Hi all, I'm one of the people who worked on the analysis for the LCLS data. I only recently signed up for this list, so I'll start by saying "Hi". It's nice to see a lot of familiar names around here already.
Regarding "pseudo-twinning": > Though ~15440 is a big number, and a twinning factor of 0.5 is > obtained, should data processing for all sets with randomly 1 out of > 2 indexing, followed by reindexing all in the other way, and then > merging all 2X~15440 give a better control of perfect twinning? The reality is even simpler than this. If the point group of the structure is known (which it was in this test case, of course), then we know for sure what the apparent symmetry after applying all allowable "twinning operators" in exactly equal proportions will be. We simply merge according to this higher apparent symmetry instead of the lower true symmetry, and get an exactly known "twin fraction" of 50%. In other words, we already do something very similar to what you suggest. Tom -- Thomas White <thomas.wh...@desy.de> Direct telephone: +49 (0) 40 8998-5786
