March 2002 - Ifeffit - millenia.cars.aps.anl.gov

Re: [Ifeffit] Some thoughts on gifeffit
by Bruce Ravel 11 Mar '02

11 Mar '02

Paul said: > Also one of the things I used to do with the unix environment was to > set up a series of different autobak files with different rback > settings so I could see the effect of moving the fit window. It was > nice to be able to compare all of these in the same graphic window For what it's worth --- this can be done in Athena fairly easily. 1. Read in your mu(E) data. 2. Using the Group menu, make several copies of that data group. 3. Set Rbkg as desired in each copy. 4. Mark all the copies. 5. Click the purple k or R buttons. It would be nice to have a way of probing individual parameters for single groups without explicitly making copies. If anyone pipes up in support of that idea, I'll throw it on the todo list. B -- Bruce Ravel ----------------------------------- ravel(a)phys.washington.edu U.S. Naval Research Laboratory, Code 6134 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

iFEFFit and Tcl
by Matt Newville 11 Mar '02

11 Mar '02

Hi Tsu-Chien, > >> Has anyone tried to use Tcl/tk 8.4a4 with G.I.FEFFit? > > > > Yes. G.I.FEFFIT works on MacOS X. Installing the python > > package from Fink and then building and installing the > > Ifeffit-python wrapper will install G.I.FEFFIT. > > > What I mean is Tcl/Tk for MacOS X. Just found this interesting page: > <http://www.oreillynet.com/pub/a/mac/2001/10/26/tcl_osx.html?page=4> Ah, sorry I misunderstood. I have not tried to build the Tcl/Tk wrapper on MacOS X, but I'm pretty sure it would be possible and fairly easy. In fact, Tcl/Tk is available from Fink, and installing the python module must also install this, since python/Tkinter uses Tcl/Tk. Right now, the Ifeffit build on MacOS X does not try to build the Tcl extension, but it would be easy to add this. I believe it could be as easy as this: create TclSetup.in in ../wrappers/tcl: #---- LIB_PLT = -L/sw -lpgplot -L/usr/lib -lX11 -lpng LIB_F77 = -lg2c -lm LIB_IFF = -L../../src/macosx -lifeffit INC_IFF = -I../../src/lib CC = gcc CFLAGS = -g -O2 LDFLAGS = #----- and type 'sh Build'. I believe that 'tclsh test.tcl' will then work, running a simple test that doesn't do much besides prove that ifeffit runs from Tcl. I do not know much about how 'packages' are properly installed with Tcl/Tk, and haven't been very good about keeping the Tcl extension up-to-date. That's mostly because I don't actually use tcl myself. If someone else wanted to use tcl and ifeffit together, and/or take over this module, that would be great! --Matt

