Smart Kanata: A Framework for Autonomous Decision Making in Rapid Follow-up Observations of Cataclysmic Variables
Makoto Uemura, \altemailmark Yuzuki Koga, Ryosuke Sazaki, Tomoya Yukino, Tatsuya Nakaoka, Ryo Imazawa, Taichi Kato, Daisaku Nogami, Keisuke Isogai, Naoto Kojiguchi, Kenta Taguchi, Yusuke Tampo, Hiroyuki Maehara, and Shiro Ikeda · 2024
Read Smart Kanata: A Framework for Autonomous Decision Making in Rapid Follow-up Observations of Cataclysmic Variables by Makoto Uemura, \altemailmark Yuzuki Koga, Ryosuke Sazaki, Tomoya Yukino, Tatsuya Nakaoka, Ryo Imazawa, Taichi Kato, Daisaku Nogami, Keisuke Isogai, Naoto Kojiguchi, Kenta Taguchi, Yusuke Tampo, Hiroyuki Maehara, and Shiro Ikeda (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-02092.txt.
Introduction
Makoto Uemura and co-authors run in 2024 Smart Kanata, a self drive watch frame for fast follow up of burst star events. Early views of such bursts tell much of their core physics. Yet the best watch mode is hard to pick at once, since the burst type is still in doubt.
The frame first grades what type of burst star a new find may be. It then picks one of three watch modes by how much each would teach. In near three hundred days it caught twenty one cases. One nova's spectra showed a fast deep dive in a key line as the burst grew.
Catalog record
| CREATOR | Makoto Uemura, \altemailmark Yuzuki Koga, Ryosuke Sazaki, Tomoya Yukino, Tatsuya Nakaoka, Ryo Imazawa, Taichi Kato, Daisaku Nogami, Keisuke Isogai, Naoto Kojiguchi, Kenta Taguchi, Yusuke Tampo, Hiroyuki Maehara, and Shiro Ikeda |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 82 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2412-02092
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Smart Kanata: A Framework for Autonomous Decision Making in Rapid Follow-up Observations of Cataclysmic Variables Makoto Uemura, \altemailmark Yuzuki Koga, Ryosuke Sazaki, Tomoya Yukino, Tatsuya Nakaoka, Ryo Imazawa, Taichi Kato, Daisaku Nogami, Keisuke Isogai, Naoto Kojiguchi, Kenta Taguchi, Yusuke Tampo, Hiroyuki Maehara, and Shiro Ikeda Abstract Studying the early stages of transient events provides crucial information about the fundamental physical processes in cataclysmic variables (CVs). However, determining an appropriate observation mode immediately after the discovery of a new transient presents challenges due to significant uncertainties regarding its nature. We developed a framework designed for autonomous decision making in prompt follow-up observations of CVs using the Kanata 1.5-m telescope. The system, named Smart Kanata, first estimates the class probabilities of variable star types using a generative model. It then selects the optimal observation mode from three possible options based on the mutual information calculated from the class probabilities. We have operated the system for ∼ 300 \sim 300 days and obtained 21 samples, among which automated observations were successfully performed for a nova and a microlensing event. In the time-series spectra of the nova V4370 Oph, we detected a rapid deepening of the absorption component of the H α \alpha line. These initial results demonstrate the capability of Smart Kanata in facilitating rapid observations and improving our understanding of outbursts and eruptions of CVs and other galactic transients. 2024 \Received 2023/11/03 \Accepted 2024/12/03 \KeyWords accretion, accretion disks — methods: observational — stars: dwarf novae — novae, cataclysmic variables 1 Introduction Cataclysmic variables (CVs) are close binary systems in which a white dwarf experiences accretion from a companion star that fills its Roche lobe (for a review, see [ 66 ] ). This accretion process results in transient phenomena, including nova eruptions and dwarf nova outbursts. Uninterrupted observations spanning the entire duration of these phenomena are essential to understand them. However, capturing the initial phase, particularly the rapidly rising phase, presents challenges due to its short time-scale of less than 1 day. Exploring the initial stage of transients, typically characterized by an unsteady transitional state, provides valuable insights into their underlying physical models. In novae, the accretion triggers a runaway thermonuclear burst on the surface of the white dwarf. This burst leads to the ejection of accumulated matter, making a photosphere that envelops the entire binary star system (for a recent review, see [ 13 ] ). The dynamic formation and subsequent evolution of the ejecta have recently been investigated through spectroscopic observations conducted during the initial stages of nova eruptions. In the pioneering work of Arai et al. (2015) , an optical spectrum was acquired within an extremely early temporal window, specifically at 0.19 days following the initial detection of the nova eruption in T Pyx. This spectrum exhibited distinct features reminiscent of Wolf-Rayet