy, P. M. Garnavich, C. Littlefield, T. R. Marsh, P. Callanan, R. P. Breton, T. Augusteijn, R. M. Wagner, R. P. Ashley, M. Neric

Abstract: Between May 2016 and September 2018, the intermediate polar (IP) FO Aquarii exhibited two distinct low states and one failed low state. We present optical spectroscopy of FO Aquarii throughout this period, making this the first detailed study of an accretion disc during a low state in any IP. Analysis of these data confirm that the low states are the result of a drop in the mass transfer rate betw… ▽ More Between May 2016 and September 2018, the intermediate polar (IP) FO Aquarii exhibited two distinct low states and one failed low state. We present optical spectroscopy of FO Aquarii throughout this period, making this the first detailed study of an accretion disc during a low state in any IP. Analysis of these data confirm that the low states are the result of a drop in the mass transfer rate between the secondary star and the magnetic white dwarf primary, and are characterised by a decrease in the system's brightness coupled with a change of the system's accretion structures from an accretion disc-fed geometry to a combination of disc-fed and ballistic stream-fed accretion, and that effects from accretion onto both magnetic poles become detectable. The failed low state only displays a decrease in brightness, with the accretion geometry remaining primarily disc-fed. We also find that the WD appears to be exclusively accretion disc-fed during the high state. There is evidence for an outflow close to the impact region between the ballistic stream and the disc which is detectable in all of the states. Finally, there is marginal evidence for narrow high velocity features in the H$α$ emission line during the low states which may arise due to an outflow from the WD. These features may be evidence of a collimated jet, a long predicted yet elusive feature of cataclysmic variables. △ Less

Submitted 19 May, 2020; originally announced May 2020.

Comments: 19 pages, 14 figures, accepted for publication in Monthly Notices of the Royal Astronomical Society

The frequency of gaseous debris discs around white dwarfs

Authors: Christopher J. Manser, Boris T. Gänsicke, Nicola Pietro Gentile Fusillo, Richard Ashley, Elmé Breedt, Mark Hollands, Paula Izquierdo, Ingrid Pelisoli

Abstract: 1-3 per cent of white dwarfs are orbited by planetary dusty debris detectable as infrared emission in excess above the white dwarf flux. In a rare subset of these systems, a gaseous disc component is also detected via emission lines of the Ca II 8600Å triplet, broadened by the Keplerian velocity of the disc. We present the first statistical study of the fraction of debris discs containing detectab… ▽ More 1-3 per cent of white dwarfs are orbited by planetary dusty debris detectable as infrared emission in excess above the white dwarf flux. In a rare subset of these systems, a gaseous disc component is also detected via emission lines of the Ca II 8600Å triplet, broadened by the Keplerian velocity of the disc. We present the first statistical study of the fraction of debris discs containing detectable amounts of gas in emission at white dwarfs within a magnitude and signal-to-noise limited sample. We select 7705 single white dwarfs spectroscopically observed by the Sloan Digital Sky Survey (SDSS) and $Gaia$ with magnitudes $g$ $\leq$ 19. We identify five gaseous disc hosts, all of which have been previously discovered. We calculate the occurrence rate of a white dwarf hosting a debris disc detectable via Ca II emission lines as 0.067$\pm$$_{0.025}^{0.042}$ per cent. This corresponds to an occurrence rate for a dusty debris disc to have an observable gaseous component in emission as 4$\pm$$_{2}^{4}$ per cent. Given that variability is a common feature of the emission profiles of gaseous debris discs, and the recent detection of a planetesimal orbiting within the disc of SDSSJ122859.93+104032.9, we propose that gaseous components are tracers for the presence of planetesimals embedded in the discs and outline a qualitative model. We also present spectroscopy of the Ca II triplet 8600Å region for 20 white dwarfs hosting dusty debris discs in an attempt to identify gaseous emission. We do not detect any gaseous components in these 20 systems, consistent with the occurrence rate that we calculated. △ Less

Submitted 5 February, 2020; originally announced February 2020.

Comments: 13 pages, 6 Figures, accepted for publication in MNRAS

Universal Successor Features for Transfer Reinforcement Learning

Authors: Chen Ma, Dylan R. Ashley, Junfeng Wen, Yoshua Bengio

Abstract: Transfer in Reinforcement Learning (RL) refers to the idea of applying knowledge gained from previous tasks to solving related tasks. Learning a universal value function (Schaul et al., 2015), which generalizes over goals and states, has previously been shown to be useful for transfer. However, successor features are believed to be more suitable than values for transfer (Dayan, 1993; Barreto et al… ▽ More Transfer in Reinforcement Learning (RL) refers to the idea of applying knowledge gained from previous tasks to solving related tasks. Learning a universal value function (Schaul et al., 2015), which generalizes over goals and states, has previously been shown to be useful for transfer. However, successor features are believed to be more suitable than values for transfer (Dayan, 1993; Barreto et al.,2017), even though they cannot directly generalize to new goals. In this paper, we propose (1) Universal Successor Features (USFs) to capture the underlying dynamics of the environment while allowing generalization to unseen goals and (2) a flexible end-to-end model of USFs that can be trained by interacting with the environment. We show that learning USFs is compatible with any RL algorithm that learns state values using a temporal difference method. Our experiments in a simple gridworld and with two MuJoCo environments show that USFs can greatly accelerate training when learning multiple tasks and can effectively transfer knowledge to new tasks. △ Less

Submitted 4 January, 2020; originally announced January 2020.

Fast quasi-periodic oscillations in the eclipsing polar VV Puppis from VLT and XMM-Newton observations

Authors: J. M. Bonnet-Bidaud, M. Mouchet, E. Falize, L. Van Box Som, C. Busschaert, D. A. H. Buckley, H. Breytenbach, T. R. Marsh, R. P. Ashley, V. S. Dhillon

