tag:blogger.com,1999:blog-193178369124442285.post34771604803771728..comments2022-11-12T23:40:40.957-08:00Comments on Jessica Kirkpatrick Research: Using the BOSS Randomsberkeleyjesshttp://www.blogger.com/profile/00255252868558666987noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-193178369124442285.post-38206919479168955372011-03-04T07:57:03.502-08:002011-03-04T07:57:03.502-08:00Jess, this is great news, I am so happy we now hav...Jess, this is great news, I am so happy we now have a matching set of data and randoms!! Awesome :) I think the next steps are (roughly in order of importance):<br /><br />1) Compute the 2DCF and see if it still looks weirdly flat, and negative in strange places. <br /><br />2) Ask Shirley or Eric to compute 2DCF and 3DCF with their code so you can double check what your code is doing. Be aware, if they are computing it in subdivisions so as to get bootstrapped error bars, there can be a difference of a few percent form computing the CF in the whole volume. It shouldn't be this way (the randoms should compensate) but it just is. I've spent weeks obsessing about this, but I don't think you need to. <br /><br />3) Do some of the tests we talked about on the 3DCF: check sensitivity to bin width (narrowness) and shift the center around by a few percent to see if there are big changes in the measured 3DCF. <br /><br />4) Run the CFs through the reconstruction code. Do this last, I want to check the CFs to make sure we REALLY REALY believe them before we spend a bunch of time fiddling with the reconstruction parameters. The reconstruction won't work if the CFs are wrong.Alexiahttps://www.blogger.com/profile/06694777215874417074noreply@blogger.com