stars, including N \emissiontype V, \emissiontype IV, C \emissiontype IV and He \emissiontype II emission lines, which were unprecedented for novae. In particular, these highly excited lines disappeared within 2 days. A similar spectrum was identified during the rising phase of Gaia22alz ( Aydi et al., 2023 ) and V1405 Cas ( Taguchi et al., 2023 ) . Even in ordinary emission lines, Yamanaka et al. (2010) detected the rapid evolution of high-velocity components within the first day of the nova eruption in U Sco. These observations indicate the presence of previously uncharacterized conditions in the early phases of nova eruptions, suggesting that further investigation is needed to understand the underlying mechanisms. Dwarf novae exhibit a repetitive pattern of outbursts that are triggered by the thermal instability of the accretion disk ( Osaki, 1996 ) . Within the category of dwarf novae, WZ Sge-type stars exhibit distinctive features: large outburst amplitudes ( \sim 8 mag), long recurrence time ( \sim 10 yr), and the emergence of early superhumps during the initial week of their outburst ( Howell et al., 1995 ; Kato, 2015 ) . The early superhumps are short-term periodic fluctuations characterized by a doubly-peaked profile of the orbital period. They arise from geometric irregularities within the disk structure. However, the underlying mechanism responsible for the disk structure remains poorly understood. WZ Sge stars have the most extreme binary mass ratio ( M_{2}/M_{1}\lesssim 0.1 ) among CVs. Consequently, the accretion disk in these systems is expected to approach the 2:1 resonance radius. At this radius, tidal dissipation would influence the disk structure ( Osaki, Meyer, 2002 ) . This scenario emphasizes the importance of early superhumps for studying the tidal effect in accretion disks in binary systems with extreme mass ratios. In the case of both novae and dwarf novae, it is necessary to observe the very early stages of the transient event to understand the aforementioned phenomena. The appropriate mode of the follow-up observation depends on the nature of the transient. For instance, photometric observations are employed to identify early superhumps, while spectroscopic observations are used to investigate the formation process of the nova ejecta. Unlike objects such as supernovae, where spectroscopic observations are consistently prioritized for follow-up, the availability of this diverse selection complicates immediate follow-up observations of CVs. Determining the suitable observation mode for a new transient is a challenging task because of the limited information available, leading to substantial uncertainties in its nature immediately after its discovery. This decision-making process has traditionally been the responsibility of experts in CVs; however, there exists a risk of overlooking critical phases of the event without the guidance of these experts during nighttime observations. Another problem arises from the substantial increase in the count of transient events suitable for potential follow-up observation due to large survey projects in recent years (e.g. ASAS-SN in Jayasinghe et al. (2018) ; Zwicky Transient Facility in Bellm et al. (2019) ). The use of manual decision making by experts has become constrained as a result of this increase. In this paper, we present a framework designed for the autonomous decision-making process in rapid follow-up observations of CVs using the Kanata 1.5-m telescope at the Higashi-Hiroshima Observatory. We call this system Smart Kanata. The novelty of our Smart Kanata as an observation system is that it automatically determines and executes the observation expected to have the largest information gain using the framework of information theory, rather than simply automatically executing a predetermined observation. Previous studies have explored machine learning techniques for classifying transient events and variable stars (for a review, see Förster et al. (2021) ). However, most existing models are designed for broad classifications of variable star types or focus exclusively on particular types, such as supernovae (e.g., Muthukrishna et al. (2019) ) and pulsating variables (e.g., Jayasinghe et al. (2019) ). In contrast, our study targets finer classifications of CVs, including subtypes of dwarf novae, necessitating the development of a new and specialized classification model. The automation of observations, from the detection of transient events to their follow-up, has advanced significantly, particularly in studies of gamma- ray bursts and supernovae (e.g., Gal-Yam et al. (2011) ). These automated systems have made substantial contributions to our understanding of such phenomena; however, they often rely on static decision trees or predefined observation priorities. Mahabal et al. (2008) and Djorgovski et al. (2011) introduced the concept of using probabilistic models for dynamic,