The Most Interesting Anomalies Discovered in ZTF DR3 from the SNAD-III Workshop

, , , , , , , , , and

Published July 2020 © 2020. The American Astronomical Society. All rights reserved.
, , Citation Patrick D. Aleo et al 2020 Res. Notes AAS 4 112 DOI 10.3847/2515-5172/aba6e8

2515-5172/4/7/112

Abstract

The search for objects with unusual astronomical properties, or anomalies, is one of the most anticipated results to be delivered by the next generation of large scale astronomical surveys. Moreover, given the volume and complexity of current data sets, machine learning algorithms will undoubtedly play an important role in this endeavor. The SNAD team is specialized in the development, adaptation and improvement of such techniques with the goal of constructing optimal anomaly detection strategies for astronomy. We present here the preliminary results from the third annual SNAD workshop (https://snad.space/2020/) that was held on-line in 2020 July.

Export citation and abstract BibTeX RIS

SNAD9 is a project devoted to tackling the anomaly detection problem in large astronomical data sets using machine learning methods (Malanchev et al. 2020; Ishida et al. 2019; Kornilov et al. 2019; Pruzhinskaya et al. 2019). The main focus of the third SNAD workshop was anomaly detection in the public Data Release 3 (DR3)10 of the Zwicky Transient Facility (ZTF, Bellm et al. 2019), using the data from the first 9.4 months of ZTF science operations, dated from 2018 March 17th to December 31st. We analyzed two ∼47 sq. deg fields (fieldid = 795 and fieldid = 807), and a third ∼3 sq. deg. field (fieldid = 695, ccdid = 11) which contains the M31 galaxy. Each light curve was characterized by 42 features (e.g., amplitude, periodogram peak, etc.) which have proven their effectiveness in describing variable starlight curves (Kim et al. 2014; D'Isanto et al. 2016). Four different anomaly detection algorithms were applied to the data: Gaussian Mixture Model (Reynolds 2009), Local Outlier Factor (Breunig et al. 2000), One-Class Support Vector Machine (Khan & Madden 2010), and Isolation Forest (Liu et al. 2008, 2012). The outliers given by the algorithms were summarized in a common table and were ordered in the decreasing order of their anomaly scores. Since not all of the outliers are necessarily interesting astrophysical anomalies (for example, they could be artifacts generated during image pre-processing, etc.), they were also investigated by our team experts.

For each field, we obtained a list with potentially interesting astronomical objects which will be subjected to further analysis. We report some of the interesting non-cataloged anomalies in Table 1. The first column contains the object identifiers from ZTF DR3 (some of the targets also appeared in the ZTF alert stream and, therefore, have an additional identifier). The equatorial coordinates (J2000.0) are given in the second column while the maximum r-magnitudes from ZTF automatic photometry are given in column 3 (the magnitude is not corrected for Milky Way dust extinction). In a few cases, an SDSS galaxy has been identified at the source position. For these objects we estimated the r-band absolute magnitude at maximum light assuming a flat ΛCDM cosmology with ΩΛ = 0.702 ± 0.022 (Scolnic et al. 2018) and H0 = 70 km s−1 Mpc−1 (column 4). Candidate host galaxies from SDSS DR16 11 (if available) and their corresponding photometric redshifts (zph, obtained via the KD-tree method) are given in the columns 5 and 6. The last column contains comments about the possible origin of the source.

Table 1.  List of Selected Non-cataloged Anomalies

OID R.A. (deg), Decl. (deg) mr,max Mr,max Possible Host zph Comments
795209200003484/ZTF18abbpebf 251.54866 56.33124 19.0       SN Iaa/SN/AGN
795212100007964/ZTF18aanbksg 242.93762 55.96133 19.3 −21.6 SDSS J161144.90+555740.7 0.29 Blazara/SN/AGN
795205100007271/ZTF18aayatjf 252.30216 54.11178 19.5       SN Iaa/SN/AGN
795202100005941/ZTF18aanbnjh 248.65767 52.27841 19.4 −22.4 SDSS J163437.92+521642.2 0.42 QSO-Ia/SN/SLSN/AGN
795213200000671/ZTF18aaincjv 251.93077 58.50559 18.9       AGN-Ia/SN/AGN
795204100013041/ZTF18abgvctp 242.30742 52.21426 20.0 −21.6 SDSS J160913.83+521251.3 0.38 SN/AGN
695211300004359 10.45371 41.15395 17.6 −6.9b M31    
695211100131796 11.04581 41.55487 16.8 −7.7b M31   LBV/SDOR
695211100022045/ZTF18abgpztr 10.98339 41.53641 18.7 −5.8b M31   AGN-Ia/PNV

Notes.

aAccording to the machine learning light curve classification of the ALeRCE broker (http://alerce.science/). bAssuming that distance to M31 is ∼780 kpc, http://leda.univ-lyon1.fr/.

Among the anomalies listed in Table 1, there are six objects that are likely extragalactic in origin. These are potential supernovae or active galactic nuclei (AGNs) candidates. Three other detected sources possibly belong to the M31 galaxy.

  • 1.  
    ZTF DR3 contains only the descending part of the light curve for object 795202100005941. Considering the photometric redshift of the possible host galaxy, its absolute magnitude is ∼−22.4 which makes it a potential superluminous supernova (SLSN) candidate.
  • 2.  
    The object 795213200000671 also appeared in Gaia Alerts12 on 2018 March 12 with an apparent magnitude of 18.83 mag in G filter which was ∼8 days before ZTF first detection (AT 2018afr/Gaia18apj).
  • 3.  
    The object 695211100131796 is located near the ionized hydrogen region [AMB2011] H ii 2692 (Azimlu et al. 2011). It was previously detected as an object of unknown nature PSO J011.0457+41.5548 (Lee et al. 2014). We consider it as a candidate for luminous blue variables (LBV) or for variables of the S Doradus type (SDOR).
  • 4.  
    According to the Transient Name Server13 the object 695211100022045 was first seen on 2017 October 29 as AT 2017ixs. It was detected a second time on 2017 December 15 with 19.5 mag in Clear filter and classified as a possible nova (PNV; Carey 2017). Six months latter, on 2018 June 20, MASTER-Kislovodsk auto-detection system discovered MASTER OT J004355.89+413209.9 with an unfiltered magnitude of 19.0 mag at AT 2017ixs position (Balanutsa et al. 2018). The behavior of its light curve is not typical for dwarf novae or for cataclysmic variables.

M. Kornilov, V. Korolev, K. Malanchev, A. Volnova, and M. Pruzhinskaya are supported by RFBR grant according to the research project 20-02-00779 for preparing ZTF DR3 data, anomaly analysis, and evaluation of different anomaly detection algorithms. E. E. O. Ishida and S. Sreejith acknowledge support from CNRS 2017 MOMENTUM grant under project Active Learning for Large Scale Sky Surveys. We used the equipment funded by the Lomonosov Moscow State University Program of Development. The authors acknowledge the support from the Program of Development of M.V. Lomonosov Moscow State University (Leading Scientific School "Physics of stars, relativistic objects and galaxies").

Footnotes

Please wait… references are loading.
10.3847/2515-5172/aba6e8