Slight Problem with the new ARTEMIS
Dear Friends, I found a slight problem with the new ARTEMIS - I use the latest package for ifeffit (1.2.3) and HORAE (0.24). When you do a fit and look at the fit results for the current fit it seems that the program does not update all fit statistics the first time you do a fit. For example, when I switch from a single-shell fit to a dual-shell mode (from 4 to 8 variables), I get still 4 variables in the first run. When I re-calculate, it updates everything correctly. It's just for the number of variables - the rest seems to be updated in the first run. By the way: The chi-square maps are doing well. I am at the state of applying them to my systems. The GUI is horrible still, but the algorithm works. Thanks again to Matt for coding the library which does this work for me :) Cheers, Norbert -- Dr. rer. nat. Norbert Weiher (norbertweiher@yahoo.de) Institute for Chemical and Bioengineering - ETH Hönggerberg HCI E 117 - 8093 Zürich - Phone: +41 1 63 3 48 32
Hi all,
I found a slight problem with the new ARTEMIS - I use the latest package for ifeffit (1.2.3) and HORAE (0.24).
When you do a fit and look at the fit results for the current fit it seems that the program does not update all fit statistics the first time you do a fit. For example, when I switch from a single-shell fit to a dual-shell mode (from 4 to 8 variables), I get still 4 variables in the first run. When I re-calculate, it updates everything correctly. It's just for the number of variables - the rest seems to be updated in the first run.
I believe this is not an ARTEMIS problem but an IFEFFIT bug: I have seen the same behaviour on updating of "#(fit variables)" while using GIFEFFIT. Best, Peter -- -------------------------------------------------------------- Peter Pfalzer Universitaet Augsburg Tel: +49-821-598-3215 Lehrstuhl fuer Experimentalphysik II Fax: +49-821-598-3411 Universitaetsstr. 1 D-86135 Augsburg Germany Peter.Pfalzer@physik.uni-augsburg.de --------------------------------------------------------------
Hi, On Tue, 7 Oct 2003, Peter Pfalzer wrote:
Hi all,
I found a slight problem with the new ARTEMIS - I use the latest package for ifeffit (1.2.3) and HORAE (0.24).
When you do a fit and look at the fit results for the current fit it seems that the program does not update all fit statistics the first time you do a fit. For example, when I switch from a single-shell fit to a dual-shell mode (from 4 to 8 variables), I get still 4 variables in the first run. When I re-calculate, it updates everything correctly. It's just for the number of variables - the rest seems to be updated in the first run.
I believe this is not an ARTEMIS problem but an IFEFFIT bug: I have seen the same behaviour on updating of "#(fit variables)" while using GIFEFFIT.
Yep, this is my fault. The feffit() command is not setting n_varys correctly after a fit. The minimize() command is doing this ok. It's an easy fix, and I should be able to post an update within a day or two. This may be somewhat unrelated, but I'll go into anyway: For what it's worth, n_varys is set _after_ the fit is completed (assuming it works correctly!). So if you do a fit, and then guess another variable, n_varys will still hold the number of fit variables in the most recent feffit() or minimize() fit. It's been suggested (by Bruce) that knowing the number of guessed set, and defined variables would also be helpful. I'll try to add this as well, but won't hold up the update for this... I'll also look into addressing the column label bug in the other thread as well. And finally, since there were no complaints about the Windows update procedure or the new feff6, I'll make that an automated update when I make these other fixes as well. Thanks for all the bug reports!! --Matt
On Tuesday 07 October 2003 05:18 am, Norbert Weiher wrote:
I found a slight problem with the new ARTEMIS - I use the latest package for ifeffit (1.2.3) and HORAE (0.24).
When you do a fit and look at the fit results for the current fit it seems that the program does not update all fit statistics the first time you do a fit. For example, when I switch from a single-shell fit to a dual-shell mode (from 4 to 8 variables), I get still 4 variables in the first run. When I re-calculate, it updates everything correctly. It's just for the number of variables - the rest seems to be updated in the first run.
I've not seent his behavior. Nor have I seen what Peter reported in his followup to Norbert's message. If you could make a zip file containing the project file, the data, and the feff.inp file, then I could look at this myself. Make sure to explain to me the steps you took to produce the problem. B -- Bruce Ravel ----------------------------------- ravel@phys.washington.edu Code 6134, Building 3, Room 222 Naval Research Laboratory phone: (1) 202 767 5947 Washington DC 20375, USA fax: (1) 202 767 1697 NRL Synchrotron Radiation Consortium (NRL-SRC) Beamlines X11a, X11b, X23b National Synchrotron Light Source Brookhaven National Laboratory, Upton, NY 11973 My homepage: http://feff.phys.washington.edu/~ravel EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/
Ciao Bruce, here are all files necessary to reconstruct the problem. I do the following: - open a data file - open a feff calculation - set all fitting parameters (window, kmin, kmax, etc) - include the first path and set the nr. of neighbours to 4 - fit it, keep s02 fixed to 0.82, the other parameters vary => 3 Variables. until here everything is fine. then I do: - add the 2nd path, set the nr. of atoms in path0001 to 1 and path0002 to 3 - insert the corresponding GUESSes - fit it. as i set the same e0 and keep s02 fixed, it should be 5 variables now, but artemis reports 3 :) when i redo the same fit, it reports the correct value. This behaviour seems to be uniform because I tested it with 3 datafiles already. Hope that helps to track the problem down. Cheers, Norbert -- Dr. rer. nat. Norbert Weiher (norbertweiher@yahoo.de) Institute for Chemical and Bioengineering - ETH Hönggerberg HCI E 117 - 8093 Zürich - Phone: +41 1 63 3 48 32
Sorry... here is the attachment ;) Norbert -- Dr. rer. nat. Norbert Weiher (norbertweiher@yahoo.de) Institute for Chemical and Bioengineering - ETH Hönggerberg HCI E 117 - 8093 Zürich - Phone: +41 1 63 3 48 32
participants (4)
-
Bruce Ravel -
Matt Newville -
Norbert Weiher -
Peter Pfalzer