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

new releases of Athena and Artemis
by Bruce Ravel 18 Sep '02

18 Sep '02

Hi folks, Today I posted new source tarballs of Athena and Artemis on my web site. A new Windows installer will be available later today from Matt's web site. This time there are some big changes. 1. I am now distributing Athena and Artemis in a single source tarball called `horae-001.tar.gz'. http://homepage.mac.com/cparada/GML/HORAE.html 2. Windows users should download and install the new installer package because there are several new library and example files. I will continue the practice of posting replacement executables when I _next_ update Athena and Artemis. 3. I have started numbering Athena and Artemis differently. Both programs are getting within striking distance of becoming the magical 1.0. Because of that and other reasons, I want to start using actual version numbers rather than dates to identify the versions. Athena is starting at 0.8 and Artemis at 0.5. The version number on the source tarball will simply be an incremented counter unrelated to the version numbers of the programs. 4. I have made a very large number of changes to Artemis. I'll refer you to the change logs for all the details. In addition to many, many squished bugs, the most important change is that reading in feffNNNN.dat files now works in a much more sensible manner. You can read in individual feffNNNN.dat files or read in all at once by selecting the feff.inp file. In either case, the feffNNNN.dat files should be grouped in a sensible manner in the list in the skinny part of the window. 5. The other big new feature in Artemis is templates. This is a scheme I am working on to help automate the creation of fitting models. The idea is that after reading in data and a feff calculation, you can select, for example, "FirstShell" from the Templates menu. Artemis reads an external file which described a prototypical first shell fit and generates guess, set, and path parameters that hopefully are close to an appropriate fitting model. For copper data, you can hit the big green fit button right after reading in the template file and get a reasonable fit. Some tweaking is usually necessary, but the templates are a decent start. 6. The most interesting change to Athena is that you can now save your data as a "chi(k) record". This is a special file that has the same structure as a normal Athena project file, but which contains only a single entry. That entry is saved as chi(k) rather than as mu(E). The purpose of this kind of file is for reading into Artemis. The advantage of a chi(k) record over chi(k) column data is that the values for things like the Fourier transform parameters that you chose in Athena will be set in Artemis. As always, let me know about problems and feature requests, Enjoy, Bruce -- 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

