May 2003 - Ifeffit - millenia.cars.aps.anl.gov

new versions of Athena and Artemis using Ifeffit 1.2.0
by Bruce Ravel 16 May '03

16 May '03

Hi folks, Today I posted a new source code tarball for Athena and Artemis to my web site. I did *not* post new Windows executables, however. Linux, Unix, and OSX users should first upgrade Ifeffit by grabbing the Ifeffit 1.2.0 tarball from Matt's web site and then grabbing the tarball from my web site or run the horae_update script. Windows users should wait until Matt announces the new Windows installer (which is almost ready to be released) and download that. Future Windows executables of Athena and Artemis will assume that you have already updated the base Ifeffit package. New stuff this time: 1. The horae_update script is a bit smarter in that it compares the version on the server to the version on your computer and does nothing if you seem to be up to date. 2. New features of Ifeffit 1.2.0 are used, such as rebinning in Athena and fits in Artemis with multiple-data-sets, multiple-k-weights, and/or background co-refinements. 3. Better reporting of ill-posed variables in Artemis (based on Ifeffit's reporting of the same). 4. Better reporting of best-fit path lengths after a fit finishes. 5. Check for mismatched parentheses in math expressions in Artemis (a HUGE problem in prior versions). 6. Examples of multiple-data-set fits in the form of artemis project files. Look for the files examples/artemis/EuTiO3/eto.apj and examples/artemis/Cu/temparatures.apj in the source code tarball or C:\Program Files\Ifeffit\examples\EuTiO3\eto.apj and C:\Program Files\Ifeffit\examples\Cu\temparatures.apj on Windows. Be sure to read the comments in the project journals of those two files. + several bug fixes Regards, B -- Bruce Ravel ----------------------------------- ravel(a)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, X24c, U4b 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/

1 0

Artemis questions
by Bruce Ravel 16 May '03

16 May '03

Hi Carlo, Hope you don't mind my responding to the list. These are questions that I think will be of general interest. I am actually going to split my answers into two emails, both of which will still be rather long -) > I have been working with Artemis over the past few days and I have a > couple of questions that I cannot seem to find answers to in the manuals > for FEFFIT and Artemis. OK! You have me up against the wall ;-) Documentation is currently the Artemis' great shortcoming. I apologize that the doc is coming along slowly. FWIW, I have been thinking alot about the doc, but haven't found a lot of time to write. > 1. What is the difference between a SET and a DEF quantity. I suspect > that it is an issue of being updated or not but I am not sure. A SET parameter is evaluated at the start of a fit and remains invarient throughout the fit. A DEF parameter is re-evaluated at every iteration of the fit. Typically a def parameter depends on one or more guess parameters. Let's consider a simple example. Suppose that we have a crystal with a nominal lattice constant "a0" and we want to consider the possibility in the fit that the actuall lattice constant deviates from "a0". To do so, we may consider a parameter "delta_a": set a0 3.61 guess delta_a 0.0 Now suppose we want to use "a" (the measured lattice constant) in math expressions for some of our path parameters. Do we say set a a0 + delta_a or def a a0 + delta_a ? The answer is the second one. If make "a" a SET parameter, then a will be fixed to 3.61 at the beginning of the fit and not change. If, however, we make "a" a DEF parameter, then it will be updated every time that "delta_a" is updated during the fit. At the risk of confusing the issue, I feel I should make a connection with Matt's old feffit program. In the old feffit, set parameters behaved like Ifeffit's def parameters. That is, in the old feffit, set parameters were updated throughout the fit. As a rule of thumb for Artemis, if your parameter is a number, make it a set parameter. If it is a math expression, make it a def parameter. HTH, B -- Bruce Ravel ----------------------------------- ravel(a)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, X24c, U4b 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/