Abstract: We present high time resolution optical photometric data of the polar VV Puppis obtained simultaneously in three filters (u', HeII $λ$4686, r') with the ULTRACAM camera mounted at the ESO-VLT telescope. An analysis of a long 50 ks XMM-Newton observation of the source, retrieved from the database, is also provided. Quasi-periodic oscillations (QPOs) are clearly detected in the optical during the so… ▽ More We present high time resolution optical photometric data of the polar VV Puppis obtained simultaneously in three filters (u', HeII $λ$4686, r') with the ULTRACAM camera mounted at the ESO-VLT telescope. An analysis of a long 50 ks XMM-Newton observation of the source, retrieved from the database, is also provided. Quasi-periodic oscillations (QPOs) are clearly detected in the optical during the source bright phase intervals when the accreting pole is visible, confirming the association of the QPOs with the basis of the accretion column. QPOs are detected in the three filters at a mean frequency of $\sim$ 0.7 Hz with a similar amplitude $\sim$ 1\%. Mean orbitally-averaged power spectra during the bright phase show a rather broad excess with a quality factor Q= $ν$/$Δν$ = 5-7 but smaller data segments commonly show a much higher coherency with Q up to 30. The XMM (0.5--10 keV) observation provides the first accurate estimation of the hard X-ray component with a high kT $\sim$ 40 keV temperature and confirms the high EUV-soft/hard ratio in the range of (4--15) for VV Pup. The detailed X-ray orbital light curve displays a short $Δφ\simeq 0.05$ ingress into self-eclipse of the active pole, indicative of a accretion shock height of $\sim$ 75 km. No significant X-ray QPOs are detected with an amplitude upper limit of $\sim$30\% in the range (0.1--5) Hz. Detailed hydrodynamical numerical simulations of the post-shock accretion region with parameters consistent with VV Pup demonstrate that the expected frequencies from radiative instability are identical for X-rays and optical regime at values $ν$ $\sim$ (40--70) Hz, more than one order magnitude higher than observed. This confirms previous statements suggesting that present instability models are unable to explain the full QPO characteristics within the parameters commonly known for polars. △ Less

Submitted 17 June, 2019; originally announced June 2019.

Comments: 11 pages, 10 figures, accepted for publication in A&A

Journal ref: A&A 633, A145 (2020)

Near-Earth asteroids spectroscopic survey at Isaac Newton Telescope

Authors: M. Popescu, O. Vaduvescu, J. de León, R. M. Gherase, J. Licandro, I. L. Boacă, A. B. Şonka, R. P. Ashley, T. Močnik, D. Morate, M. Predatu, Mário De Prá, C. Fariña, H. Stoev, M. Díaz Alfaro, I. Ordonez-Etxeberria, F. López-Martínez, R. Errmann

Abstract: The population of near-Earth asteroids (NEAs) shows a large variety of objects in terms of physical and dynamical properties. They are subject to planetary encounters and to strong solar wind and radiation effects. Their study is also motivated by practical reasons regarding space exploration and long-term probability of impact with the Earth. We aim to spectrally characterize a significant sample… ▽ More The population of near-Earth asteroids (NEAs) shows a large variety of objects in terms of physical and dynamical properties. They are subject to planetary encounters and to strong solar wind and radiation effects. Their study is also motivated by practical reasons regarding space exploration and long-term probability of impact with the Earth. We aim to spectrally characterize a significant sample of NEAs with sizes in the range of $\sim$0.25 - 5.5 km (categorized as large), and search for connections between their spectral types and the orbital parameters. Optical spectra of NEAs were obtained using the Isaac Newton Telescope (INT) equipped with the IDS spectrograph. These observations are analyzed using taxonomic classification and by comparison with laboratory spectra of meteorites. A total number of 76 NEAs were observed. We classified 44 of them as Q/S-complex, 16 as B/C-complex, eight as V-types, and another eight belong to the remaining taxonomic classes. Our sample contains 27 asteroids categorized as potentially hazardous and 31 possible targets for space missions including (459872) 2014 EK24, (436724) 2011 UW158, and (67367) 2000 LY27. The spectral data corresponding to (276049) 2002 CE26 and (385186) 1994 AW1 shows the 0.7 $μ$m feature which indicates the presence of hydrated minerals on their surface. We report that Q-types have the lowest perihelia (a median value and absolute deviation of $0.797\pm0.244$ AU) and are systematically larger than the S-type asteroids observed in our sample. We explain these observational evidences by thermal fatigue fragmentation as the main process for the rejuvenation of NEA surfaces. In general terms, the taxonomic distribution of our sample is similar to the previous studies and matches the broad groups of the inner main belt asteroids. Nevertheless, we found a wide diversity of spectra compared to the standard taxonomic types. △ Less

Submitted 4 June, 2019; v1 submitted 30 May, 2019; originally announced May 2019.

Comments: Accepted in Astronomy & Astrophysics (A&A)

Journal ref: A&A 627, A124 (2019)

The evolutionary status of Cataclysmic Variables: Eclipse modelling of 15 systems

Authors: M. McAllister, S. P. Littlefair, S. G. Parsons, V. S. Dhillon, T. R. Marsh, B. T. Gaensicke, E. Breedt, C. Copperwheat, M. J. Green, C. Knigge, D. I. Sahman, M. J. Dyer, P. Kerry, R. P. Ashley, P. Irawati, S. Rattanasoon

Abstract: We present measurements of the component masses in 15 Cataclysmic Variables (CVs) - 6 new estimates and 9 improved estimates. We provide new calibrations of the relationship between superhump period excess and mass ratio, and use this relation to estimate donor star masses for 225 superhum** CVs. With an increased sample of donor masses we revisit the implications for CV evolution. We confirm th… ▽ More We present measurements of the component masses in 15 Cataclysmic Variables (CVs) - 6 new estimates and 9 improved estimates. We provide new calibrations of the relationship between superhump period excess and mass ratio, and use this relation to estimate donor star masses for 225 superhum** CVs. With an increased sample of donor masses we revisit the implications for CV evolution. We confirm the high mass of white dwarfs in CVs, but find no trend in white dwarf mass with orbital period. We argue for a revision in the location of the orbital period minimum of CVs to $79.6 \pm 0.2$ min, significantly shorter than previous estimates. We find that CV donors below the gap have an intrinsic scatter of only 0.005 R$_{\odot}$ around a common evolutionary track, implying a correspondingly small variation in angular momentum loss rates. In contrast to prior studies, we find that standard CV evolutionary tracks - without additional angular momentum loss - are a reasonable fit to the donor masses just below the period gap, but that they do not reproduce the observed period minimum, or fit the donor radii below 0.1 M$_{\odot}$. △ Less

Submitted 3 April, 2019; originally announced April 2019.

Comments: 27 pages, 15 figures, accepted for publication in MNRAS

Phase-Resolved Spectroscopy of Gaia14aae: Line Emission From Near the White Dwarf Surface

Authors: M. J. Green, T. R. Marsh, D. Steeghs, E. Breedt, T. Kupfer, P. Rodriguez-Gil, J. van Roestel, R. P. Ashley, L. Wang, E. Cukanovaite, S. Outmani

