Improved two-point correlation function estimates using glass-like distributions as a reference sample

All estimators of the two-point correlation function are based on a random catalogue, set points with no intrinsic clustering following selection survey. High-accuracy estimates require use large catalogues, which imply high computational cost. We propose to replace standard catalogues by glass-like point distributions or glass characterized power spectrum $P(k)\propto k^4$ and exhibit significantly less than Poisson distribution same number scales larger mean inter-particle separation. show that these can be obtained iteratively applying technique Zeldovich reconstruction commonly used in studies baryon acoustic oscillations (BAO). provide modified version widely Landy-Szalay estimator adapted compare its performance results using samples. Our samples do not add any bias respect distributions. On separation leads significant reduction variance Legendre multipoles $\xi_\ell(s)$ points. The size catalogue required achieve given accuracy is smaller when Even considering small additional cost constructing their could help drastically reduce configuration-space analysis future surveys while maintaining high-accuracy requirements.