1 0

An Athena mini-tutorial
by Bruce Ravel 11 Mar '02

11 Mar '02

Hi folks, In the absence of a real document, I want to write something down to help people get started with Athena. Here goes.... First, Athena, like gifeffit, is an attempt to automate the creation of ifeffit commands. The beauty of ifeffit's command interface is that it is very flexible, both in terms of functional flexibility and in terms of writing wrappers. The annoying part of ifeffit's command interface is that you have to remember a lot of syntax and do a lot of typing. Essentially, Athena does a lot of typing for you. Second, Athena has two features that may not be obvious to folks who have used Mac-xafs or win-xafs or some other extant package. (1) There are no buttons that say "Remove background" or "Fourier transform". There are only buttons that say "Plot in energy" or "plot in R". Athena is clever about keeping track of what needs to be done and doing all necessary analysis steps in the background before plotting. So, if you change the k-weight for the Fourier transform, Athena knows to redo the forward and backward transforms as needed but not to redo the background removal. If you change Rbkg and plot in energy only, the background removal will be done. And so on. (2) You can "mark" groups and perform various actions on single groups or on ensembles of marked groups. So it is easy to plot single groups or ensembles of groups with a few mouse clicks. What's more the analysis steps are taken care of automagically for single groups or for ensembles of groups. Now I am going to describe specific features of Athena: ***** Reading in data Athena reads ASCII data using ifeffit, so all the rules that apply to ifeffit also apply to Athena. Basically, Athena expects that data will be in the form of column data, possibly with headers. Athena can directly read mu(E) or chi(k) data. It can also read raw data straight from the beamline. See http://leonardo.phys.washington.edu/~ravel/software/exafs/images/athena/col… You can then construct mu(E) data by selecting columns for the numerator and denominator. In the screen-shot, I show selecting It and I0 columns as the numerator and denominator for transmission data. There is a natural log button which should be clicked for transmission data. You can also handle multi-element detector data in a very rudimentary way by selecting multiple numerator columns for fluorescence data. You will notice that there are options in the File menu for reading single or multiple files. This is annoying but necessary. The standard file dialog does not do extended selection, the motif-like file dialog does extended selection but is unfamiliar to most people. Extended selection is very useful -- if you extendedly select a group of identically formatted files, Athena will only give the column selection dialog for the first one. The button labeled "Set preprocessing parameters" is beyond the scope of this post, but may become clear to you once you grok all the features of Athena. ***** Selecting and marking groups Once you have read in some data, you will probably want to plot it. You select a group by clicking on its name in the skinny frame on the right. When you select a group, its name will be highlighted in orange and its parameter values will displayed in the form on the left. When you click on a *red* plotting button, the selected group will be plotted. To the left of the group's name is a little check button. You mark a group by clicking the checkbutton on. When you click a *purple* plot button, all marked groups will be over-plotted. ***** Setting parameter values You can only directly edit the parameters for the selected group because only the selected group has its parameters displayed. There are ways of setting the parameters of other groups to the values of the selected group. In the Group menu is and option for setting the values of all parameters of all other groups to the values of the currently selected group. There is another option for setting all parameters of all marked groups to the values of the currently selected group. There is also fine grained control over setting parameters which is described in the next section. ***** Mouse clicks on labels When you pass the mouse over a label in the frame displaying all the parameter values, you will notice that the text and background change color. This is a visual cue that the text is sensitive to mouse clicks. Clicking the left mouse button on a word that has changed color will display a little bit of text in the echo area at the bottom of the window which explains in a few words the purpose of that parameter. Clicking the right mouse button will pop up a menu which allows you to set that parameter for other groups to the value of the currently selected group. The functions in the Group menu explained above work on all parameters, this right-click function works only on that specific parameter. ***** Plotting options You can independently control how single and marked group plots are made by choosing different options in the Plotting Options section at the lower right hand corner of the window. There is one notecard for each plotting space. Red buttons are for single group plots and purple buttons are for marked group plots. I suggest poking at these buttons to see what kinds of plots you can make. The plot ranges for the x-axis are set in this space as well. You may need to think about these values. For example, the data alignment function (described below) uses the values of Emin and Emax. Typically, you would want to restrict the plot to a few tens of volts above and below the edge so you can see the details of the edge for alignment. ***** Palettes The options in the Edit menu all pop up a new window with notecards which display all sorts of textual information. Perhaps the most useful is the ifeffit buffer which allows you to monitor the communication between Athena and ifeffit. ***** Project files You have the options of saving data in a variety of column formats, including mu(E), norm(E), chi(k), chi(R), and chi(q). You can also save a "project file". Which is an Athena specific which records the state of your Athena session. It will record data and parameters in a big ASCII file that is hard for a human but easy for Athena to read. Athena's project files are different from ifeffit's PAD files. They exploit the object oriented nature of its internal data structures to save the state of an ensemble of groups, even if the data in that ensemble is unrelated one to the other. ***** Deglitching The deglitch menu has two options in it, deglitching and truncating. Deglitching is for removing spurious points throughout the data set, truncating is for removing all data points after a specified point. Deglitching can be done point-by-point or algorithmically. The point-by-point is done by clicking the "Select a point" button, then clicking on the plot. The "remove point" button will then become active. Clicking it will remove the point from the group. The algorithm is a bit more complicated. There are three parameters which set margins above and below the data in a range either in the pre-edge or the post-edge. Any data points which extend outside these margins will be removed. To understand this better, try reading the data file uhup.003 and plying around with the algorithmic deglitching parameters. Eventually you will make it look like this: http://leonardo.phys.washington.edu/~ravel/software/exafs/images/athena/deg… Clicking the "Remove glitches" button will remove the four points outside the margins. ***** Aligning Alignment can be done using raw mu(E), normalized mu(E), or the derivative of mu(E). I think this one is pretty easy to understand. The "Standard" is the fixed reference. The "Other" is the group to be energy shifted. When you click the + or - buttons, the other group's e0 shift is incremented and the two data sets are replotted. See this: http://leonardo.phys.washington.edu/~ravel/software/exafs/images/athena/ali… ***** Merging Merging works on all marked groups. You can merge in any of the plotting spaces. The average and standard deviation of marked groups are calculated. The average, the average + the std. dev., and the average - the std. dev. will be plotted. The standard deviation will be written to output column data and to project files. There is more stuff to explain. I have not discussed macros, plot multipliers, clamps, phase corrections, Cromer-Liberman normalization, hints, the Plot and Mark menus, data preprocessing, detector data, pluck buttons, autobk standards, feff data, and a few other things. But hopefully this is enough to get people started until I write a proper document/tutorial. Regards, B -- Bruce Ravel ----------------------------------- ravel(a)phys.washington.edu U.S. Naval Research Laboratory, Code 6134 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

