On detecting variables using ROTSE-IIId archival data

Size: px

Start display at page:

Download "On detecting variables using ROTSE-IIId archival data"

Stella Dickerson
8 years ago
Views:

1 Second Workshop on Robotic Autonomous Observatories ASI Conference Series, 2012, Vol. 7, pp Edited by Sergey Guziy, Shashi B. Pandey, Juan C. Tello & Alberto J. Castro-Tirado On detecting variables using ROTSE-IIId archival data C. Yesilyaprak 1, S. K. Yerli 2, N. Aksaker 3, B. B. Gucsav 4, U. Kiziloglu 2, E. Dikicioglu 1, D. Coker 4, E. Aydin 4 and F. F. Ozeren 5 1 Ataturk Universitesi, Science Faculty, Physics Department, Erzurum, Turkey 2 Orta Dogu Teknik Universitesi, Physics Department, Ankara, Turkey 3 Cukurova Universitesi, Technical Science Vocational School, Adana, Turkey 4 Ankara Universitesi, Astronomy and Space Science Department, Ankara, Turkey 5 Erciyes Universitesi, Astronomy and Space Sciences Department, Kayseri, Turkey Abstract. ROTSE (Robotic Optical Transient Search Experiment) telescopes can also be used for variable star detection. As explained in the system description (Akerlof et al. 2003), they have a good sky coverage and they allow a fast data acquisition. The optical magnitude range varies between 7 m to 19 m. Thirty percent of the telescope time of north-eastern leg of the network, namely ROTSE-IIId (located at TUBITAK National Observatory, Bakirlitepe, Turkey is owned by Turkish researchers. Since its first light (May 2004) considerably a large amount of data has been collected (around 2 TB) from the Turkish time and roughly one million objects have been identified from the reduced data. A robust pipeline has been constructed to discover new variables, transients and planetary nebulae from this archival data. In the detection process, different statistical methods were applied to the archive. We have detected thousands of variable stars by applying roughly four different tests to light curve of each star. In this work a summary of the pipeline is presented. It uses a high performance computing (HPC) algorithm which performs inhomogeneous ensemble photometry of the data on a 36 core cluster. This study is supported by TUBITAK (Scientific and Technological Research Council of Turkey) with the grant number TBAG-108T475. Keywords : ROTSE software pipeline variable analysis cahity@atauni.edu.tr

F. Ozeren 5 1 Ataturk Universitesi, Science Faculty, Physics Department, Erzurum, Turkey 2 Orta Dogu Teknik Universitesi, Physics Department, Ankara, Turkey 3 Cukurova Universitesi, Technical Science

2 54 C. Yesilyaprak et al. 1. Introduction Robotic Optical Transient Search Experiment (ROTSE, Akerlof et al. (2003)) is a network of telescopes located all around the world. The primary goal of the ROTSE- III project is to observe Gamma-Ray Bursts (GRB) in unfiltered optical light. Each ROTSE-III telescope consists of 45-cm mirror with a wide field of view (1.85 o ). The ROTSE-III collaboration uses 70% of each ROTSE-III telescope s observation time. The rest of the time is allocated for discretion by the local organization. The ROTSE- IIId telescope is located at TUBITAK National Observatory (TUG), Bakrltepe, Antalya, Turkey. The aim of this work is to detect variability as well as finding new variables using the ROTSE-IIId archival data. To achieve this main goal multistage algorithms were developed. 2. Archive The ROTSE-IIId has a CCD with a pixel scale of 3.3 /pixel. Even though ROTSE-IIId has no filters, it has a wide pass-band which peaks at 550 nm (Akerlof et al. 2003). Throughout Turkish share of observations between 2004 and 2010, approximately CCD frames were collected from 645 different pointings which corresponds to approximately 2 TB of data; when reduced approximately one million objects have been obtained. The limiting magnitudes of ROTSE-IIId with three different exposure times of 5 s, 20 s and 60 s are 18.0m, 18.7m and 19.3m, respectively. The majority of CCD frames analyzed in this study were taken with 5 s exposure time. 3. The pipeline We have constructed a new, fast, robust and reliable pipeline to detect variable stars from the ROTSE-IIId archival data. The main stages of the pipeline were as follows: Object identification (Bertin & Arnouts 1996), astrometry of the objects (Bertin 2006), light curve generation and inhomogeneous ensemble photometry (Saesen et al. 2010). For the first time in the ROTSE-III archive, a high performance computing (HPC) algorithm has been implemented into the pipeline (Gucsav et al. 2012). A Main stage of the pipeline is given in Fig.1. Depending on the data quality, either corrected CCD images or the calibrated object catalogs produced by the ROTSE-IIId pipeline could be used in the pipeline. The last stage of the pipeline, namely inhomogeneous ensemble photometry (implemented to the ROTSE-III archive for the first time), gives relative light variations of the measured stars with high precision. The new pipeline works on a small cluster called Infinitus which made use of

The ROTSE-III collaboration uses 70% of each ROTSE-III telescope s observation time. The rest of the time is allocated for discretion by the local organization.