Abstract: AM CVn binaries are a class of ultracompact, hydrogen-deficient binaries, each consisting of a white dwarf accreting helium-dominated material from a degenerate or semi-degenerate donor star. Of the 56 known systems, only Gaia14aae undergoes complete eclipses of its central white dwarf, allowing the parameters of its stellar components to be tightly constrained. Here, we present phase-resolved opt… ▽ More AM CVn binaries are a class of ultracompact, hydrogen-deficient binaries, each consisting of a white dwarf accreting helium-dominated material from a degenerate or semi-degenerate donor star. Of the 56 known systems, only Gaia14aae undergoes complete eclipses of its central white dwarf, allowing the parameters of its stellar components to be tightly constrained. Here, we present phase-resolved optical spectroscopy of Gaia14aae. We use the spectra to test the assumption that the narrow emission feature known as the `central spike' traces the motion of the central white dwarf. We measure a central spike velocity amplitude of $13.8 \pm 3.2$ km/s, which agrees at the 1 $σ$ level with the predicted value of $17.6 \pm 1.0$ km/s based on eclipse-derived system parameters. The orbital phase offset of the central spike from its expected position is $4 \pm 15$ $^\circ$, consistent with 0 $^\circ$. Doppler maps of the He I lines in Gaia14aae show two accretion disc bright spots, as seen in many AM CVn systems. The formation mechanism for the second spot remains unclear. We detect no hydrogen in the system, but we estimate a 3 $σ$ limit on H$α$ emission with an equivalent width of -1.14 Å. Our detection of nitrogen and oxygen with no corresponding detection of carbon, in conjunction with evidence from recent studies, mildly favours a formation channel in which Gaia14aae is descended from a cataclysmic variable with a significantly evolved donor. △ Less

Submitted 4 February, 2019; originally announced February 2019.

Comments: 16 pages, accepted by MNRAS

Evidence for bimodal orbital separations of white dwarf-red dwarf binary stars

Authors: R. P. Ashley, J. Farihi, T. R. Marsh, D. J. Wilson, B. T. Gaensicke

Abstract: We present the results of a radial velocity survey of 20 white dwarf plus M dwarf binaries selected as a follow up to a \textit{Hubble Space Telescope} study that aimed to spatially resolve suspected binaries. Our candidates are taken from the list of targets that were spatially unresolved with \textit{Hubble}. We have determined the orbital periods for 16 of these compact binary candidates. The p… ▽ More We present the results of a radial velocity survey of 20 white dwarf plus M dwarf binaries selected as a follow up to a \textit{Hubble Space Telescope} study that aimed to spatially resolve suspected binaries. Our candidates are taken from the list of targets that were spatially unresolved with \textit{Hubble}. We have determined the orbital periods for 16 of these compact binary candidates. The period distribution ranges from 0.14 to 9.16\,d and peaks near 0.6\,d. The original sample therefore contains two sets of binaries, wide orbits ($\approx100-1000$\,au) and close orbits ($\lesssim1-10$\,au), with no systems found in the $\approx10-100$\,au range. This observational evidence confirms the bimodal distribution predicted by population models and is also similar to results obtained in previous studies. We find no binary periods in the months to years range, supporting the post common envelope evolution scenario. One of our targets, WD\,1504+546, was discovered to be an eclipsing binary with a period of 0.93\,d. △ Less

Submitted 25 January, 2019; originally announced January 2019.

The scatter of the M dwarf mass-radius relationship

Authors: S. G. Parsons, B. T. Gänsicke, T. R. Marsh, R. P. Ashley, E. Breedt, M. R. Burleigh, C. M. Copperwheat, V. S. Dhillon, M. J. Green, J. J. Hermes, P. Irawati, P. Kerry, S. P. Littlefair, A. Rebassa-Mansergas, D. I. Sahman, M. R. Schreiber, M. Zorotovic

Abstract: M dwarfs are prime targets in the hunt for habitable worlds around other stars. This is due to their abundance as well as their small radii and low masses and temperatures, which facilitate the detection of temperate, rocky planets in orbit around them. However, the fundamental properties of M dwarfs are difficult to constrain, often limiting our ability to characterise the planets they host. Here… ▽ More M dwarfs are prime targets in the hunt for habitable worlds around other stars. This is due to their abundance as well as their small radii and low masses and temperatures, which facilitate the detection of temperate, rocky planets in orbit around them. However, the fundamental properties of M dwarfs are difficult to constrain, often limiting our ability to characterise the planets they host. Here we test several theoretical relationships for M dwarfs by measuring 23 high precision, model-independent masses and radii for M dwarfs in binaries with white dwarfs. We find a large scatter in the radii of these low-mass stars, with 25 per cent having radii consistent with theoretical models while the rest are up to 12 per cent over-inflated. This scatter is seen in both partially- and fully-convective M dwarfs. No clear trend is seen between the over-inflation and age or metallicity, but there are indications that the radii of slowly rotating M dwarfs are more consistent with predictions, albeit with a similar amount of scatter in the measurements compared to more rapidly rotating M dwarfs. The sample of M dwarfs in close binaries with white dwarfs appears indistinguishable from other M dwarf samples, implying that common envelope evolution has a negligible impact on their structure. We conclude that theoretical and empirical mass-radius relationships lack the precision and accuracy required to measure the fundamental parameters of M dwarfs well enough to determine the internal structure and bulk composition of the planets they host. △ Less

Submitted 23 August, 2018; originally announced August 2018.

Comments: 12 pages, 10 figures, accepted for publication in MNRAS

A 15.7-Minute AM CVn Binary Discovered in K2

Authors: M. J. Green, J. J. Hermes, T. R. Marsh, D. T. H. Steeghs, Keaton J. Bell, S. P. Littlefair, S. G. Parsons, E. Dennihy, J. T. Fuchs, J. S. Reding, B. C. Kaiser, R. P. Ashley, E. Breedt, V. S. Dhillon, N. P. Gentile Fusillo, P. Kerry, D. I. Sahman

Abstract: We present the discovery of SDSS J135154.46-064309.0, a short-period variable observed using 30-minute cadence photometry in K2 Campaign 6. Follow-up spectroscopy and high-speed photometry support a classification as a new member of the rare class of ultracompact accreting binaries known as AM CVn stars. The spectroscopic orbital period of $15.65 \pm 0.12$\,minutes makes this system the fourth-sho… ▽ More We present the discovery of SDSS J135154.46-064309.0, a short-period variable observed using 30-minute cadence photometry in K2 Campaign 6. Follow-up spectroscopy and high-speed photometry support a classification as a new member of the rare class of ultracompact accreting binaries known as AM CVn stars. The spectroscopic orbital period of $15.65 \pm 0.12$\,minutes makes this system the fourth-shortest period AM CVn known, and the second system of this type to be discovered by the Kepler spacecraft. The K2 data show photometric periods at $15.7306 \pm 0.0003$\,minutes, $16.1121 \pm 0.0004$\,minutes and $664.82 \pm 0.06$\,minutes, which we identify as the orbital period, superhump period, and disc precession period, respectively. From the superhump and orbital periods we estimate the binary mass ratio $q = M_2/M_1 = 0.111 \pm 0.005$, though this method of mass ratio determination may not be well calibrated for helium-dominated binaries. This system is likely to be a bright foreground source of gravitational waves in the frequency range detectable by LISA, and may be of use as a calibration source if future studies are able to constrain the masses of its stellar components. △ Less

Submitted 19 April, 2018; originally announced April 2018.

