Measurement of baryon acoustic oscillations in the Lyman-α forest fluctuations in BOSS data release 9

Anže Slosar, Vid Iršič, D. Kirkby, S. Bailey, Nicolás G. Busca, Timothée Delubac, J. Rich, É. Aubourg, Julian Bautista, Vaishali Bhardwaj, Michael Blomqvist, A. Bolton, Jo Bovy, Joel R. Brownstein, Bill Carithers, Rupert A. C. Croft, Kyle Dawson, Andreu Font-Ribera, J.m. Le Goff, Shirley Ho, Klaus Honscheid, Khee‐Gan Lee, Daniel Margala, Patrick McDonald, Bumbarija Medolin, Jordi Miralda‐Escudé, Adam D. Myers, Robert C. Nichol, P. Noterdaeme, N. Palanque‐Delabrouille, Isabelle Pâris, Patrick Petitjean, Matthew M. Pieri, Yodovina Piškur, Natalie A. Roe, Nicholas P. Ross, Graziano Rossi, David J. Schlegel, Donald P. Schneider, N. Suzuki, Erin Sheldon, Uroš Seljak, Matteo Viel, David H. Weinberg, Christophe YècheJournal of Cosmology and Astroparticle Physics 2013 · DOI · arXiv

Abstract

We use the Baryon Oscillation Spectroscopic Survey (BOSS) Data Release 9 (DR9) to detect and measure the position of the Baryonic Acoustic Oscillation (BAO) feature in the three-dimensional correlation function in the Lyman-alpha forest flux fluctuations at a redshift z=2.4. The feature is clearly detected at significance between 3 and 5 sigma (depending on the broadband model and method of error covariance matrix estimation) and is consistent with predictions of the standard LCDM model. We assess the biases in our method, stability of the error covariance matrix and possible systematic effects. We fit the resulting correlation function with several models that decouple the broadband and acoustic scale information. For an isotropic dilation factor, we measure 100x(alpha_iso-1) = -1.6 ^{+2.0+4.3+7.4}{-2.0-4.1-6.8} (stat.) +/- 1.0 (syst.) (multiple statistical errors denote 1,2 and 3 sigma confidence limits) with respect to the acoustic scale in the fiducial cosmological model (flat LCDM with Omega_m=0.27, h=0.7). When fitting separately for the radial and transversal dilation factors we find marginalised constraints 100x(alpha_par-1) = -1.3 ^{+3.5+7.6 +12.3}{-3.3-6.7-10.2} (stat.) +/- 2.0 (syst.) and 100x(alpha_perp-1) = -2.2 ^{+7.4+17}_{-7.1-15} +/- 3.0 (syst.). The dilation factor measurements are significantly correlated with cross-correlation coefficient of ~ -0.55. Errors become significantly non-Gaussian for deviations over 3 standard deviations from best fit value. Because of the data cuts and analysis method, these measurements give tighter constraints than a previous BAO analysis of the BOSS DR9 Lyman-alpha forest sample, providing an important consistency test of the standard cosmological model in a new redshift regime.

Introduction

The expansion history of the Universe is a fundamental pillar of modern cosmology. To probe this history, researchers utilize "standard rulers" like Baryon Acoustic Oscillations (BAO)—remnants of sound waves from the early universe. While BAO have been extensively measured using galaxy distributions at lower redshifts ($z < 1$), reaching higher redshifts is challenging as galaxies become increasingly difficult to detect.

This paper, a key product of the Baryon Oscillation Spectroscopic Survey (BOSS), provides a high-precision measurement of the expansion of the Universe when it was only about 3 billion years old ($z \approx 2.4$). By using the "Lyman-$\alpha$ forest"—a series of absorption lines in quasar spectra—as a tracer of large-scale structure, the authors successfully detect the BAO signal, confirming that the standard $\Lambda$CDM cosmological model holds firm even in this distant epoch.

Methods

Extracting a clean cosmological signal from the Lyman-$\alpha$ forest is technically demanding due to intrinsic quasar variability, instrumental noise, and metal contamination. The researchers employed several advanced statistical and data-processing techniques:

  • Continuum Fitting: Modeling the intrinsic spectrum of each quasar to isolate fluctuations caused by intervening neutral hydrogen gas.
  • Optimal Quadratic Estimator: A sophisticated statistical tool used to calculate the 3D correlation function, weighting data based on noise levels and the 1D power spectrum of the forest.
  • Distortion Marginalization: Addressing the loss of large-scale signals during continuum fitting by marginalizing over the specific data modes most affected by the fitting process.
  • Covariance Estimation: Utilizing both theoretical Fisher matrices and empirical bootstrapping (resampling the data) to ensure robust error estimation, accounting for non-Gaussianities in the data.

Data and Sample

The study utilizes the BOSS Data Release 9 (DR9), part of the Sloan Digital Sky Survey III (SDSS-III). The primary sample consists of 58,227 quasars selected after rigorous quality cuts, such as removing Broad Absorption Line (BAL) quasars and spectra with low signal-to-noise ratios.

Sky distribution of the BOSS DR9 quasar sample used in the analysis.

The distribution of these quasars across the sky and their redshift range are critical for the 3D correlation analysis. The Lyman-$\alpha$ forest pixels used in the analysis typically fall in the redshift range $2.0 < z < 3.5$, providing a dense sampling of the high-redshift universe.