Some thoughts on gifeffit
by Dr. Paul Fons 11 Mar '02

11 Mar '02

I just brought up gifeffit this morning -- and I can already say I like it. I was wondering about some simple changes (perhaps I can do some of these myself once I get a little more sophisticated) 1. It would be nice to have legends on the pgplot window -- colors are nice but if one is comparing different data sets, it is really hard to tell them apart (especially if the page is printed). Also one of the things I used to do with the unix environment was to set up a series of different autobak files with different rback settings so I could see the effect of moving the fit window. It was nice to be able to compare all of these in the same graphic window -- it doesn't seem like this can be done anymore (am I right about this?) 2. In terms of documentation, a nice intro would be to highlight the differences between gifeffit and the more traditional unix command line version to help new users get up to speed -- again I am willing to write some of this (if help is needed), but I still need to get up to speed myself! 3. I am not 100% sure of this, but it seemed that when, for example, I had the plotter open and subsequently read in a new data set in another window or defined a new array, the pop-up menus in the plotter window weren't aware of the new data set (at least not immediately). One last thing, on the ifeffit homepage, it is stated that feffit is included in the distribution, but is not stable -- what is unstable and what are the limitations of the feffit module within gifeffit currently, e.g. is it something I can use to fit data? Last thing -- but really should have been first. Thank you for putting all of this effort into ifeffit and then sharing it with the rest of the community. Many people are benefiting from your efforts. Bruce's course (which is still a developing entity I believe) is also of great help in learning more about the different approaches to data analysis. It is my impression that while there is a fairly good collection of knowledge regarding exafs (and less so for xanes), this information is scattered in many different locations (and some of it is very dated). Having a reference in a easy-to-find place is a real benefit to the community at large. It also would have the advantage of being dynamic in that it doesn't have the birth (and death) cycles of commercially published material. I guess all of this is obvious, but I just wanted to say thanks (again). Paul

2 1

Athena runs
by Bruce Ravel 11 Mar '02

11 Mar '02

Chuck asked: CB> BTW, does Artemis have any requirements, other than its own perl CB> files, that are beyond ifeffit/atoms/athena? Nope. It works just like athena. There are a lot of widgets that are used to generate ifeffit text which gets piped to the ifeffit library. Then stuff gets plotted by pgplot. So if athena works, artemis will work. CB> Now to actually learn a bit about how to -use- ifeffit, tkatoms, CB> and athena! Matt's ifeffit tutorial (see the documents page on the ifeffit site) is quite nice and will give you an overview of all the main functions. The reference is also good -- I consult it all the time as I write my codes. It is, however, a bit behind the current version. Still, it's very useful. If you have ever used atoms, tkatoms will be quite straightforward. Athena is lacking documentation. I am going to post something this morning under a different subject line which will, I hope, get you started. B -- Bruce Ravel ----------------------------------- ravel(a)phys.washington.edu U.S. Naval Research Laboratory, Code 6134 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