Comments: 13 pages, 11 figures

Towards an improvement in the spectral description of central stars of planetary nebulae

Authors: W. Weidmann, R. Gamen, D. Mast, C. Fariña, G. Gimeno, E. O. Schmidt, R. P. Ashley, L. Peralta de Arriba, P. Sowicka, I. Ordonez-Etxeberria

Abstract: Context. There are more than 3000 known Galactic planetary nebulae (PNe), but only 492 central stars of Galactic planetary nebulae (CSPN) have known spectral types. It is vital to increase this number in order to have reliable statistics, which will lead to an increase of our understanding of these amazing objects. Aims. We aim to contribute to the knowledge of central stars of planetary nebulae… ▽ More Context. There are more than 3000 known Galactic planetary nebulae (PNe), but only 492 central stars of Galactic planetary nebulae (CSPN) have known spectral types. It is vital to increase this number in order to have reliable statistics, which will lead to an increase of our understanding of these amazing objects. Aims. We aim to contribute to the knowledge of central stars of planetary nebulae and stellar evolution. Methods. This observational study is based on Gemini Multi-Object Spectrographs (GMOS) and with the Intermediate Dispersion Spectrograph (IDS) at the Isaac Newton Telescope (INT) spectra of 78 CSPN. The objects were selected because they did not have any previous classification, or the present classification is ambiguous. These new high quality spectra allowed us to identify the key stellar lines for determining spectral classification in the Morgan-Keenan (MK) system. Results. We have acquired optical spectra of a large sample of CSPN. From the observed targets, 50 are classified here for the first time while for 28 the existing classifications have been improved. In seven objects we have identified a P-Cygni profile at the He i lines. Six of these CSPN are late O-type. The vast majority of the stars in the sample exhibit an absorption-type spectrum, and in one case we have found wide emission lines typical of [WR] stars. We give a complementary, and preliminary, classification criterion to obtain the sub-type of the O(H)-type CSPN. Finally, we give a more realistic value of the proportion of CSPN that are rich or poor in hydrogen. △ Less

Submitted 6 February, 2018; originally announced February 2018.

Comments: Accepted to be published in the A&A

Journal ref: A&A 614, A135 (2018)

High-Speed Photometry of Gaia14aae: An Eclipsing AM CVn That Challenges Formation Models

Authors: M. J. Green, T. R. Marsh, D. T. H. Steeghs, T. Kupfer, R. P. Ashley, S. Bloemen, E. Breedt, H. C. Campbell, A. Chakpor, C. M. Copperwheat, V. S. Dhillon, G. Hallinan, L. K. Hardy, J. J. Hermes, P. Kerry, S. P. Littlefair, J. Milburn, S. G. Parsons, N. Prasert, J. van Roestel, D. I. Sahman, N. Singh

Abstract: AM CVn-type systems are ultra-compact, hydrogen-deficient accreting binaries with degenerate or semi-degenerate donors. The evolutionary history of these systems can be explored by constraining the properties of their donor stars. We present high-speed photometry of Gaia14aae, an AM CVn with a binary period of 49.7 minutes and the first AM CVn in which the central white dwarf is fully eclipsed by… ▽ More AM CVn-type systems are ultra-compact, hydrogen-deficient accreting binaries with degenerate or semi-degenerate donors. The evolutionary history of these systems can be explored by constraining the properties of their donor stars. We present high-speed photometry of Gaia14aae, an AM CVn with a binary period of 49.7 minutes and the first AM CVn in which the central white dwarf is fully eclipsed by the donor star. Modelling of the lightcurves of this system allows for the most precise measurement to date of the donor mass of an AM CVn, and relies only on geometric and well-tested physical assumptions. We find a mass ratio $q = M_2/M_1 = 0.0287 \pm 0.0020$ and masses $M_1 = 0.87 \pm 0.02 M_\odot$ and $M_2 = 0.0250 \pm 0.0013 M_\odot$. We compare these properties to the three proposed channels for AM CVn formation. Our measured donor mass and radius do not fit with the contraction that is predicted for AM CVn donors descended from white dwarfs or helium stars at long orbital periods. The donor properties we measure fall in a region of parameter space in which systems evolved from hydrogen-dominated cataclysmic variables are expected, but such systems should show spectroscopic hydrogen, which is not seen in Gaia14aae. The evolutionary history of this system is therefore not clear. We consider a helium-burning star or an evolved cataclysmic variable to be the most likely progenitors, but both models require additional processes and/or fine-tuning to fit the data. Additionally, we calculate an updated ephemeris which corrects for an anomalous time measurement in the previously published ephemeris. △ Less

Submitted 1 February, 2018; originally announced February 2018.

Comments: 18 pages, 13 figures, accepted for publication by MNRAS

Directly Estimating the Variance of the λ-Return Using Temporal-Difference Methods

Authors: Craig Sherstan, Brendan Bennett, Kenny Young, Dylan R. Ashley, Adam White, Martha White, Richard S. Sutton

Abstract: This paper investigates estimating the variance of a temporal-difference learning agent's update target. Most reinforcement learning methods use an estimate of the value function, which captures how good it is for the agent to be in a particular state and is mathematically expressed as the expected sum of discounted future rewards (called the return). These values can be straightforwardly estimate… ▽ More This paper investigates estimating the variance of a temporal-difference learning agent's update target. Most reinforcement learning methods use an estimate of the value function, which captures how good it is for the agent to be in a particular state and is mathematically expressed as the expected sum of discounted future rewards (called the return). These values can be straightforwardly estimated by averaging batches of returns using Monte Carlo methods. However, if we wish to update the agent's value estimates during learning--before terminal outcomes are observed--we must use a different estimation target called the λ-return, which truncates the return with the agent's own estimate of the value function. Temporal difference learning methods estimate the expected λ-return for each state, allowing these methods to update online and incrementally, and in most cases achieve better generalization error and faster learning than Monte Carlo methods. Naturally one could attempt to estimate higher-order moments of the λ-return. This paper is about estimating the variance of the λ-return. Prior work has shown that given estimates of the variance of the λ-return, learning systems can be constructed to (1) mitigate risk in action selection, and (2) automatically adapt the parameters of the learning process itself to improve performance. Unfortunately, existing methods for estimating the variance of the λ-return are complex and not well understood empirically. We contribute a method for estimating the variance of the λ-return directly using policy evaluation methods from reinforcement learning. Our approach is significantly simpler than prior methods that independently estimate the second moment of the λ-return. Empirically our new approach behaves at least as well as existing approaches, but is generally more robust. △ Less

Submitted 14 February, 2018; v1 submitted 25 January, 2018; originally announced January 2018.