1 0

Anyone out there do SEXAFS?
by k-kupiecki＠northwestern.edu 15 May '03

15 May '03

HI, I'm trying to analyze my SEXAFS data... if anyone out there has done this before could you please drop me a line at k-kupiecki(a)northwestern.edu.... I could really use someone that I can directly question and bounce procedures off of.... Inparticular I'm currently looking for suggestions for how to simulate with FEFF SEXAFS in the "grazing incidence" orientation (with the E vector of the synchrotron beam close (~0.2 deg) to the surface normal of the sample (incident beam ~.2 deg incident on surface). Thus far I've been trying to analyze my samples with SEXAFS taken from a "normal geometry" (~0.2 angle incident to the beam and the E vector of the synchrotron beam in the plane of the surface)... I managed to simulate this with FEFF by giving it a hemispherical cluster representing my surface... But I'm a little confused what I need to change in FEFF to replicate the information coming from just the inter planar bonding (in "grazing incidence") instead of both inter and intra planar bonding as found in the "normal incidence" situation.... Any suggestions would be greatly appreciated! Thanks, Kristine (Kupiecki) Witkowski Northwestern University

3 2

FEFF6 errors?
by Carlo U. Segre 15 May '03

15 May '03

Matt: I have been having problems with the FEFF6 that is distributed with Ifeffit. If I try to run the executable with the feff.inp the TEST directory, it gives the following output ---- copper Calculating potentials and phases... free atom potential and density for atom type 0 free atom potential and density for atom type 1 overlapped potential and density for unique potential 0 overlapped potential and density for unique potential 1 muffin tin radii and interstitial parameters phase shifts for unique potential 0 phase shifts for unique potential 1 Preparing plane wave scattering amplitudes... nncrit in prcrit 9 Searching for paths... Rmax 0.0000 keep and heap limits 0.0000000 0.0000000 Preparing neighbor table nfound nheap nheapx nsc r Internal path finder limit exceeded -- path list may be incomplete. Paths found 0 (nheapx, nbx 1 0) Eliminating path degeneracies... Plane wave chi amplitude filter 2.34% STOP no input statement executed ---- It only generates the 4 files: geom.dat paths.bin paths.dat phase.bin If I look at paths.dat, I see: ---- copper Rmax 0.0000, keep limit 0.000, heap limit 0.000 ^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@ Plane wave chi amplitude filter 2.34% ------------------------------------------------------------------------------- ---- I am a bit confused about this because when building the Debian package, everything compiles properly and all the libraries seem to be there properly. Any ideas? Carlo -- Carlo U. Segre -- Professor of Physics Associate Dean for Research, Armour College Illinois Institute of Technology Voice: 312.567.3498 Fax: 312.567.3494 Carlo.Segre(a)iit.edu http://www.iit.edu/~segre

2 1

Athena - plotting limits
by Carlo U. Segre 15 May '03

15 May '03

Bruce: Why does Athena only plot from 0-15 in k-space even though there is data beyond that. Now that I look back at the mu(E) plots, they are restricted in energy too. I understand that we would normally not use data to far beyond k=15 but it still might be nice to see it. Perhaps, I am just not seeing some kind of parameter that could be set? Carlo -- Carlo U. Segre -- Professor of Physics Associate Dean for Research, Armour College Illinois Institute of Technology Voice: 312.567.3498 Fax: 312.567.3494 Carlo.Segre(a)iit.edu http://www.iit.edu/~segre

2 1

ifeffit 1.2.0
by Matt Newville 15 May '03

15 May '03

Hi Everyone, At the suggestion of several people, I've repackaged the latest version of Ifeffit and called it version 1.2.0, to indicate that this is the stable version. By this, I mean that bugs and errors will be fixed in this series, but that no new features will be added. As of now, only the source kit for ifeffit-1.2.0 is available from the Ifeffit web site. The Windows installer package should be available within a few days. In addition to this, it was recommended to start using the SourceForge.net services. This will happen, and an empty page for Ifeffit is available there at http://sourceforge.net/projects/ifeffit/ It may take much of the summer to get this working completely. This will provide multiple download sites with mirroring. In addition, the development/unstable versions will be available through Sourceforge's CVS server -- if anyone is interested in this, letting me know! There are other services (such as bug tracking) available from Sourceforge as well, and it is possible that the mailing list may eventually move to sourceforge.net as well. Thanks, --Matt

