residual in non-equally spaced data
Hope it'n not a known bug. Anyway, I have to report a strange behavior in Artemis 0.8.006 & Ifeffit 1.2.8 (everything updated today). I have extracted a chi(k) with EDA, which returns non-equally spaced data (equally-spaced in energy, I think), but the fix_chik macro in Artemis fixes everything (or so it seems) interpolating the data over an equally-spaced grid. Next, when I try to plot the residual functions, both in k and R-space, very weird things appear. Anyway, as expected, if I do myself the interpolation and I use the clean data, everything is ok. Bye, -- Francesco Giannici Dept. of Inorganic Chemistry University of Palermo giannici at pa dot ismn dot cnr dot it
Hi Francesco, Yes, I believe that Artemis calculates the residual incorrectly for chi(k) data which is not on the strictly uniform k-grid [0.0,0.05,0.10,...] Ang^-1. Ifeffit does the fit correctly (it internally interpolates the data to the uniform grid), and writes the best-fit result with this k-grid (and this is a feature, not a bug). Artemis calculates the residual itself, and doesn't check if the k-array for data and fit match. Sorry for the trouble, and hope that helps. Both Bruce and I have been swamped with other work the last year or so that such bug fixes take a lot of time to get fully resolved and released. --Matt
participants (2)
-
Francesco Giannici -
Matt Newville