### [1004.0221] High resolution spectroscopy of the three dimen

Posted:

**April 06 2010**This is an interesting paper. They have a bunch of close pairs of quasars and then they do various 2-point statistics (which they phrase in somewhat complicated manners, but it is really just correlation function at various angles). They also do some higher-order statistics.

What I find most intriguing is their claim that peculiar velocities in absorbing gas are smaller than 100 km/s. It is not clear, where exactly is this number coming from. Figure 4 compares auto and cross-correlation functions. Auto-correlation function is essentially correlation function at mu=1 and cross-correlation is integrated over all the other angles (and would give monopole in the limit of many pairs, but they are not quite there). In any case, in the presence of just Kaiser compression, one would expect auto-correlation function to be always larger than cross-correlation function. Since they have high SNR spectra, they must be dominated by the sample variance, but since the two have been measured over the *same* volume, this errors is irrelevant and I think that auto-correlation should be larger than cross-correlation. The only way to wash it out is to have peculiar velocities. So, from the fact that the auto-corr is lower than cross-corr, I'd say you could put a lower limit on the pec velocities, not the upper one. But I guess you can always check this with simulations....

Also, it would be good to get some more formal chi2 to compare simulations with data... As always with corr func - errorbars are heavily correlated (as one can see from the fact that scatter in the points is always much smaller than errorbars).

What I find most intriguing is their claim that peculiar velocities in absorbing gas are smaller than 100 km/s. It is not clear, where exactly is this number coming from. Figure 4 compares auto and cross-correlation functions. Auto-correlation function is essentially correlation function at mu=1 and cross-correlation is integrated over all the other angles (and would give monopole in the limit of many pairs, but they are not quite there). In any case, in the presence of just Kaiser compression, one would expect auto-correlation function to be always larger than cross-correlation function. Since they have high SNR spectra, they must be dominated by the sample variance, but since the two have been measured over the *same* volume, this errors is irrelevant and I think that auto-correlation should be larger than cross-correlation. The only way to wash it out is to have peculiar velocities. So, from the fact that the auto-corr is lower than cross-corr, I'd say you could put a lower limit on the pec velocities, not the upper one. But I guess you can always check this with simulations....

Also, it would be good to get some more formal chi2 to compare simulations with data... As always with corr func - errorbars are heavily correlated (as one can see from the fact that scatter in the points is always much smaller than errorbars).