280 one-opposition near-Earth asteroids recovered by the EURONEAR with the Isaac Newton Telescope

Authors: O. Vaduvescu, L. Hudin, T. Mocnik, F. Char, A. Sonka, V. Tudor, I. Ordonez-Etxeberria, M. Diaz Alfaro, R. Ashley, R. Errmann, P. Short, A. Moloceniuc, R. Cornea, V. Inceu, D. Zavoianu, M. Popescu, L. Curelaru, S. Mihalea, A. -M. Stoian, A. Boldea, R. Toma, L. Fields, V. Grigore, H. Stoev, F. Lopez-Martinez , et al. (58 additional authors not shown)

Abstract: One-opposition near-Earth asteroids (NEAs) are growing in number, and they must be recovered to prevent loss and mismatch risk, and to improve their orbits, as they are likely to be too faint for detection in shallow surveys at future apparitions. We aimed to recover more than half of the one-opposition NEAs recommended for observations by the Minor Planet Center (MPC) using the Isaac Newton Teles… ▽ More One-opposition near-Earth asteroids (NEAs) are growing in number, and they must be recovered to prevent loss and mismatch risk, and to improve their orbits, as they are likely to be too faint for detection in shallow surveys at future apparitions. We aimed to recover more than half of the one-opposition NEAs recommended for observations by the Minor Planet Center (MPC) using the Isaac Newton Telescope (INT) in soft-override mode and some fractions of available D-nights. During about 130 hours in total between 2013 and 2016, we targeted 368 NEAs, among which 56 potentially hazardous asteroids (PHAs), observing 437 INT Wide Field Camera (WFC) fields and recovering 280 NEAs (76% of all targets). Engaging a core team of about ten students and amateurs, we used the THELI, Astrometrica, and the Find_Orb software to identify all moving objects using the blink and track-and-stack method for the faintest targets and plotting the positional uncertainty ellipse from NEODyS. Most targets and recovered objects had apparent magnitudes centered around V~22.8 mag, with some becoming as faint as V~24 mag. One hundred and three objects (representing 28% of all targets) were recovered by EURONEAR alone by Aug 2017. Orbital arcs were prolonged typically from a few weeks to a few years; our oldest recoveries reach 16 years. The O-C residuals for our 1,854 NEA astrometric positions show that most measurements cluster closely around the origin. In addition to the recovered NEAs, 22,000 positions of about 3,500 known minor planets and another 10,000 observations of about 1,500 unknown objects (mostly main-belt objects) were promptly reported to the MPC by our team. Four new NEAs were discovered serendipitously in the analyzed fields, increasing the counting to nine NEAs discovered by the EURONEAR in 2014 and 2015. △ Less

Submitted 3 November, 2017; v1 submitted 2 November, 2017; originally announced November 2017.

Comments: Accepted for publication in Astronomy and Astrophysics (11 Oct 2017). Version 2 adding two co-authors and fixing the affiliation page overflow

Journal ref: A&A 609, A105 (2018)

A search for optical bursts from the repeating fast radio burst FRB 121102

Authors: Liam K Hardy, Vik S Dhillon, Laura G Spitler, Stuart P Littlefair, Richard P Ashley, Annalisa De Cia, Matthew J Green, Phrudth Jaroenjittichai, Evan F Keane, Paul Kerry, Michael Kramer, Daniele Malesani, Tom R Marsh, Steven G Parsons, Andrea Possenti, Somsawat Rattanasoon, David I Sahman

Abstract: We present a search for optical bursts from the repeating fast radio burst FRB 121102 using simultaneous observations with the high-speed optical camera ULTRASPEC on the 2.4-m Thai National Telescope and radio observations with the 100-m Effelsberg Radio Telescope. A total of 13 radio bursts were detected, but we found no evidence for corresponding optical bursts in our 70.7-ms frames. The 5-sigma… ▽ More We present a search for optical bursts from the repeating fast radio burst FRB 121102 using simultaneous observations with the high-speed optical camera ULTRASPEC on the 2.4-m Thai National Telescope and radio observations with the 100-m Effelsberg Radio Telescope. A total of 13 radio bursts were detected, but we found no evidence for corresponding optical bursts in our 70.7-ms frames. The 5-sigma upper limit to the optical flux density during our observations is 0.33 mJy at 767nm. This gives an upper limit for the optical burst fluence of 0.046 Jy ms, which constrains the broadband spectral index of the burst emission to alpha < -0.2. Two of the radio pulses are separated by just 34 ms, which may represent an upper limit on a possible underlying periodicity (a rotation period typical of pulsars), or these pulses may have come from a single emission window that is a small fraction of a possible period. △ Less

Submitted 21 August, 2017; originally announced August 2017.

Comments: 8 pages, 4 figures, accepted for publication in MNRAS

Testing the white dwarf mass-radius relationship with eclipsing binaries

Authors: S. G. Parsons, B. T. Gänsicke, T. R. Marsh, R. P. Ashley, M. C. P. Bours, E. Breedt, M. R. Burleigh, C. M. Copperwheat, V. S. Dhillon, M. Green, L. K. Hardy, J. J. Hermes, P. Irawati, P. Kerry, S. P. Littlefair, M. J. McAllister, S. Rattanasoon, A. Rebassa-Mansergas, D. I. Sahman, M. R. Schreiber

Abstract: We present high precision, model independent, mass and radius measurements for 16 white dwarfs in detached eclipsing binaries and combine these with previously published data to test the theoretical white dwarf mass-radius relationship. We reach a mean precision of 2.4 per cent in mass and 2.7 per cent in radius, with our best measurements reaching a precision of 0.3 per cent in mass and 0.5 per c… ▽ More We present high precision, model independent, mass and radius measurements for 16 white dwarfs in detached eclipsing binaries and combine these with previously published data to test the theoretical white dwarf mass-radius relationship. We reach a mean precision of 2.4 per cent in mass and 2.7 per cent in radius, with our best measurements reaching a precision of 0.3 per cent in mass and 0.5 per cent in radius. We find excellent agreement between the measured and predicted radii across a wide range of masses and temperatures. We also find the radii of all white dwarfs with masses less than 0.48M$_\odot$ to be fully consistent with helium core models, but they are on average 9 per cent larger than those of carbon-oxygen core models. In contrast, white dwarfs with masses larger than 0.52M$_\odot$ all have radii consistent with carbon-oxygen core models. Moreover, we find that all but one of the white dwarfs in our sample have radii consistent with possessing thick surface hydrogen envelopes ($10^{-5} \ge M_\mathrm{H}/M_\mathrm{WD} \ge 10^{-4}$), implying that the surface hydrogen layers of these white dwarfs are not obviously affected by common envelope evolution. △ Less

Submitted 15 June, 2017; originally announced June 2017.