athena and artemis on Mac OS X 10.2
by Dr. Paul Fons 06 Sep '02

06 Sep '02

I have successfully installed the latest tarballs for athena and artemis on Mac OS 10.2 (Jaguar) and thought I would share with the list the necessary steps to do the installation. 0. Install 10.2 and the latest development tools (included with 10.2) 1. install the latest patch for xterm from the sourceforge site: ceforge.net/projects/xonx/ and install Xfree86 4.2.0.1 2. rebuild the readline libraries within fink after updating from cvs (I use the fink unstable branch) 3. recompile Tk after making the following modification: in Tk/MMutil.pm replace the subroutine const_config with the following code: sub const_config { my $self = shift; my $name; foreach $name (grep /(%|\.(old|bak|q4|orig|rej))$/,keys %{$self->{PM}}) { delete $self->{PM}->{$name}; } my $flags = $self->{'CCCDLFLAGS'}; $flags =~ s/(-[fK]?\s*)pic\b/${1}PIC/; $self->{'CCCDLFLAGS'} = $flags; if ($^O eq 'MSWin32' && $Config{'ccflags'} =~ /-DPERL_OBJECT/) { $self->{'LDFLAGS'} =~ s/-(debug|pdb:\w+)\s+//g; $self->{'LDDLFLAGS'} =~ s/-(debug|pdb:\w+)\s+//g; } elsif ($^O eq 'darwin' ) { $self->{'LDDLFLAGS'} =~ s/-flat_namespace\s+//; $self->{'LDDLFLAGS'} =~ s/-undefined\s+suppress\s+//; if ( -e "$Config{'archlib'}/CORE/$Config{'libperl'}" ) { $self->{'LDDLFLAGS'} .= " -L\${PERL_ARCHLIB}/CORE -lperl "; } elsif ( -e "/System/Library/Perl/darwin/CORE/libperl.dylib" ) { $self->{'LDDLFLAGS'} .= " -L/System/Library/Perl/darwin/CORE -lperl "; } else { warn "Can't find libperl.dylib"; } $self->{'LDFLAGS'} =~ s/-flat_namespace\s+//; $self->{'LDFLAGS'} =~ s/-undefined\s+suppress\s+//; } elsif ($^O =~ /(openbsd)/i) { # -Bforcearchive is bad news for Tk - we don't want all of libpTk.a in all .s o-s. $self->{'LDDLFLAGS'} =~ s/-Bforcearchive\s*//g; } return $self->MM::const_config; } The only thing that is different is a new conditional block in a if block, but I thought it would be easier just to copy and paste. This hack was taken from Bug's Macintosh Page at http://www.lehigh.edu/~sol0/Macintosh/X/ptk/. It turns out it is no longer necessary to hack the dynamic library loader dylib. After changing the subroutine const_config in MMutil.pm, mumble the usual incantations: perl Makefile.PL make make test sudo make install 4. rebuild the ifeffit source cd ifeffit/src/macosx make clean make sudo make install cd ../../wrappers/perl perl Makefiile.PL make sudo make install cd ../python sudo python setup.py install 5. Rebuild athena and artemis cd athena perl Makefile.PL make sudo make install cd artemis perl Makefile.PL make sudo make install You are done! Caveats: I (being perverse) have not done the installation exactly this way. I have compiled xfree86 4.2.99.1 from the online cvs sources and have in addition installed perl 5.8.0. This is a little more complicated (but not much). I also found that ifeffit was looking for a library libg2c.a and made a symbolic link from /usr/lib/gcc/darwin/2.95.2/libg2c.a to /usr/local/lib so that the ifeffit compilation could proceed normally. I am not sure if this step is necessary for all. I don't think perl 5.8.0 and the newest build of Xfree86 are necessary (but on the other hand that is all I have and can't check the alternatives). The version of perl installed by 10.2 is still 5.6. Any questions? Send me email if you have problems and I will help if I can. Dr. Paul Fons Photonics Research Institute National Institute for Advanced Industrial Science and Technology Ministry of Economics and Trade Umezono 1-1-1 Tsukuba, Japan 305-8568 tel. 81-298-61-5636 fax. 81-298-61-5615 paul-fons(a)aist.go.jp Lines below are in Japanese 独立行政法人産業技術総合研究所光技術部門〒305-8568 茨城県つくば市梅園1-1-1 産総研つくば中央第2 E-mail: paul-fons(a)aist.go.jp, Tel. 0298-61-5636, Fax. 0298-61-5615

1 1

athena and artemis updates
by Bruce Ravel 26 Aug '02

26 Aug '02

Hi Folks, I have just posted new source tarballs and windows executables for both athena and artemis. Both packages have significant new features and bug fixes. Please read the two special notes at the end of this announcement regarding a couple of bugs that I know about. I will be on vacation the rest of this week -- so I don't want anyone finding new bugs until next week! Athena: 1. fixed a problem that prevented one from plotting a chi(k) record (such as from feff's chi.dat file) 2. fixed a problem in which unmatched parentheses in the header of a data file would seemingly cause Athena to hang. Please note, that this was fixed by having Athena append the necessary close parens to title lines with unmatched open parens (i.e. by having Athena alter some title lines!) 3. Several usability improvements, including ability to specify a data or project file from the command line, remembering preprocessing parameters between files, and remembering the name of an input project file and using it as the default for saving a project file. 4. When records are saved as mu(E) or norm(E) column data, the e0 shift is applied before saving 5. project files now identify the operating system, perl version, and ifeffit version used to create the file. This makes project files more useful to me when submitting them with bug reports. (This was done without affecting compatibility in either direction.) (This is also a new feature in Artemis.) 6. several minor bug fixes Artemis: 1. The big box on the feff page is now filled with a schematic interpretation of the feff calculation. SS and colinear MS paths are highlighted and right-mouse clicking over a path description pops up a useful menu http://leonardo.phys.washington.edu/~ravel/software/exafs/images/artemis/ar… 2. Began work on a templating system. This means that you can read in some data and a feff calculation, then select, say, the "FirstShell" template and Artemis will fill in guess, set, and path parameters with a model that should run to completion and may even be a decent stab at the fit you want to do. Note that the "Templates" menu does nothing on Windows at this time. Although this is undocumented, you can poke at it to see how it works. 3. Changed the color of the "fit" and "ff2chi" buttons. (See the screenshot mentioned above) 4. Several bug fixes and minor usability improvements. For complete details see the change logs in the tarballs or at http://leonardo.phys.washington.edu/~ravel/software/exafs/notes/ OSX users: It is the intended behavior of Athena and Artemis to remember the path whenever you use a file dialog and to use that path as the default the next time you use a file dialog. I am told that Athena does not behave this way on OSX. I am aware of the problem, but have not yet worked to fix it. All users: On your computer, there are probably a few seconds between the outline of Athena or Artemis window appearing on your screen and having all the features of the window be completely rendered. In that time, it is tempting to "jump the gun" and hit control-f so that the file dialog will appear just as soon as perl/Tk finishes rendering all the widgets. Doing this cause a minor race condition that will impede Athena's ability to monitor how much memory you are using and cause many scary-looking messages to be written to standard output. I haven't figured out how to prevent one from doing this, so my advice for the moment is to be patient and let the program finish rendering before using keyboard shortcuts. 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