3 ROTSE-IIId pipeline 55 Figure 1. Main stages of the ROTSE pipeline.

4 56 C. Yesilyaprak et al. 8 different computers with 36 cores each having 8 GHz CPU speed. GNU/Linux operating system and Lustre file system is used on all computers. Computers are interconnected with a gigabyte Ethernet. The pipeline is mainly written in C language. Parallel algorithms have been used in every step of the pipeline which made use of MPICH2 library. 4. Analysis Scatter-Error Analysis In finding the variable stars there exists a very simple and common method: a plot of standard deviation of the light curve versus mean magnitude. Red line shows moving average of the standard deviation. Stars located above the red line ( 10000) show high variability. Analysis of Variance The analysis of graph of variance versus RMS magnitude can be used to obtain the variables. Two groups arise from the figure: 1) possible variables with P 15 days. 2) no variability with P < 15 days. In the analysis, light curves must have at least 100 data points and then sigma of the light curve should be > 0.1. Lastly, for the light curves (σ/mag) should be > 5 (Wozniak et al. 2004). Abbe Index The third test is the Abbe test. This test gave another indication of variability. It depends on the point-topoint variations and is sensible to the derivative of the light curve. Abbe index give three possibilities: 1) 1 constant stars. 2) > 1 variables. 3) << 1 variability on longer time scales (Saesen et al. 2010).

4. Analysis Scatter-Error Analysis In finding the variable stars there exists a very simple and common method: a plot of standard deviation of the light curve versus mean magnitude.

5 ROTSE-IIId pipeline Results and discussion According to the above analysis we made some concluding remarks listed below: 1,000,000 light curves analyzed. 43,000 light curves passed for further analysis according to the data points. 9,000 stars are passed from 3 tests at the same time. 2,000 identified as variable stars. 7,000 unidentified variable stars. The work is in progress on different techniques PDM, Lomb-Scargle, SigSpec to find variable stars. We expect to find new periods of hundreds of SR and Mira variables. New SR and Mira stars will be searched within unidentified variable stars. Unidentified variable stars will be classified and more detailed analysis will be done. Unidentified variable stars will be searched in other telescope archives. Pipeline is still being developed. In the near future trend extraction algorithm will be used. Acknowledgments This project utilizes data obtained by the Robotic Optical Transient Search Experiment (ROTSE). ROTSE is a collaboration of Lawrence Livermore National Lab, Los Alamos National Lab and the University of Michigan ( The archival data of ROTSE-IIId were obtained from TÜBİTAK (Turkish Scientific and Research Council) National Observatory (TUG) which was observed by the Turkish observers. This study was supported by TÜBİTAK with the project TBAG-108T475. Background Photo: ROTSE-IIId at TUG, Bakirlitepe, Turkey - References Akerlof C. W., Kehoe R. L., McKay T. A., et al., 2003, PASP, 115, 132 Bertin E., Arnouts S., 1996, A&AS, 117, 393 Bertin E, 2006, ASP Conference Series, 351, 112 Saesen S., Carrier F., Pigulski A., et al., 2010, A&A, 55, 16 Güçsav, B. B., Yeşilyaprak C., Yerli S. K., et al., 2012, Experimental Astronomy, 33, 1, 197 Wozniak P. R., Williams S. J., Vestrand W. T., et al., AJ, 128, 2965

7,000 unidentified variable stars. The work is in progress on different techniques PDM, Lomb-Scargle, SigSpec to find variable stars.

Software challenges in the implementation of large surveys: the case of J-PAS

Software challenges in the implementation of large surveys: the case of J-PAS 1/21 Paulo Penteado - IAG/USP pp.penteado@gmail.com http://www.ppenteado.net/ast/pp_lsst_201204.pdf (K. Taylor) (A. Fernández-Soto)