Comments: Accepted for publication in MNRAS

The molecular environment of the pillar-like features in the HII region G46.5-0.2

Authors: S. Paron, M. Celis Peña, M. E. Ortega, C. Fariña, M. Rubio, R. P. Ashley

Abstract: At the interface of HII regions and molecular gas peculiar structures appear, some of them with pillar-like shapes. Understanding their origin is important for characterizing triggered star formation and the impact of massive stars on the interstellar medium. In order to study the molecular environment and the influence of the radiation on two pillar-like features related to the HII region G46.5-0… ▽ More At the interface of HII regions and molecular gas peculiar structures appear, some of them with pillar-like shapes. Understanding their origin is important for characterizing triggered star formation and the impact of massive stars on the interstellar medium. In order to study the molecular environment and the influence of the radiation on two pillar-like features related to the HII region G46.5-0.2, we performed molecular line observations with the Atacama Submillimeter Telescope Experiment, and spectroscopic optical observations with the Isaac Newton Telescope. From the optical observations we identified the star that is exciting the HII region as a spectral type O4-6. The molecular data allowed us to study the structure of the pillars and a HCO+ cloud lying between them. In this HCO+ cloud, which have not any well defined 12CO counterpart, we found direct evidence of star formation: two molecular outflows and two associated near-IR nebulosities. The outflows axis orientation is perpendicular to the direction of the radiation flow from the HII region. Several Class I sources are also embedded in this HCO+ cloud, showing that it is usual that the YSOs form large associations occupying a cavity bounded by pillars. On the other hand, it was confirmed that the RDI process is not occurring in one of the pillar tips. △ Less

Submitted 13 June, 2017; originally announced June 2017.

Comments: Accepted in MNRAS (2017 June 13)

Two white dwarfs in ultrashort binaries with detached, eclipsing, likely substellar companions detected by K2

Authors: S. G. Parsons, J. J. Hermes, T. R. Marsh, B. T. Gänsicke, P. -E. Tremblay, S. P. Littlefair, D. I. Sahman, R. P. Ashley, M. Green, S. Rattanasoon, V. S. Dhillon, M. R. Burleigh, S. L. Casewell, D. A. H. Buckley, I. P. Braker, P. Irawati, E. Dennihy, P. Rodríguez-Gil, D. E. Winget, K. I. Winget, K. J. Bell, M. Kilic

Abstract: Using data from the extended Kepler mission in K2 Campaign 10 we identify two eclipsing binaries containing white dwarfs with cool companions that have extremely short orbital periods of only 71.2 min (SDSS J1205-0242, a.k.a. EPIC 201283111) and 72.5 min (SDSS J1231+0041, a.k.a. EPIC 248368963). Despite their short periods, both systems are detached with small, low-mass companions, in one case a b… ▽ More Using data from the extended Kepler mission in K2 Campaign 10 we identify two eclipsing binaries containing white dwarfs with cool companions that have extremely short orbital periods of only 71.2 min (SDSS J1205-0242, a.k.a. EPIC 201283111) and 72.5 min (SDSS J1231+0041, a.k.a. EPIC 248368963). Despite their short periods, both systems are detached with small, low-mass companions, in one case a brown dwarf, and the other case either a brown dwarf or a low-mass star. We present follow-up photometry and spectroscopy of both binaries, as well as phase-resolved spectroscopy of the brighter system, and use these data to place preliminary estimates on the physical and binary parameters. SDSS J1205-0242 is composed of a $0.39\pm0.02$M$_\odot$ helium-core white dwarf which is totally eclipsed by a $0.049\pm0.006$M$_\odot$ ($51\pm6$M$_J$) brown dwarf companion, while SDSS J1231+0041 is composed of a $0.56\pm0.07$M$_\odot$ white dwarf which is partially eclipsed by a companion of mass $\lesssim 0.095$M$_\odot$. In the case of SDSS J1205-0242 we look at the combined constraints from common-envelope evolution and brown dwarf models; the system is compatible with similar constraints from other post common-envelope binaries given the current parameter uncertainties, but has potential for future refinement. △ Less

Submitted 26 June, 2017; v1 submitted 16 May, 2017; originally announced May 2017.

Comments: Accepted for publication in MNRAS

An artificial neural network to discover Hypervelocity stars: Candidates in Gaia DR1/TGAS

Authors: T. Marchetti, E. M. Rossi, G. Kordopatis, A. G. A. Brown, A. Rimoldi, E. Starkenburg, K. Youakim, R. Ashley

Abstract: The paucity of hypervelocity stars (HVSs) known to date has severely hampered their potential to investigate the stellar population of the Galactic Centre and the Galactic Potential. The first Gaia data release gives an opportunity to increase the current sample. The challenge is the disparity between the expected number of hypervelocity stars and that of bound background stars. We have applied a… ▽ More The paucity of hypervelocity stars (HVSs) known to date has severely hampered their potential to investigate the stellar population of the Galactic Centre and the Galactic Potential. The first Gaia data release gives an opportunity to increase the current sample. The challenge is the disparity between the expected number of hypervelocity stars and that of bound background stars. We have applied a novel data mining algorithm based on machine learning techniques, an artificial neural network, to the Tycho-Gaia astrometric solution (TGAS) catalogue. With no pre-selection of data, we could exclude immediately $\sim 99 \%$ of the stars in the catalogue and find 80 candidates with more than $90\%$ predicted probability to be HVSs, based only on their position, proper motions, and parallax. We have cross-checked our findings with other spectroscopic surveys, determining radial velocities for 30 and spectroscopic distances for 5 candidates. In addition, follow-up observations have been carried out at the Isaac Newton Telescope for 22 stars, for which we obtained radial velocities and distance estimates. We discover 14 stars with a total velocity in the Galactic rest frame > 400 km/s, and 5 of these have a probability $>50\%$ of being unbound from the Milky Way. Tracing back their orbits in different Galactic potential models we find one possible unbound HVS with velocity $\sim$ 520 km/s, 5 bound HVSs, and, notably, 5 runaway stars with median velocity between 400 and 780 km/s. At the moment, uncertainties in the distance estimates and ages are too large to confirm the nature of our candidates by narrowing down their ejection location, and we wait for future Gaia releases to validate the quality of our sample. This test successfully demonstrates the feasibility of our new data mining routine. △ Less

Submitted 29 May, 2017; v1 submitted 26 April, 2017; originally announced April 2017.

