Multimessenger astronomy

An invited leture about the transient universe and the TOROS project.

Transcript

1. Marcelo Lares (IATE, OAC) UNAB, 6 oct 2020 The transient universe and the TOROS project
3. Gravitational waves A solution to Einstein’s field equations has a wave form, with two polarizations. Assume a first order perturbation of the Minkowsky space. Free falling test point particles react as the wave passes.
4. Gravitational waves 1905 – Poincaré suggests gravity is transmited by GWs 1915 – GR theory 1922 – Eddington paper 1936 – Einstein & Rosen paper 1968 – Weber experiments
5. Gravitational waves 1974 – PSR B1913+16, “Hulse & Taylor pulsar” Discovered by Russel Hulse & Joseph Taylor using Arecibo radio telescope. Nobel Prize in 1993. “Orbital decay” observed as period shift, predicted by GR
6. GW observatories 0
7. GW observatories
8. GW observatories
10. 2015 – First observation of GW, LIGO GW observatories
11. 1900 Km LIGO Hanford LIGO Livingston
12. LIGO Livingston GEO600 LIGO India KAGRA LIGO Hanford Virgo
13. LIGO Livingston GEO600 LIGO India KAGRA LIGO Hanford Virgo 2015 2020 2025 2015 2020 2025 O1 O2 O3 O4 O5
15. Gravitational waves
16. Gravitational waves Adhikari 2013 RMS sensitivity for antenna pattern X polarization X polarization + polarization + polarization
19. Ears & eyes to the universe The dawn of multimessenger astronomy
20. kilonovas Collision of neutron stars Hypothesized for a long time a.k.a. macronovas, mergernovas 1000 brighter than novas
21. Lippuner 2017
22. Lippuner 2017
23. Kilonova ? Tanvir et al. 2011
24. Center for Gravitational Wave Astronomy, UTRGV IATE-Observatorio Astronómico Córdoba-UNC Texas A&M Astronomical Instrumentation Lab. Universidad Nacional de Salta TOROS Transients Observatory Robotic Of the South Transients Observatory Robotic Of the South
25. Cerro Macón astronomical reserve ELT candidate site Solar energy supply Astronomical reserve Roads and infrastructure Site of ABRAS, TOROS and ISON Operations center in Tolar Grande TOROS site
27. TOROS pilot toritos Development of the site: energy, conectivity Operations center (Salta) Telescope network: EABA, CASLEO, Gemini, Falcon ABRAS
28. ///////////////////////////////////////////////////////////////////////////// TITLE: GCN CIRCULAR NUMBER: 18330 SUBJECT: LIGO/Virgo G184098: Burst candidate in LIGO engineering run data DATE: 15/09/20 00:53:16 GMT FROM: Leo Singer at NASA/GSFC <leo.p.singer@nasa.gov> Dear colleagues, We would like to bring to your attention a trigger identified by the online Burst analysis during the ongoing Engineering Run 8 (ER8). Normally, we would send this in the form of a private GCN Circular, but the LIGO/Virgo GCN Circular list is not ready yet. The LIGO Scientific Collaboration and Virgo report that the cWB unmodeled burst analysis identified candidate G184098 during real-time processing of data from LIGO Hanford Observatory (H1) and LIGO Livingston Observatory (L1) at 2015-09-14 09:50:45 UTC (GPS time: 1126259462.3910). Alerts were not sent in real-time because the candidate occurred in ER8 data; however, we have now sent GCN notices through our normal channel. G184098 is an unvetted event of interest, as the false alarm rate (FAR) determined by the online analysis would have passed our stated alert threshold of ~1/month. The event's properties can be found at this URL: https://gracedb.ligo.org/events/G184098 Available through a LoU
29. ///////////////////////////////////////////////////////////////////////////// TITLE: GCN CIRCULAR NUMBER: 18330 SUBJECT: LIGO/Virgo G184098: Burst candidate in LIGO engineering run data DATE: 15/09/20 00:53:16 GMT FROM: Leo Singer at NASA/GSFC <leo.p.singer@nasa.gov> Dear colleagues, We would like to bring to your attention a trigger identified by the online Burst analysis during the ongoing Engineering Run 8 (ER8). Normally, we would send this in the form of a private GCN Circular, but the LIGO/Virgo GCN Circular list is not ready yet. The LIGO Scientific Collaboration and Virgo report that the cWB unmodeled burst analysis identified candidate G184098 during real-time processing of data from LIGO Hanford Observatory (H1) and LIGO Livingston Observatory (L1) at 2015-09-14 09:50:45 UTC (GPS time: 1126259462.3910). Alerts were not sent in real-time because the candidate occurred in ER8 data; however, we have now sent GCN notices through our normal channel. G184098 is an unvetted event of interest, as the false alarm rate (FAR) determined by the online analysis would have passed our stated alert threshold of ~1/month. The event's properties can be found at this URL: https://gracedb.ligo.org/events/G184098 Available through a LoU
30. inspiral merger merger ring-down Strain * 1.e21 GW150914 First detection of a gravitational wave
31. inspiral merger merger ring-down
32. Follow up Time since the merger, days TOROS TOROS
33. Follow up
34. First detection using HLV First detection of a GW First observation of a Kilonova
36. 150914 – First detection of a GW 170817 – First detection of a GW with an EM counterpart 190521 – First evidence of intermediate mass black holes 190814 – Lightest black hole (or heaviest neutron star?) 190412 – First detection of unequal mass binary merger
41. GW170817The birth of multimessenger astronomy
45. Transient detections Classification problem Real/bogus Machine learning approach
46. TOROS dome
47. TOROS telescope 0.60m diameter focal corrector CCD 10Kx10K, 9 microns/ pixel 1815 mm FD 1 arcsec/pixel FOV: 3 x 3 deg mag 22~23 w/ 15 min exposure
48. TOROS aims and scope Galaxy survey for reference imaging Follow up of gravitational wave alerts Multimessenger astronomy Develop automatic data reduction pipelines Develop astrostatistics and astroinformatics Develop robotics and advanced electronics ABRAS, EABA, CASLEO and GEMINI sinergy Big data and machine learning Training of a new generation of astronomers Develop the site Favour the installation of new facilities • • •
49. TOROS pipeline We need an automatic data reduction pipeline, which: Is robust to conection cuts Work without human intervention Mange large volumes of data Automatic processing of images Output list of possible candidates • •
50. Reference image New image difference Zackay Bramich Hotpants Scorr Z B H S Methods:
51. Zackay Bramich Hotpants Scorr Methods: deconvolution of the kernel function into base functions, Bayesian Pixel based hypothesis testing Implementation of the Alard-Lupton method Cross-correlation of the image with its PSF
52. Zackay et al. 2016
53. Andreoni et al. 2017
54. CODE COMPUTER RESULTS DATA CODE COMPUTER RESULTS DATA Traditional paradigm Machine Learning Models Statistics Numerical methods HPC Interpretation Models Statistics Numerical methods HPC Architectures Traing data
55. Support vector machines Decision boundary
56. K nearest neighbors
57. Random forests . . .
58. TP FP TN FN TP FP TN FN TP FN FP TN predicted actual P N N P Precision, purity Recall, TPR, or sensitivity Metrics of preformance: F1 score = Harmonic mean of precision and sensitivity Accuracy = classified right / total
59. TP (event) FP (bogus) Probability of true event 0 1 threshold 0 1 1 0 FPR (contamination) TPR (purity) 0 1 precision 1 0 recall
60. New image Reference image REALS BOGUS Classification by machine learning methods real bogus Zackay Bramich Hotpants Scorr
61. real bogus
66. Time domain and multimessenger astronomy