athena and artemis under Mac OS X 10.2
by Dr. Paul Fons 26 Aug '02

26 Aug '02

I have successfully installed 10.2 on my Mac and found that, after a recompile readline is just fine. The only problem I am having is with Tk which is broken due to trouble with the new dynamic linker library. The author of the last hacked dyld loader Stephen.O.Lidie(a)lehigh.edu is working on a new version of dyld as we speak (write). It should be done tomorrow (hopefully) he says. What I wanted to add here is to warn people *not* to use to old hacked version of dyld on the new 10.2 as it will almost certainly break the system (e.g. every command as well as the OS uses the dynamic library loader). I will let the list know when things are working. For the moment I have my pismo laptop running 10.1.5 until I get this sorted out on my dual 500 MHz G4 desktop. Mac OS X 10.2 does seem nice and fast! Paul

1 0

Warning: ifeffit users on Mac OS X
by Jeff Terry 24 Aug '02

24 Aug '02

Hello, If you are dependent upon ifeffit under OS X, DO NOT install the 10.2 upgrade. This upgrade breaks the readline library which is needed by ifeffit. A quick web search showed this to be a known problem with no current solution. Readline does not currently compile under 10.2. I tried a cursory effort compilation which resulted in many missing symbols. Jeff

1 0

Warning:ifeffit on OS X
by jterry94＠mac.com 24 Aug '02

24 Aug '02

Do not install OS X 10.2 if you rely on ifeffit under OS X. The readline libraries break and ifeffit cannot be used. A quick web search revealed that others have seen the problem and a quick attempt to recompile readline by myself has failed: missing symbols. Jeff

1 0

ifeffit 1.0073 released
by Matt Newville 20 Aug '02

20 Aug '02

Hi Everyone, I've posted Ifeffit 1.0073, both source code and Windows installer, on the ifeffit download page. Following the earlier discussion of interpolation schemes for continuous scan data, this source code has added 'interp=' options ('interp=line', 'interp=quad', 'interp=cubic') to the spline() and bkg_cl() commands. There are no other major changes to the base library, but a few minor improvements and bug fixes in the source. The Windows installer includes the latest (I believe!!) of athena, artemis, sixpack, and even gifeffit. Also, on Windows, the GrWin plot window now raises to the top at every plot(), cursor(), and zoom() command. Enjoy, --Matt

