Time series exploration with Stingray

Invited talk given remotely at the 9th Microquasar Workshop: https://sites.google.com/inaf.it/microquasar-2020/home

Abstract:
New ideas about how to analyze time-domain X-ray astronomy data have initiated the “spectral-timing revolution,” leading to a surge in developments of analysis techniques. Many individual tools and libraries exist, and some are even publicly available, but we lacked a coherent set for a complete analysis that provides documentation and tutorials of the techniques themselves, not just the syntax for implementation. Stingray is a community-developed, open-source software package in Python for spectral-timing analysis of astrophysical data. This code provides a library of spectral-timing analysis tools while following the Astropy guidelines for modern open-source scientific software development. Our aim is to provide the community with a package that eases the learning curve for state-of-the-art spectral-timing techniques, with a correct statistical framework, to make maximal use of data from NuSTAR, NICER, and potentially STROBE-X and eXTP. In this talk, I will review the spectral-timing analysis tools in the Stingray library and show examples from ongoing research. For more information on Stingray, see: https://stingray.science

More about Dr. Abbie Stevens: https://abigailstevens.com/

Transcript

1. Abbie Stevens, on behalf of Stingray contributors and developers Various astronomy things in the state of Michigan Instructor in general sciences, paused NSF A&A postdoc fellow alstev@msu.edu @abigailstev github.com/abigailstev Time series exploration with Stingray
2. Features in X-ray light curves Strohmayer 01 Abbie Stevens • Michigan State U. & U. Michigan § Coherent pulsations from NSs ⇾ Burst oscillations, spin-up, spin-down, glitches, transition/switching § Quasi-periodic oscillations (QPOs) from BHs and NSs ⇾ Low-freq. (~0.1-20 Hz): Precession of corona/hot flow? Magnetic warps in disk? ⇾ High-freq. (~100-1000 Hz): Hot blobs in Keplerian orbit at inner disk edge? Radial oscillations in NS boundary layer? § Broadband/band-limited noise at ≲ 1 Hz Stevens
3. Fourier transforms (for evenly sampled data) § X-ray light curves of BHs and NSs vary on timescales from microseconds to years § Shorter (< 1 minute) variability: Fourier analysis! ⇾ Study time domain f in the frequency domain f ⇾ Break down light curve into sine waves, take amplitude of sines at each frequency ^ Image: L. Barbosa via wikiMedia Abbie Stevens • Michigan State U. & U. Michigan
4. But it’s not just X-ray data! Abbie Stevens • Michigan State U. & U. Michigan § A.Tetarenko+21: Multi- wavelength fast timing in radio, sub-mm, optical, and X-ray § Paice+21: optical/X-ray lags § Kimura+21: power spectra of a dwarf nova in a CV § Crinquand+21: power spectra of γ-ray light curves from 2D GR-PIC § Nimmo+21: FRB repeaters
5. Slide adapted from J. VanderPlas SciPy Why Python? Abbie Stevens • Michigan State U. & U. Michigan
6. Stingray core library § Library of time series analysis methods ⇾Light curve manipulation, GTIs ⇾Power spectra, cross spectra, bispectra ⇾Lag-frequency & lag-energy spectra ⇾Rms & covariance spectra ⇾Coherence, cross-correlation ⇾Phase-resolved spectroscopy of QPOs ⇾Pulsation searches Abbie Stevens • Michigan State U. & U. Michigan
7. Stingray core library Abbie Stevens • Michigan State U. & U. Michigan § Library of time series analysis methods § Simulation tools ⇾Timmer & Koenig power spectra ⇾Pulsations ⇾Impulse response, 2D transfer functions
8. Stingray core library Abbie Stevens • Michigan State U. & U. Michigan § Library of time series analysis methods § Simulation tools § Modeling tools ⇾Power spectra, cospectra: pulsations, QPOs, noise • Built on scipy and emcee, including Laplace distribution for cospectra (Huppenkothen+Bachetti 18) • Integrating with Astropy.modeling ⇾Bayesian parameter estimation, maximum likelihood
9. § Inform at which frequencies the light curve is varying Plots courtesy of M. Brumback Power spectra (periodograms) Abbie Stevens • Michigan State U. & U. Michigan
10. NICER data of BH MAXI J1631-479 from 2019 (sudden jumps are gaps due to spacecraft orbit) Stevens Transition to intermediate state, Type C QPO appears! Dynamical power spectra (spectrograms) Abbie Stevens • Michigan State U. & U. Michigan
11. § Can get around timing systematics in NuSTAR with cospectra! § Cross spectrum and its products like the cospectra are used extensively in spectral-timing analysis Huppenkothen+Bachetti 18 Bachetti+Huppenkothen 18 Cospectra Abbie Stevens • Michigan State U. & U. Michigan
12. § Can get around timing systematics in NuSTAR with cospectra! § Cross spectrum and its products like the cospectra are used extensively in spectral-timing analysis Huppenkothen+Bachetti 18 Bachetti+Huppenkothen 18 Cospectra Abbie Stevens • Michigan State U. & U. Michigan