iFEFFit/PGPlot/AquaTerm
by Tsu-Chien Weng 11 Mar '02

11 Mar '02

> 3. Has anyone run ifeffit using aquaterm? > Yes, I just tried newplot() with PGPlot/AQT, not in rootless XFree86. Since I'm a new user, I've not tried many functions in iFEFFit. Has anyone tried to use Tcl/tk 8.4a4 with G.I.FEFFit? Tsu-Chien <Chien>001000</Chien> === PGPlot/AquaTerm === 1. cvs AquaTerm 0.3 from aquaterm.sourceforge.net. 2. cp AQTProtocol.h, aqdriv.m to /usr/local/src/pgplot/drivers 3. mkdir sys_macosx, then cp g77_cc_AQT.conf and iand.c to it. 4. mkdir /usr/local/pgplot and copy over drivers.list. Add a line: AQDRIV 0 /AQT 5. /usr/local/src/pgplot/makemake /usr/local/pgplot 6. Modify makefile: DRIVERS=nudriv.o aqdriv.o aqdriv.o: $(DRVDIR)/aqdriv.m $(MCOMPL) -c $(MFLAGC) $(XINCL) $(DRVDIR)/aqdriv.m 7. make === iFEFFit === 1. download readline-4.2a from GNU and install libreadline.a to /usr/local/lib. 2. modify ifeffit_1.0b/src/macosx/Makefile a. #X_LIBS b. READLINE_LIBS = -L/usr/local/lib -lreadline c. NSLIBS = -framework Foundation -framework AppKit d. PGPLOT_LIB = -L$(BASEDIR)/pgplot -lpgplot -L/usr/lib $(NSLIBS) Note: I downloaded gcc3.1/g77 from <http://floyd.liunet.edu/~gkhanna/g77.dmg.gz>. The version of /usr/bin/cc is 2.95.2 and it has problem like /usr/bin/ld: /usr/lib/crt1.o relocation overflow for relocation entry 33 in section (__TEXT,__text) (displacement too large)

3 4

Mac OS X: athena, tkatoms, gifeffit all working!
by Matt Newville 09 Mar '02

09 Mar '02

OK, After installing perl/Tk, tkatoms (3.0beta5) installed without problems. Since I had already installed the ifeffit perl extension, and perl -e 'use Ifeffit;' was already working without complaint, the tkatoms interface to ifeffit ran fine. I also verified that a 'Powder' calculation ran and plotted to its canvas, and that the Chantler, Elam, and CL resources worked for absorption calcs. I couldn't get a plot in the DAFS window, but that could be operator error. After this, I reinstalled the ifeffit perl extension (though it may not have been necessary), and ... athena runs. It seemed to run perfectly, though I didn't try everything. There is a problem with athena saving hardcopies, but I think that will be easy to fix. In short, I believe everything is working on MacOS X. The necessary ingredients seem to be: From http://fink.sourceforge.net/ fink components: rootless X-windows, an X window manager (eg, orobus, sawfish, windowmaker), g77, python, and PGPLOT. From http://www.Lehigh.EDU/~sol0/Macintosh/X/ptk/ the perl/Tk fix for dyld follow instructions carefully!!! From http://cpan.perl.com/ perl's Tk800 build from source: > Perl Makefile.PL ; sudo make install From http://feff.phys.washington.edu/~ravel/software/exafs/ tkatoms build from source: > Perl Makefile.PL ; sudo make install From http://cars.uchicago.edu/ifeffit ifeffit.tar.gz, unpack and install: base lib: > tar xvzf ifeffit.tar.gz > cd ifeffit_1.0b/src/macosx > make > sudo make install perl extension + athena: > cd ../../wrappers/perl > perl Makefile.PL > sudo make install python extension + gifeffit: > cd ../../wrappers/python > sudo python setup.py install Edit .cshrc, adding if (! $?DISPLAY) setenv DISPLAY 0.0 setenv PGPLOT_DEV /xwindow # PGPLOT window dies with ifeffit setenv PGPLOT_DEV /xserve # PGPLOT window stays up. Then, with X-Darwin running, tkatoms, ifeffit, gifeffit, and athena should all run from a normal Terminal.app. I haven't tried these steps on a fresh MacOS X box, so there may be a missing step or two. Thanks Chuck, Francois, Paul, and Bruce for all the help. Now, can anyone else get all this to work? --Matt