1 0

symmetry in Atoms
by Bruce Ravel 19 Aug '02

19 Aug '02

Here is an interesting question that was waiting in my mailbox this morning. I thought folks on the ifeffit list might also find it interesting and, in any case, it seems useful to have the question and the answer in the ifeffit list's archive. B The attached EXCEL file is the structure data for Fe17Y2 referred from Pearson's Handbook. Actually I get FEFF.inp file by Atoms using the input data as attached list. I wondered, howevere, how Atoms recognize site symmetry thorough the Atoms.inp data. My question is that the input data for each symmetric site should fulfill the site symmetry like as following list. In case of Fe17Y2 (No. 194), when I input (0 0 1/4) then Atoms will automatically recognizes this is 2b site. But when I input the number (0.1672 0.3344 0.9874) for 12k-site, how Atoms recognizes this is 12k site. Or if input numerical numbers do not fulfill the symmetry condition, then how Atoms deal with it. Thank you for your attention. space group number : 194 setting : 1 Hermann-Mauguin : P 6_3/m 2/m 2/c Laue group : 6/mmm point group : 6/m mm positions 24 l x,y,z -y,x-y,z -x+y,-x,z -x,-y,z+1/2 y,-x+y,z+1/2 x-y,x,z+1/2 y,x,-z x-y,-y,-z -x,-x+y,-z -y,-x,-z+1/2 -x+y,y,-z+1/2 x,x-y,-z+1/2 -x,-y,-z y,-x+y,-z x-y,x,-z x,y,-z+1/2 -y,x-y,-z+1/2 -x+y,-x,-z+1/2 -y,-x,z -x+y,y,z x,x-y,z y,x,z+1/2 x-y,-y,z+1/2 -x,-x+y,z+1/2 12 k x,2x,z -2x,-x,z x,-x,z -x,-2x,z+1/2 2x,x,z+1/2 -x,x,z+1/2 2x,x,-z -x,-2x,-z -x,x,-z -2x,-x,-z+1/2 x,2x,-z+1/2 x,-x,-z+1/2 12 j x,y,1/4 -y,x-y,1/4 -x+y,-x,1/4 -x,-y,3/4 y,-x+y,3/4 x-y,x,3/4 y,x,3/4 x-y,-y,3/4 -x,-x+y,3/4 -y,-x,1/4 -x+y,y,1/4 x,x-y,1/4 12 i x,0,0 0,x,0 -x,-x,0 -x,0,1/2 0,-x,1/2 x,x,1/2 -x,0,0 0,-x,0 x,x,0 x,0,1/2 0,x,1/2 -x,-x,1/2 6 h x,2x,1/4 -2x,-x,1/4 x,-x,1/4 -x,-2x,3/4 2x,x,3/4 -x,x,3/4 6 g 1/2,0,0 0,1/2,0 1/2,1/2,0 1/2,0,1/2 0,1/2,1/2 1/2,1/2,1/2 4 f 1/3,2/3,z 2/3,1/3,z+1/2 2/3,1/3,-z 1/3,2/3,-z+1/2 4 e 0,0,z 0,0,z+1/2 0,0,-z 0,0,-z+1/2 2 d 1/3,2/3,3/4 2/3,1/3,1/4 2 c 1/3,2/3,1/4 2/3,1/3,3/4 2 b 0,0,1/4 0,0,3/4 2 a 0,0,0 0,0,1/2 The attachment contained this: Fe17Y2 "structure type" "Pearson Smmbol" "space group" No Fe17Ho2 hP44 P6_3/mmc 194 Wyckoff x y z occupncy Y1 2b 0 0 0.25 0.41 Y2 2c 0.3333 0.6666 0.25 0.35 Y3 2d 0.3333 0.6666 0.75 1 Fe1 4e 0 0 0.1103 0.28 Fe2 4f 0.3333 0.6666 0.1049 0.71 Fe3 6g 0.5 0 0 1 Fe4 12j 0.3341 0.964 0.25 1 Fe5 12k 0.1672 0.3344 0.9871 1 And my response: Well, you made it very easy for me to answer your question by including the symmetry table for your space group in your email. The answer is that Atoms reads a data base of the space group symmetries. For your example, #194, the data base entry contains this information: x,y,z -y,x-y,z -x+y,-x,z -x,-y,z+1/2 y,-x+y,z+1/2 x-y,x,z+1/2 y,x,-z x-y,-y,-z -x,-x+y,-z -y,-x,-z+1/2 -x+y,y,-z+1/2 x,x-y,-z+1/2 -x,-y,-z y,-x+y,-z x-y,x,-z x,y,-z+1/2 -y,x-y,-z+1/2 -x+y,-x,-z+1/2 -y,-x,z -x+y,y,z x,x-y,z y,x,z+1/2 x-y,-y,z+1/2 -x,-x+y,z+1/2 which you will recognize as Wyckoff's position "l". Atoms doesn't exactly distinguish between Wyckoff's position l and position b. Instead it assumes that everything is position l and discards repeated positions. To understand this, apply the symmetry operations in the list above to the position (0,0,1/4). You will notice that 12 of them generate the position (0,0,1/4) and 12 of them generate (0,0,3/4). Atoms notices that and discards 11 of each. So, you see, position b is not actually different from position l, it's just degenerate due to symmetry. Atoms will notice these degeneracies a posteriori and does not need to know about the a priori. So, where did Wyckoff get his notation scheme from? Well, the Wyckoff positions are simply ordered in terms of decreasing degeneracy. Position a is the most degenerate and the highest lettered position is the non-degenerate one. This is not actually a useful notation scheme because there is no consistency between space groups. That is, position "c", for example, in group 194 has no relationship to position "c" in any other space group. The last thing worth mentioning is that Atoms works with 5 significant digits. If you use, as you suggest, (0.1672 0.3344 0.9874) then Atoms will find the correct amount of degeneracy. However, how you write (1/3, 2/3, z) is very important. Atoms 3.0 is able to read the string "1/3" as a coordinate. However if you choose to write coordinates in decimal notation, you must be careful. (0.33333, 0.66667, z) is fine and will generate 4 positions. (0.33, 0.67, z) will have problems related to numerical precision. Hope that's clear, Bruce -- 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