13. Fractional covariance NuSTAR data of a pulsar from Stevens+in prep § Fractional variance ⇾ rms: of one light curve (Revnivtzev+99) ⇾ covariance: of one with respect to a reference light curve (Wilkinson+Uttley 09) § Energy resolution of the light curve energy bands (often binned up to be broader than detector resolution) § [Full disclosure: this feature needs more testing!] RMS and covariance spectra Abbie Stevens • Michigan State U. & U. Michigan
14. § Stingray.modeling has range of statistical models for Gaussian, Poisson, χ2, and Laplace distributed data QPO Harmonic Broadband noise MAXI J1535-571 Stevens+18 Modeling power spectra Abbie Stevens • Michigan State U. & U. Michigan
15. § Stingray.modeling has range of statistical models for Gaussian, Poisson, χ2, and Laplace distributed data QPO Harmonic Broadband noise MAXI J1535-571 Huppenkothen+19 Modeling power spectra Abbie Stevens • Michigan State U. & U. Michigan Figure 4. from Stingray: A Modern Python Library for Spectral Timing null 2019 APJ 881 39 doi:10.3847/1538-4357/ab258d http://dx.doi.org/10.3847/1538-4357/ab258d © 2019. The American Astronomical Society. All rights reserved.
16. § Simulates power spectra with specific shapes (white noise, red noise, broadband noise, QPO) § Can be used to estimate errors on power spectra § Warning: does not preserve energy-dependent phase relationships! Though energy channels are supported Flicker noise Poisson noise Brightness Time Time Example in the literature: K2 data of PSR J1023 orbit Kennedy+18 Also have more sophisticated simulation tools with impulse response functions! Figures from Stingray docs Timmer & Koenig simulations Abbie Stevens • Michigan State U. & U. Michigan
17. Lead: M. Bachetti § Evolution of MaLTPyNT, analysis software for NuSTAR timing Stingray + HENRICS + DAVE Abbie Stevens • Michigan State U. & U. Michigan
18. Stingray + HENRICS + DAVE Abbie Stevens • Michigan State U. & U. Michigan
19. HENDRICS: Interactive phaseograms Abbie Stevens • Michigan State U. & U. Michigan null 2019 APJ 881 39 doi:10.3847/1538-4357/ab258d http://dx.doi.org/10.3847/1538-4357/ab258d © 2019. The American Astronomical Society. All rights reserved. Huppenkothen+19 Times Of Arrival can be exported for use with pulsar timing software like PINT or Tempo2
20. Data Analysis for Variable Events Leads: P. Balm, S. Migliari Image from M. Bachetti Stingray + HENRICS + DAVE Abbie Stevens • Michigan State U. & U. Michigan
21. Data Analysis for Variable Events Stingray + HENRICS + DAVE Abbie Stevens • Michigan State U. & U. Michigan Figure 7. from Stingray: A Modern Python Library for Spectral Timing null 2019 APJ 881 39 doi:10.3847/1538-4357/ab258d http://dx.doi.org/10.3847/1538-4357/ab258d © 2019. The American Astronomical Society. All rights reserved. Huppenkothen+19 Leads: P. Balm, S. Migliari See also: Migliari+20, ASPC
22. Documentation Abbie Stevens • Michigan State U. & U. Michigan We also have Jupyter notebook tutorials: github.com/StingraySoftware/notebooks
23. Get involved with Stingray! Abbie Stevens • Michigan State U. & U. Michigan We also have a Slack workspace for discussing features and troubleshooting
24. § Conferences: ⇾ Python in Astronomy conference series ⇾ AstroHackWeek, .Astronomy ⇾ Special sessions at large symposia ⇾ Hack days at AAS, EWASS § Training: ⇾ Software Carpentry and Data Carpentry workshops ⇾ Paid summer internships via Google Summer of Code (Bachelors) ⇾ LSSTC Data Science Fellowship Program (Masters and PhD) § Databases and affiliations: ⇾ Astrophysical Source Code Library: ascl.net ⇾ Astropy affiliated packages: astropy.org/affiliated Facilitating astro research software Abbie Stevens • Michigan State U. & U. Michigan
25. Following best practices for software If what you want isn’t already out there: § Open-source with version control (on, e.g., GitHub) AND LICENSED! § Documentation on how to implement the code *and* what the physics is (cite papers in the docs) § Tutorials with examples of standard use cases § Able to be imported into our existing codes in python (though could use a python wrapper around, e.g., C or Fortran) § Able to run on a standard/nice desktop computer § Plausible that a sysadmin for a university-level computing cluster will install the necessary dependencies so you can run it there for more/faster analysis, fitting, etc. § Accompanying paper for proper citation and credit Abbie Stevens • Michigan State U. & U. Michigan
26. § Stingray is a tool for timing and spectral-timing analysis of interesting astronomical time series ⇾ Stingray version 0.3 was released on June 1st, 2021 ⇾ HENDRICS version 6.0 released on May 31st, 2021 § Install instructions: docs.stingray.science/ § Code & tutorials: github.com/StingraySoftware —Papers: Huppenkothen, Bachetti, ALS+2019, ApJ & JOSS —Cite scientific papers in code documentation —Software is vital for the future of astronomy Summary Abbie Stevens • Michigan State U. & U. Michigan