Comments: Published in MNRAS, 17 pages, 10 figures

A VLT-ULTRACAM study of the fast optical quasi-periodic oscillations in the polar V834 Centauri

Authors: M. Mouchet, J. -M. Bonnet-Bidaud, L. Van Box Som, E. Falize, D. A. H. Buckley, J. B. Breytenbach, R. P. Ashley, T. R. Marsh, V. S. Dhillon

Abstract: Quasi-periodic oscillations (QPOs) of a few seconds have been detected in some Polars, the synchronised subclass of cataclysmic systems containing a strongly magnetised white dwarf (WD) which accretes matter from a red dwarf companion. The QPOs are thought to be related to instabilities of a shock formed in the accretion column, close to the WD photosphere above the impact region. We present optic… ▽ More Quasi-periodic oscillations (QPOs) of a few seconds have been detected in some Polars, the synchronised subclass of cataclysmic systems containing a strongly magnetised white dwarf (WD) which accretes matter from a red dwarf companion. The QPOs are thought to be related to instabilities of a shock formed in the accretion column, close to the WD photosphere above the impact region. We present optical observations of the polar V834 Centauri performed with the fast ULTRACAM camera mounted on the ESO-VLT simultaneously in three filters (u', He II 4686A, r') to study these oscillations and characterise their properties along the orbit. Fast Fourier transforms and wavelet analysis have been performed and the mean frequency, rms amplitude, and coherence of the QPOs are derived; a detailed inspection of individual pulses has also been performed. The observations confirm the probable ubiquity of the QPOs for this source at all epochs when the source is in a high state, with observed mean amplitude of 2.1 percent (r'), 1.5 percent (He II), and 0.6 percent (u'). Trains of oscillations are clearly observed in the r' light curve and can be mimicked by a superposition of damped sinusoids with various parameters. The QPO energy distribution is comparable to that of the cyclotron flux, consistent for the r' and He II filters but requiring a significant dilution in the u' filter. New 1D hydrodynamical simulations of shock instabilities, adapted to the physical parameters of V834 Cen, can account for the optical QPO amplitude and X-ray upper limit assuming a cross section of the accretion column in the range (4-5) E14 cm2. However, the predicted frequency is larger than the observed one by an order of magnitude. This shortcoming indicates that the QPO generation is more complex than that produced in a homogeneous column and calls for a more realistic 3D treatment of the accretion flow in future modelling. △ Less

Submitted 10 January, 2017; originally announced January 2017.

Comments: 14 pages, 15 figures, accepted for publication in A&A

Journal ref: A&A 600, A53 (2017)

The Astropy Problem

Authors: Demitri Muna, Michael Alexander, Alice Allen, Richard Ashley, Daniel Asmus, Ruyman Azzollini, Michele Bannister, Rachael Beaton, Andrew Benson, G. Bruce Berriman, Maciej Bilicki, Peter Boyce, Joanna Bridge, Jan Cami, Eryn Cangi, Xian Chen, Nicholas Christiny, Christopher Clark, Michelle Collins, Johan Comparat, Neil Cook, Darren Croton, Isak Delberth Davids, Éric Depagne, John Donor , et al. (129 additional authors not shown)

Abstract: The Astropy Project (http://astropy.org) is, in its own words, "a community effort to develop a single core package for Astronomy in Python and foster interoperability between Python astronomy packages." For five years this project has been managed, written, and operated as a grassroots, self-organized, almost entirely volunteer effort while the software is used by the majority of the astronomical… ▽ More The Astropy Project (http://astropy.org) is, in its own words, "a community effort to develop a single core package for Astronomy in Python and foster interoperability between Python astronomy packages." For five years this project has been managed, written, and operated as a grassroots, self-organized, almost entirely volunteer effort while the software is used by the majority of the astronomical community. Despite this, the project has always been and remains to this day effectively unfunded. Further, contributors receive little or no formal recognition for creating and supporting what is now critical software. This paper explores the problem in detail, outlines possible solutions to correct this, and presents a few suggestions on how to address the sustainability of general purpose astronomical software. △ Less

Submitted 10 October, 2016; originally announced October 2016.

Using Gaussian processes to model light curves in the presence of flickering: the eclipsing cataclysmic variable ASASSN-14ag

Authors: M. J. McAllister, S. P. Littlefair, V. S. Dhillon, T. R. Marsh, R. P. Ashley, M. C. P. Bours, E. Breedt, L. K. Hardy, J. J. Hermes, S. Kengkriangkrai, P. Kerry, S. Rattanasoon, D. I. Sahman

Abstract: The majority of cataclysmic variable (CV) stars contain a stochastic noise component in their light curves, commonly referred to as flickering. This can significantly affect the morphology of CV eclipses and increases the difficulty in obtaining accurate system parameters with reliable errors through eclipse modelling. Here we introduce a new approach to eclipse modelling, which models CV flickeri… ▽ More The majority of cataclysmic variable (CV) stars contain a stochastic noise component in their light curves, commonly referred to as flickering. This can significantly affect the morphology of CV eclipses and increases the difficulty in obtaining accurate system parameters with reliable errors through eclipse modelling. Here we introduce a new approach to eclipse modelling, which models CV flickering with the help of Gaussian processes (GPs). A parameterised eclipse model - with an additional GP component - is simultaneously fit to 8 eclipses of the dwarf nova ASASSN-14ag and system parameters determined. We obtain a mass ratio $q$ = 0.149 $\pm$ 0.016 and inclination $i$ = 83.4 $^{+0.9}_{-0.6}$ $^{\circ}$. The white dwarf and donor masses were found to be $M_{w}$ = 0.63 $\pm$ 0.04 $M_{\odot}$ and $M_{d}$ = 0.093 $^{+0.015}_{-0.012}$ $M_{\odot}$, respectively. A white dwarf temperature $T_{w}$ = 14000 $^{+2200}_{-2000}$ K and distance $d$ = 146 $^{+24}_{-20}$ pc were determined through multicolour photometry. We find GPs to be an effective way of modelling flickering in CV light curves and plan to use this new eclipse modelling approach going forward. △ Less

Submitted 20 September, 2016; originally announced September 2016.

Comments: 13 pages, 7 figures, to be published in Monthly Notices of the Royal Astronomical Society (MNRAS)

NSV 1907 - A new eclipsing, nova-like cataclysmic variable

Authors: Stefan Hümmerich, Rainer Gröbel, Franz-Josef Hambsch, Franky Dubois, Richard Ashley, Boris T. Gänsicke, Siegfried Vanaverbeke, Klaus Bernhard, Patrick Wils

Abstract: NSV 1907, formerly listed as an irregular variable in variability catalogues, was classified as an Algol-type eclipsing binary in the Catalina Surveys Periodic Variable Star Catalogue. We have identified NSV 1907 as an ultraviolet (UV) bright source using measurements from the GALEX space telescope and detected obvious out-of-eclipse variability in archival photometric data from the Catalina Sky S… ▽ More NSV 1907, formerly listed as an irregular variable in variability catalogues, was classified as an Algol-type eclipsing binary in the Catalina Surveys Periodic Variable Star Catalogue. We have identified NSV 1907 as an ultraviolet (UV) bright source using measurements from the GALEX space telescope and detected obvious out-of-eclipse variability in archival photometric data from the Catalina Sky Survey, which instigated a closer examination of the object. A spectrum and extensive multicolour photometric observations were acquired, from which we deduce that NSV 1907 is a deeply eclipsing, nova-like cataclysmic variable. Apart from the orbital variations (deep eclipses with a period of P ~ 6.63 hours), changes in mean brightness and irregular short-term variability (flickering) were observed. The presence of a secondary minimum at phase phi ~ 0.5 was established, which indicates a significant contribution of the companion star to the optical flux of the system. We find possible evidence for sinusoidal variations with a period of P ~ 4.2 d, which we interpret as the nodal precession period of the accretion disc. No outbursts or VY Scl-like drops in brightness were detected either by the CSS or during our photometric monitoring. Because of its spectral characteristics and the observed variability pattern, we propose NSV 1907 as a new moderately bright long-period SW Sextantis star. Further photometric and spectroscopic observations are encouraged. △ Less

Submitted 26 August, 2016; originally announced August 2016.