2 3

Athena runs
by Charles Bouldin 08 Mar '02

08 Mar '02

>From: Bruce Ravel <ravel(a)phys.washington.edu> > > >Well, it is certainly a step in the right direction to actually >install the perl interface to Ifeffit before attempting to use Athena ;-) Well, okay(!) :), I never said I knew what I'm doing here, although I certainly know a lot more than 2 weeks ago. I did what Matt said, and also cleaned up what may have been a small version difference in the ifeffit distribution, and then, as Matt suggested: > cd ..... src/macosx; make clean ; sudo make install > cd ../../wrappers/perl ; perl Makefile.PL ; sudo make install and it all works! At least I get the tk panel for Athena...I don't know how to drive it yet. So, I agree with Matt...ifeffit, atoms and tk interfaces seem to be working. We should now put our heads together and try to refine the directions for all this....I'm pretty sure I could repeat it all from scratch, but it would take a while. Now to actually learn a bit about how to -use- ifeffit, tkatoms, and athena! BTW, does Artemis have any requirements, other than its own perl files, that are beyond ifeffit/atoms/athena? Chuck

1 0

Re: [Ifeffit] The road to Athena... (and Artemis too!)
by Bruce Ravel 07 Mar '02

07 Mar '02

The fact that the wrapper for python worked as it was occurred to me after I hit the send button. How does the loader work on BSD? Does it expect libraries to be in certain standard places? Perhaps Chuck ended up with the ifeffit library in some non-standard place and so needs to fix that, set an environment variable (like on SGI), or fix the loader configuration (like on linux). Anyway, I am glad to hear that such good progress is being made. I wish I could contribute more to the immediate problem. Do post the screen messages from the build, as Matt suggested. I mostly grok perl's strange and numerous messages (which must say something bad about me .... ;-) I wanted to mention that this week I made some good progress on Artemis (I had all of Monday at home watching a contractor poke holes in my walls!) There are still alot of clunky interface problems, but Artemis will run to completion a fit with one data set using any number of feff calculations and it will plot the results in a manner that rsembles Athena. That is, you can select the data, the fit, and/or any number of paths from the list in the skinny panel on the right and click a button for plotting in k, R, or q. Model building needs a lot of work, but its usable. I would love if the Mac effort could succeed soon -- that would somewhat increase the number of alpha testers ;-) B -- Bruce Ravel ----------------------------------- ravel(a)phys.washington.edu U.S. Naval Research Laboratory, Code 6134 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

The road to Athena...
by Charles Bouldin 07 Mar '02

07 Mar '02

Bruce sez to try this to test if things are in place for Athena: >#!/usr/bin/perl -w >use Tk; >use Ifeffit qw(get_scalar); >use Tk::Dialog; >my $x = get_scalar('etok'); >my $top=MainWindow->new(); >$top->Label(-text=>"Ifeffit sez etok=$x") ->pack(); >$top->Button(-text=>'Done', -command=>sub{exit})->pack(); >MainLoop; I tried it and got: >[h196132:~/Desktop] charlesb% perl test_ifeffit.pl >Can't locate Ifeffit.pm in @INC (@INC contains: /sw/lib/perl5/darwin >/sw/lib/perl5 /System/Library/Pe >rl/darwin /System/Library/Perl /Library/Perl/darwin /Library/Perl >/Library/Perl /Network/Library/Perl >/darwin /Network/Library/Perl /Network/Library/Perl .) at >test_ifeffit.pl line 3. >BEGIN failed--compilation aborted at test_ifeffit.pl line 3. But then I realized I needed to At 12:01 PM -0600 3/7/02, ifeffit-request(a)millenia.cars.aps.anl.gov wrote: > cd ../../wrappers/perl > perl Makefile.PL > sudo make install So, I did that, but now I get >[h196132:~/Desktop] charlesb% perl test_ifeffit.pl >dyld: perl Undefined symbols: >_ifeffit_ >_iff_exec >_iff_get_array >_iff_get_echo >_iff_get_scalar >_iff_get_string >_iff_put_array >_iff_put_scalar >_iff_put_string >_iffgetarr_ >_iffgetecho_ >_iffgetsca_ >_iffgetstr_ >_iffputarr_ and I'm not sure why this is happening. However, this feels close to working. More suggestions?

3 2