1 0

upcoming EXAFS course at NSLS
by Bruce Ravel 12 May '03

12 May '03

Hi folks, It occurs to me that not everyone who reads this list is or ever has been a user of NSLS, so you may not be aware of an upcoming event. This summer, Simon Bare and I are running a course on EXAFS Data Collection and Analysis at the NSLS. The course will be July 14-17 and will include lectures, beamline practicals, and lots of time in front of computers using atoms, feff, ifeffit, athena, and artemis. The speaker list includes eight 75 minute lectures on a variety of topics covering most aspects of XAS measurement and EXAFS analysis. The speaker list is really great this year and includes some familiar names from our little list, including Matt, Simon, Shelly Kelly, Scott Calvin, and John Rehr. Last year's course was lots of fun for everyone involved. Here is the URL: http://nslsweb.nsls.bnl.gov/nsls/org/workshops/exafs/exafs_workshop.htm We have limited space and received about a factor of two too many application last year. Still, it would be great to see some of you folks this summer at the NSLS! Or you might pass the URL along to someone else in your research group who is looking for an intense but fun introduction to XAFS. Regards, B -- Bruce Ravel ----------------------------------- ravel(a)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, X24c, U4b 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/

1 0

Re: [Ifeffit] Re: ifeffit_1.0077 compilation problem
by Carlo U. Segre 09 May '03

09 May '03

Matt: I very much like the idea of a "stable" and "testing" line for the programs. I could keep both of them updated in my Debian packages. Carlo -- Carlo U. Segre -- Professor of Physics Associate Dean for Research, Armour College Illinois Institute of Technology Voice: 312.567.3498 Fax: 312.567.3494 Carlo.Segre(a)iit.edu http://www.iit.edu/~segre

1 0

RE: [Ifeffit] Re: ifeffit_1.0077 compilation problem
by SCHLEGEL Michel 177447 09 May '03

09 May '03

Dear Everybody, > I can then start a beta series for new features and build > systems, and expect only beta testers to be affected. Any > comments on this? Just a suggestion (discard it right away if it looks like it will take too much time): Why not copying the well-proved Linux bumbering of versions, and reserve odd numbers for test versions, and even numbers for "stable" versions? This would allow simple laymen to benefit from the outstanding capabilities of Ifeffit, while allowing the "clergy" to summon the debugging deities on well-iddntified newborn versions... Best regards, Michel Schlegel

1 0

Re: ifeffit_1.0077 compilation problem
by Matt Newville 08 May '03

08 May '03

Hi Paul, I'm responding to this one on the mailing list. As many people have noticed, there are several problems with the 1.0077 distribution of Ifeffit. It seems that it should be treated as a 'test' version. Most of the problems are small from my end (that is, easy to fix, even if show-stoppers for functionality or building). I'll be changing the links to ifeffit.tar.gz to point to versions ifeffit-1.0077a.tar.gz, ifeffit-1.0077b.tar.gz, .... over the next few days for testing. If you're brave, you can try the latest test version and let me know what's not working. Otherwise, we should have a stable, working 1.0078 sometime next week, possibly with some of the wrappers using some of the new features. As more people are using Ifeffit/Athena/Sixpack for real work, it seems that having a stable version is good idea. So I'm leaning toward calling the next release (ie, 1.0078) Ifeffit-1.100 and only fixing bugs, and build problem for this version so that this can be the 'stable, recommended' version, and can get better tests and documentation to really reflect this stable version. I can then start a beta series for new features and build systems, and expect only beta testers to be affected. Any comments on this? On Thu, 8 May 2003, Paul Fons wrote: > In attempting to compile 1.0077 of ifeffit, I encountered some > unresolved symbols. A grep search seems to indicate that atomic.f has > the definitions, but it doesn't seem to be in the macosx makefile. > Any suggestions? Yes, two suggestions: 1. Run 'configure; make ' work from the base ifeffit directory. If this doesn't work, please let me know what fails. I'd much refer this to work, and it seems close. 2. Change line 134 of src/macosx/makefile.in to read $(srcdir)/iff_unguess.o $(srcdir)/iff_window.o $(srcdir)/atomic.o and re-run 'configure ; make'. > I am doing the build using the Makefile in src/macosx. One more thing, > under 10.2 I had some problems with overlapping definitions in multiple > libraries (in Apple's and Fink). The solution was to delete the "-l > termcap" as well as the "-L/usr/lib" from the library options in the > makefile. As I have yet to get things to link successfully, I will > send you the final modifications when I resolve the symbols (at_kedge, > etc.) problem. Which is the preferred version: Apple's or Fink's? Is there any hope of getting a binary distribution for Mac OS X or at least clear and complete instructions for how to do a complete installation using Fink? --Matt

2 1