• DOI:10.1093/mnras/sty422
• Corpus ID: 88503118

Cosmographic analysis with Chebyshev polynomials

@article{Capozziello2017CosmographicAW,
  title={Cosmographic analysis with Chebyshev polynomials},
  author={Salvatore Capozziello and Rocco D’Agostino and Orlando Luongo},
  journal={Monthly Notices of the Royal Astronomical Society},
  year={2017},
  volume={476},
  pages={3924-3938}
}

• S. Capozziello, R. D’Agostino, O. Luongo
• Published 12 December 2017
• Computer Science
• Monthly Notices of the Royal Astronomical Society

The limits of standard cosmography are here revised addressing the problem of error propagation during statistical analyses. To do so, we propose the use of Chebyshev polynomials to parameterize cosmic distances. In particular, we demonstrate that building up rational Chebyshev polynomials significantly reduces error propagations with respect to standard Taylor series. This technique provides unbiased estimations of the cosmographic parameters and performs significatively better than previous numerical approximations. To figure this out, we compare rational Chebyshev polynomials with Pad\'e series. In addition, we theoretically evaluate the convergence radius of (1,1) Chebyshev rational polynomial and we compare it with the convergence radii of Taylor and Pad\'e approximations. We thus focus on regions in which convergence of Chebyshev rational functions is better than standard approaches. With this recipe, as high-redshift data are employed, rational Chebyshev polynomials remain highly stable and enable one to derive highly accurate analytical approximations of Hubble's rate in terms of the cosmographic series. Finally, we check our theoretical predictions by setting bounds on cosmographic parameters through Monte Carlo integration techniques, based on the Metropolis-Hastings algorithm. We apply our technique to high-redshift cosmic data, using the JLA supernovae sample and the most recent versions of Hubble parameter and baryon acoustic oscillation measurements. We find that cosmography with Taylor series fails to be predictive with the aforementioned data sets, while turns out to be much more stable using the Chebyshev approach.