On 20 June 2013 20:46, Douglas Theobald <dtheob...@brandeis.edu> wrote:
> Well, I tend to think Ian is probably right, that doing things the > "proper" way (vs French-Wilson) will not make much of a difference in the > end. > > Nevertheless, I don't think refining against the (possibly negative) > intensities is a good solution to dealing with negative intensities --- > that just ignores the problem, and will end up overweighting large negative > intensities. Wouldn't it be better to correct the negative intensities > with FW and then refine against that? > > Hmmm, I seem to recall suggesting that a while back (but there were no takers!). I also think that using corrected Is, as opposed to corrected Fs, (however you choose to do it) is the right way to do twinning & other statistical tests. For example the Padilla/Yeates L test uses the cumulative distribution of |I1 - I2| / (I1 + I2) where I1 & I2 are intensities of unrelated reflections (but close in reciprocal space). The denominator of this expression is clearly going to have problems if you feed it negative intensities! Also I believe (my apologies if I'm wrong!) that the UCLA twinning server obtains the Is by squaring the Fs (presumably obtained by F-W). This is a formally invalid procedure (the expectation of I is not the square of the expectation of F). See here for an explanation of the difference: http://xtal.sourceforge.net/man/bayest-desc.html . Cheers -- Ian