Redshift distribution of the quasars (QSOs) and the Lyman-alpha pixels used in the study.

Key Results and Insights

The analysis yielded a clear detection of the BAO feature with a significance between 3 and 5 sigma, depending on the specific model used.

  • Isotropic Measurement: The measured isotropic dilation factor $\alpha_{iso}$ was found to be $100 \times (\alpha_{iso} - 1) = -1.6^{+2.0}_{-2.0} \text{ (stat.)} \pm 1.0 \text{ (syst.)}$, indicating the scale is within ~1.6% of the flat $\Lambda$CDM prediction.
  • Anisotropic Constraints: Fitting for radial ($\alpha_{\parallel}$, probing the Hubble expansion rate) and transverse ($\alpha_{\perp}$, probing angular diameter distance) components provided more granular constraints. These components are anti-correlated with a coefficient of $\sim -0.55$.
  • Cosmological Implications: The results are entirely consistent with a standard flat $\Lambda$CDM model ($\Omega_m = 0.27$), offering no evidence for exotic "early dark energy" or modified gravity at $z=2.4$.

The study reinforces the Lyman-$\alpha$ forest's role as a primary tracer for large-scale structure cosmology. By using "looser" data cuts than previous studies—keeping more of the forest even if slightly noisier—the authors significantly increased their signal-to-noise ratio, demonstrating that the forest is a powerful tool for precision cosmology.

Conclusion and Future Work

While successful, the study identifies areas for future refinement. Metal contamination (e.g., Si III) creates "shadow" correlations that must be modeled more precisely in future higher-precision surveys. Additionally, the non-Gaussian nature of errors at high significance suggests a need for more complex Bayesian analysis rather than simple $\chi^2$ statistics. This work sets a robust foundation for subsequent BOSS releases (DR11, DR12) and upcoming experiments like DESI, which will provide even larger datasets for testing the expansion rate of the Universe.

Summary updated

Abstract

We use the Baryon Oscillation Spectroscopic Survey (BOSS) Data Release 9 (DR9) to detect and measure the position of the Baryonic Acoustic Oscillation (BAO) feature in the three-dimensional correlation function in the Lyman-alpha forest flux fluctuations at a redshift z=2.4. The feature is clearly detected at significance between 3 and 5 sigma (depending on the broadband model and method of error covariance matrix estimation) and is consistent with predictions of the standard LCDM model. We assess the biases in our method, stability of the error covariance matrix and possible systematic effects. We fit the resulting correlation function with several models that decouple the broadband and acoustic scale information. For an isotropic dilation factor, we measure 100x(alpha_iso-1) = -1.6 ^{+2.0+4.3+7.4}_{-2.0-4.1-6.8} (stat.) +/- 1.0 (syst.) (multiple statistical errors denote 1,2 and 3 sigma confidence limits) with respect to the acoustic scale in the fiducial cosmological model (flat LCDM with Omega_m=0.27, h=0.7). When fitting separately for the radial and transversal dilation factors we find marginalised constraints 100x(alpha_par-1) = -1.3 ^{+3.5+7.6 +12.3}_{-3.3-6.7-10.2} (stat.) +/- 2.0 (syst.) and 100x(alpha_perp-1) = -2.2 ^{+7.4+17}_{-7.1-15} +/- 3.0 (syst.). The dilation factor measurements are significantly correlated with cross-correlation coefficient of ~ -0.55. Errors become significantly non-Gaussian for deviations over 3 standard deviations from best fit value. Because of the data cuts and analysis method, these measurements give tighter constraints than a previous BAO analysis of the BOSS DR9 Lyman-alpha forest sample, providing an important consistency test of the standard cosmological model in a new redshift regime.
BibTeX
@article{slosar2013measurement,
  author = {Anže Slosar and Vid Iršič and D. Kirkby and S. Bailey and Nicolás G. Busca and Timothée Delubac and J. Rich and É. Aubourg and Julian Bautista and Vaishali Bhardwaj and Michael Blomqvist and A. Bolton and Jo Bovy and Joel R. Brownstein and Bill Carithers and Rupert A. C. Croft and Kyle Dawson and Andreu Font-Ribera and J.M. Le Goff and Shirley Ho and Klaus Honscheid and Khee‐Gan Lee and Daniel Margala and Patrick McDonald and Bumbarija Medolin and Jordi Miralda‐Escudé and Adam D. Myers and Robert C. Nichol and P. Noterdaeme and N. Palanque‐Delabrouille and Isabelle Pâris and Patrick Petitjean and Matthew M. Pieri and Yodovina Piškur and Natalie A. Roe and Nicholas P. Ross and Graziano Rossi and David J. Schlegel and Donald P. Schneider and N. Suzuki and Erin Sheldon and Uroš Seljak and Matteo Viel and David H. Weinberg and Christophe Yèche},
  title = {{Measurement of baryon acoustic oscillations in the Lyman-α forest fluctuations in BOSS data release 9}},
  journal = {Journal of Cosmology and Astroparticle Physics},
  year = {2013},
  doi = {10.1088/1475-7516/2013/04/026},
}

Loading related content...

Figure