Comments: 7 pages, 12 figures

Journal ref: New Astronomy, Volume 50, January 2017, Pages 30 to 36

Long-term eclipse timing of white dwarf binaries: an observational hint of a magnetic mechanism at work

Authors: M. C. P. Bours, T. R. Marsh, S. G. Parsons, V. S. Dhillon, R. P. Ashley, J. P. Bento, E. Breedt, T. Butterley, C. Caceres, C. M. Copperwheat, L. K. Hardy, J. J. Hermes, P. Irawati, P. Kerry, D. Kilkenny, S. P. Littlefair, M. J. McAllister, S. Rattanasoon, D. I. Sahman, M. Vuckovic, R. W. Wilson

Abstract: We present a long-term programme for timing the eclipses of white dwarfs in close binaries to measure apparent and/or real variations in their orbital periods. Our programme includes 67 close binaries, both detached and semi-detached and with M-dwarfs, K-dwarfs, brown dwarfs or white dwarfs secondaries. In total, we have observed more than 650 white dwarf eclipses. We use this sample to search for… ▽ More We present a long-term programme for timing the eclipses of white dwarfs in close binaries to measure apparent and/or real variations in their orbital periods. Our programme includes 67 close binaries, both detached and semi-detached and with M-dwarfs, K-dwarfs, brown dwarfs or white dwarfs secondaries. In total, we have observed more than 650 white dwarf eclipses. We use this sample to search for orbital period variations and aim to identify the underlying cause of these variations. We find that the probability of observing orbital period variations increases significantly with the observational baseline. In particular, all binaries with baselines exceeding 10 yrs, with secondaries of spectral type K2 -- M5.5, show variations in the eclipse arrival times that in most cases amount to several minutes. In addition, among those with baselines shorter than 10 yrs, binaries with late spectral type (>M6), brown dwarf or white dwarf secondaries appear to show no orbital period variations. This is in agreement with the so-called Applegate mechanism, which proposes that magnetic cycles in the secondary stars can drive variability in the binary orbits. We also present new eclipse times of NN Ser, which are still compatible with the previously published circumbinary planetary system model, although only with the addition of a quadratic term to the ephemeris. Finally, we conclude that we are limited by the relatively short observational baseline for many of the binaries in the eclipse timing programme, and therefore cannot yet draw robust conclusions about the cause of orbital period variations in evolved, white dwarf binaries. △ Less

Submitted 2 June, 2016; originally announced June 2016.

Comments: 16 pages, 12 figures

High-speed photometry of the disintegrating planetesimals at WD1145+017: evidence for rapid dynamical evolution

Authors: B. T. Gaensicke, A. Aungwerojwit, T. R. Marsh, V. S. Dhillon, D. I. Sahman, Dimitri Veras, J. Farihi, P. Chote, R. Ashley, S. Arjyotha, S. Rattanasoon, S. P. Littlefair, D. Pollacco, M. R. Burleigh

Abstract: We obtained high-speed photometry of the disintegrating planetesimals orbiting the white dwarf WD1145+017, spanning a period of four weeks. The light curves show a dramatic evolution of the system since the first observations obtained about seven months ago. Multiple transit events are detected in every light curve, which have varying durations(~3-12min) and depths (~10-60%). The time-averaged ext… ▽ More We obtained high-speed photometry of the disintegrating planetesimals orbiting the white dwarf WD1145+017, spanning a period of four weeks. The light curves show a dramatic evolution of the system since the first observations obtained about seven months ago. Multiple transit events are detected in every light curve, which have varying durations(~3-12min) and depths (~10-60%). The time-averaged extinction is ~11%, much higher than at the time of the Kepler observations. The shortest-duration transits require that the occulting cloud of debris has a few times the size of the white dwarf, longer events are often resolved into the superposition of several individual transits. The transits evolve on time scales of days, both in shape and in depth, with most of them gradually appearing and disappearing over the course of the observing campaign. Several transits can be tracked across multiple nights, all of them recur on periods of ~4.49h, indicating multiple planetary debris fragments on nearly identical orbits. Identifying the specific origin of these bodies within this planetary system, and the evolution leading to their current orbits remains a challenging problem. △ Less

Submitted 9 January, 2016; v1 submitted 30 December, 2015; originally announced December 2015.

Comments: Astrophysical Journal Letters in press

Shell evolution approaching the N=20 island of inversion: structure of 26Na

Authors: G. L. Wilson, W. N. Catford, N. A. Orr, C. Aa. Diget, A. Matta, G. Hackman, S. J. Williams, I. C. Celik, N. L. Achouri, H. Al Falou, R. Ashley, R. A. E. Austin, G. C. Ball, J. C. Blackmon, A. J. Boston, H. C. Boston, S. M. Brown, D. S. Cross, M. Djongolov, T. E. Drake, U. Hager, S. P. Fox, B. R. Fulton, N. Galinski, A. B. Garnsworthy , et al. (15 additional authors not shown)

Abstract: The levels in 26Na with single particle character have been observed for the first time using the d(25Na,p gamma) reaction at 5 MeV/nucleon. The measured excitation energies and the deduced spectroscopic factors are in good overall agreement with (0+1) hbar-omega shell model calculations performed in a complete spsdfp basis and incorporating a reduction in the N=20 gap. Notably, the 1p3/2 neutron… ▽ More The levels in 26Na with single particle character have been observed for the first time using the d(25Na,p gamma) reaction at 5 MeV/nucleon. The measured excitation energies and the deduced spectroscopic factors are in good overall agreement with (0+1) hbar-omega shell model calculations performed in a complete spsdfp basis and incorporating a reduction in the N=20 gap. Notably, the 1p3/2 neutron configuration was found to play an enhanced role in the structure of the low-lying negative parity states in 26Na, compared to the isotone 28Al. Thus, the lowering of the 1p3/2 orbital relative to the 0f7/2 occurring in the neighbouring Z=10 and 12 nuclei -- 25,27Ne and 27,29Mg -- is seen also to occur at Z=11 and further strengthens the constraints on the modelling of the transition into the island of inversion. △ Less

Submitted 3 June, 2016; v1 submitted 7 August, 2015; originally announced August 2015.

Comments: 6 figures