PC 1643+4631A,B:: The Lyman-α forest at the edge of coherence

This is the first measurement and detection of coherence in the intergalactic medium (IGM) at substantially high redshift (z similar to 3.8) and on large physical scales (similar to 2.5 h(70)(-1) Mpc). We perform the measurement by presenting new observations of the high-redshift quasar pair PC 1643+4631A, B and their Ly alpha-absorber coincidences. With data collected from Keck I Low Resolution Imaging Spectrometer (LRIS) in a 10,200 s integration, we have full coverage of the Ly alpha forest over the redshift range 2.6 < z < 3.8 at a resolution of 3.6 angstrom (similar to 220 km s(-1)). This experiment extends multiple sight line quasar absorber studies to higher redshift, higher opacity, larger transverse separation, and into a regime where coherence across the IGM becomes weak and difficult to detect. Noteworthy features from these spectra are the strong damped Ly alpha absorbers (DLAs) just blueward of both Ly alpha emission peaks, each within 1000 km s(-1) of the emission redshift but separated by 2500 km s(-1) from each other. The coherence is measured by fitting discrete Ly alpha absorbers and by using pixel flux statistics. The former technique results in 222 Ly alpha absorbers in the A sight line and 211 in B. Relative to a Monte Carlo pairing test (using symmetric, nearest-neighbor matching) the data exhibit a 4 sigma excess of pairs at low velocity splitting (Delta v< 150 km s(-1)), thus detecting coherence on transverse scales of similar to 2.5 h(70)(-1) Mpc. We use spectra extracted from a smoothed particle hydrodynamic (SPH) simulation to analyze symmetric pair matching, transmission distributions as a function of redshift and compute zero-lag cross-correlations to compare with the quasar pair data. The simulations agree with the data with the same strength (similar to 4 sigma) at similarly low velocity splitting above random chance pairings. In cross-correlation tests, the simulations agree when the mean flux (as a function of redshift) is assumed to follow the prescription given by Kirkman et al. (2005). While the detection of flux correlation (measured through coincident absorbers and cross-correlation amplitude) is only marginally significant, the agreement between data and simulations is encouraging for future work in which even better quality data will provide the best insight into the overarching structure of the IGM and its understanding as shown by SPH simulations.