Atoms 3.0beta7
by Bruce Ravel 15 Aug '02

15 Aug '02

Hi folks, I released a new version of Atoms today which includes one small (but possibly important) change from the previous version. I learned that the database entry for a specific space group (I -4 3 d, #220) was in error. I fixed this and made a new release. There is a new tarball and Windows executable on my web site. If you are a windows user, you will also need to install the new version of the space_group.db file, as per the instructions on the downloads page. I also updated WebAtoms on Matt's server. This error has ABSOLUTELY NO IMPACT on any other space group. Indeed, without an atoms input file using the affected space group, you would never notice the problem. This underscores an interesting aspect of crystallography. While group theory tells us that there are 230 possible space group, nature seems to prefer using only a small subset of them. Indeed, in all the years since I started working on Atoms, I have not seen an example of a crystal in this space group until this week. The collection of input files that I distribute with Atoms for the purpose of testing the code includes examples of less than 80 space groups. Anyway, here is the input file that lead me to the error in Atoms' space group data base: title = Cu3As Domeykite title = Iglesias, J. E., and Nowacki, W., title = Zeitschrift fur kristallographie, kristallgeometrie, title = kristallphysik, kristallchemie, v. 145, pp. 334-345 (1977). space = I -4 3 d a = 9.6190 core = Cu rmax = 8.0 atoms ! elem x y z Cu -0.02930 0.12020 0.18760 As -0.02980 -0.02980 -0.02980 One might notice that the old Fortran version of Atoms handles this input file correctly. The irony of that is not lost on me, particularly in light of my frequent and loud trumpeting of the advantages of the new version over the old version. Thanks to Elspeth Barnes for bringing this problem to my attention. 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

athena 2002.08.14
by Bruce Ravel 14 Aug '02

14 Aug '02

Hi, I put a new tarball and Windows build for Athena on my website this afternoon. There are only a few changes since last week, but they are interesting enough to merit a new release. 1. I fixed a problem in the display of the group list. When deleting groups, slightly more vertical space was removed than had been allocated when the group was listed. This lead to a misalignment that was particularly apparent when deleting many groups. 2. I now do the thing where the cursor changes to a stop watch (an hourglass on Windows) whenever Athena is doing anything time consuming. That keeps you from clicking buttons as a bad time and also gives you a visual cue when lengthy procedures finish. 3. I put in more careful checks of memory usage. I had a bug report that was traced back to a large quantity of data overflowing Ifeffit's static memory space. I now check memory usage more frequently and issue a warning message when the amount of memory is running short. The scheme is not foolproof. It is still possible to overflow the memory before triggering a warning, but Athena should catch many situations in which this might happen. 4. It is now possible to create a "background group". That is a data group which is listed in the groups list and which contains as its data the background function computed from another data group. This is useful for at least three things, including directly comparing different background functions, comparing a background function to a background computed by FEFF and read from the 5th column of xmu.dat, and investigating low-R phenomena such as multi-electron effects or the so-called atomic xafs. 5. Project files now identify the version of perl, the version of ifeffit, and the operating system of the computer along with the version of Athena used to create the file. This is information I need when I am sent a project file as part of a bug report. There is no problem with backwards compatibility -- project files without this information will be read without problem. And project files continue to be transferable between platforms. The tarball and the Windows exe are in the standard places. Special thanks to Raul Barrea for a very useful bug report which lead to most of this week's new features. 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