Abstract
We present a data-adaptive spectral method – Monte Carlo Singular Spectrum Analysis (MC-SSA) – and its modification to tackle astrophysical problems. Through numerical simulations we show the ability of the MC-SSA in dealing with 1∕f β power-law noise affected by photon counting statistics. Such noise process is simulated by a first-order autoregressive, AR(1) process corrupted by intrinsic Poisson noise. In doing so, we statistically estimate a basic stochastic variation of the source and the corresponding fluctuations due to the quantum nature of light. In addition, MC-SSA test retains its effectiveness even when a significant percentage of the signal falls below a certain level of detection, e.g., caused by the instrument sensitivity. The parsimonious approach presented here may be broadly applied, from the search for extrasolar planets to the extraction of low-intensity coherent phenomena probably hidden in high energy transients.
Access provided by Autonomous University of Puebla. Download conference paper PDF
Similar content being viewed by others
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
1 Colored Noise and MC-SSA
1∕f β power-law noise is known to be highly relevant in several astrophysical systems [3–5]. It includes the well-known white noise (β = 0), pink noise (β = 1) and red or Brownian noise (β = 2). Such kind of noises are very often not tied to instrumental disturbance but rather they are property of the observed sources that emit radiation varying in a stochastic manner. In the context of our work, it is sufficient to qualitatively characterize 1∕f β to generate a parsimonious hypothesis test in MC-SSA (see below). Subsequently, the dynamical origin of noise fluctuation or the extra-noise variance – according to acceptance or rejection of our null hypothesis – has to be determined by theoretical analysis.
Singular Spectrum Analysis (SSA) is an effective, data-adaptive and non-parametric method for the decomposition of a time series into a well-defined set of independent and interpretable components that include a non-linear trend, anharmonic, amplitude-modulated oscillations, and noise [6]. In its later developments, the Monte Carlo approach to signal-to-noise separation introduced by [1] has become known as Monte Carlo SSA (MC-SSA) [2]. The distinction between what is noise and what is signal is made through the measure of the resemblance of a given noise surrogate to the original data via eigendecomposition of the time-lagged covariance matrix. For this study the noise surrogates set is generated by using an AR(1) process. In practice, the AR(1) coefficients are estimated from the time-series under consideration by using a maximum-likelihood criterion. Subsequently, these AR(1) noise surrogates are corrupted by Poisson noise to mimic the effect of a photon counting detector.
2 Simulated 1∕f β Series Test
The procedure has been tested on a large sample of artificial time series of arbitrary colored noises. As a test of validity, we show several discrete colored noise vectors of length N = 5,000, with a power-law distribution of slope β ranging from 1 to 2 in steps of two. To obtain sequences of colored noises we use the Matlab library CNOISE.Footnote 1 A threshold of signal identification is incorporated by subtracting a mean value \(\sim \) 35 % of the maximum peak-flux, C max . In this way, we take into account episodes of emission during which the emission count rate drops to the background level. Finally, the time series are corrupted by Poisson noise to simulate the effect of the shot noise based on Poisson photon statistics. The resulting test series, are shown in the supplementary video.Footnote 2 The signal to noise ratio (SNR) of the artificial series test increases as the simulations run.
3 Conclusion and Future Work
We present a modified data-adaptive spectral method – Monte Carlo Singular Spectrum Analysis (MC-SSA) – in which a parsimonious (dynamic and instrumental) noise-model H 0 is adopted for astrophysical applications, i.e. H 0 = {AR(1) + Poisson noise}. AR(1) noise takes into account the long-term variability in the power-law slope −β of a light-curve’s spectral density, and the Poisson noise considers the short-term variability imposed by the quantum nature of light. Our simulations show a remarkable effectiveness of the model for colored noises with β values between 1.5 and 2 and with C max > 1,000 counts; otherwise one should re-bin the light-curve to improve the signal-to-noise ratio of the source.
All analyses were performed by using SSA-MTM Toolkit freeware that has been developed by the Theoretical Climate Dynamics Group at UCLA.Footnote 3 To obtain presented results we relied on an advanced option in SSA-MTM Toolkit for MC-SSA that allows to read in ensemble of surrogates created off-line and corresponding to an arbitrary noise null-hypothesis H 0. In this regard, MC-SSA allows to iteratively refine the working hypothesis and adapt the null-hypothesis H 0 for various theoretical scenarios. We plan to extend further SSA-MTM Toolkit for astrophysical applications by directly including the algorithm for parsimonious H 0 in the freeware. Current and future work will focus on SSA as a promising technique for analysing complex spatio-temporal structures to detect possible periodicities and other statistical regularities due to various astrophysical processes.
References
Allen, M. R., & Smith, L. A. Monte Carlo SSA: detecting irregular oscillations in the presence of coloured noise. J. Climate. 9, 3373–3404 (1996)
Ghil, M., Allen M. R., Dettinger M. D., et al. Advanced spectral methods for climatic time series. Rev. Geophys. 40, 1–41 (2002)
Greco, G., Rosa, R., Beskin, G., et al. Evidence of deterministic components in the apparent randomness of GRBs: clues of a chaotic dynamic. Sci. Rep. 1, 91 (2011)
Pont, F., Zucker, S., & Queloz, D. The effect of red noise on planetary transit detection. MNRAS 373, 231–242 (2006)
Vaughan, S. A Bayesian test for periodic signals in red noise. MNRAS 402, 307–320. (2010)
Vautard, R., & Ghil, M. Singular spectrum analysis in nonlinear dynamics, with applications to paleoclimatic time series. Physica D 35, 395–424 (1989)
Acknowledgements
MB, GG and GS acknowledge the financial support of the Italian Ministry of Education, University and Research (MIUR) through grant FIRB 2012 RBFR12PM1F. CG acknowledges the PRIN MIUR project on “Gamma Ray Bursts: from progenitors to physics of the prompt emission process” (Prot. 2009 ERC3HT). MG and DK received support from the U.S. National Science Foundation (grant DMS-1049253) and from the U.S. Office of Naval Research (MURI grant N00014-12-1-0911).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this paper
Cite this paper
Greco, G. et al. (2016). Singular Spectrum Analysis for Astronomical Time Series: Constructing a Parsimonious Hypothesis Test. In: Napolitano, N., Longo, G., Marconi, M., Paolillo, M., Iodice, E. (eds) The Universe of Digital Sky Surveys. Astrophysics and Space Science Proceedings, vol 42. Springer, Cham. https://doi.org/10.1007/978-3-319-19330-4_16
Download citation
DOI: https://doi.org/10.1007/978-3-319-19330-4_16
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-19329-8
Online ISBN: 978-3-319-19330-4
eBook Packages: Physics and AstronomyPhysics and Astronomy (R0)