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

Re: ifeffit windowfunctions
by Matt Newville 11 Dec '02

11 Dec '02

Hi Daniel, On Wed, 11 Dec 2002, Daniel Grolimund wrote: > In case I am using non-regularly spaced k-chi data, the plot of the > chosen Fourier Transform Window is not represented correctly using the > command plot(group.k, group.win)]. > > Is there a way to get around that (except using interpolated and > therefore regularly spaced) data? Funny you mention this, I just noticed the same problem myself over the weekend. Currently, fftf() writes out the window function assuming the 0.05 Ang-1 grid, regardless of the input k-array. I believe this is fixed (that is, the group.win array corresponds to the group.k array!) in my latest test version. I still need to test other fixes in this version, but hopefully I'll get it out early next week. > Further, in a earlier version one had to use ikwindo = n to specify the > Fourier Transform Window type. In ifeffit the termology kwindow = aaaaa > is used. In addition to 'kaiser' and 'hanning' are there any other > options for kwindow ? Oh right. This is one place where Ifeffit and Feffit are different. Feffit's 'ikwindo = 2' caused some headaches (no one could remember them!) so 'kwindow=hanning' etc seemed better (if english-centric) for ifeffit. Other window functions are 'parzen' (linear sills), 'welch' (quadratic sills), and 'sine' (a sine wave over the whole range), and 'gaussian' (which is really weird: centered at dk2, with full width of 1/dk1, someone asked for it many years ago, it could probably be improved!). The next version of ifeffit will also have a window() command to make window functions without actually doing the FTs. --Matt

1 0

RE: [Ifeffit] Double dipping
by Kelly, Shelly D. 02 Dec '02

02 Dec '02

Hi Matt, I think that your reply sounds reasonable. Lets talk some more about this. > The other side of this is that I think it's more difficult than > generally acknowleged to over-interpret data. Most of the cases > I've seen are due to blatantly ignore the confidence limits or do > completely unfair things like fitting So2 to get 0.9+/-0.1, then > fix So2 to 0.90, fitting N, and claiming N to better than 10%. > Those are serious, but are the normal mistakes in analysis that > can happen with any data. If you fix s02 and then you fit N and get a error lets call it dn. Then the more generous estimation of the uncertainty for N is d(s02*n) = ds02*n + s02*dn. How would you estimate the error for N if you fix s02, R and sigma2? I will try to be more specific to my original problem. lets say that the original data set (s3) gives these results for the Ca shell. Ncas3=3.4+/-0.9 Rcas3=4.01+/-0.01 sigma2cas3=0.006+/-0.3 s02s3=1.05+/-0.10 Now I run another fit and I fix all the parameters except for Nca. This time I fit three data sets s3, s4 and s5 together and I get these results. Ncas3 = 3.2+/-0.1 Ncas4 = 2.5+/-0.4 Ncas5 = 2.0+/-0.4 What would you think would be an "appropriate" uncertainty? Surely 0.1 is too small for Ncas3 but 0.9 seems rather huge. So we have some limits but can I do any better than that? The difference is between these to extremes is quite profound. In one case we see a change in the value for Nca in the other we see nothing. Shelly > -----Original Message----- > From: Newville, Matthew G. > Sent: Friday, November 22, 2002 10:17 AM > To: 'ifeffit(a)millenia.cars.aps.anl.gov' > Subject: Re: [Ifeffit] Double dipping > > > Hi Shelly, > > > I would like to take a bunch of parameters (all of them that do > > not vary significantly from the known, as determined by > > individual fits to each data set) from this "known" data set > > and then fit the series including the "original" to see how a > > particluar parameter (the number of Ca atoms) changes as the > > ratio of U to Ca changes. I would like to hear a discussion on > > the fairness of this approach and how would you calculate the > > number of independent points? > > I think what you propose is fair, but you might need to be a > little more specific.... > > In general, I think it's fair to use parameters from one fit to > to another data set. Once one accepts the fact that there is > limited amount of information, getting hung up on trying to count > or enforce N_idp becomes pointless -- the data and estimated > uncertainties __WILL__ indicate how many parameters can be > determined with any confidence. > > N_idp is one statistic, it should not be construed as an exact > value nor a hard upper limit on how much data can be extracted > from data. Any attempt to use N_idp this way is abuse. > > The classic test is whether 100 very noisy scans (say, because > data was collected for 0.01s per point), has 100x more > 'independent points' as 1 scan that collected for 1.0s per point. > The 100 noisy scans __ARE__ independent, but when added together > with give exactly the same result as using the 1 clean scan. So > independent data does not mean you can determine you can > determine more parameters. > > The other side of this is that I think it's more difficult than > generally acknowleged to over-interpret data. Most of the cases > I've seen are due to blatantly ignore the confidence limits or do > completely unfair things like fitting So2 to get 0.9+/-0.1, then > fix So2 to 0.90, fitting N, and claiming N to better than 10%. > Those are serious, but are the normal mistakes in analysis that > can happen with any data. > > --Matt > > > _______________________________________________ > Ifeffit mailing list > Ifeffit(a)millenia.cars.aps.anl.gov > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit >

2 1

notes for figures
by Bruce Ravel 22 Nov '02

22 Nov '02

Ning, I trust you will not mind if I respond to the ifeffit mailing list. You question is of general interest. > On Fri, 22 Nov 2002, "Ning Chen" <Ning.Chen(a)lightsource.ca> said: NC> Hi Bruce: I am wondering if it is possible to add note to the NC> figure inside the "Graphic window"? Hmmm.... no one has ever asked for that before. Good question! There is a way. It's not particularly convenient, but it's not particularly difficult either. Here is an example: http://leonardo.phys.washington.edu/~ravel/software/exafs/images/athena/tex… I made this first by clicking on the red "kq" button. Then, in the edit menu, I selected "Ifeffit buffer". At the bottom of the little window that popped up, there is a box with a label that says "Ifeffit>". That text box allows the user to interact directly with ifeffit by typing in explicit commands. Here is a picture: http://leonardo.phys.washington.edu/~ravel/software/exafs/images/athena/ife… To get the text string in the plot, I typed: plot_text(x=7, y=-0.3, text="Hi Ning!") which tells Ifeffit to put the indicated text string at the indicated coordinates. The same thing could be done the same way in Artemis. Obviously, this could be made more user friendly by, for example, having Athena or Artemis prompt you to click on a spot in the plot window and then prompt you for a text string. I'll put that on the TODO list. Please note that using the plot_text command as I describe above will trigger the minor bug described by Matt earlier this week on this list: http://millenia.cars.aps.anl.gov/pipermail/ifeffit/2002-November/000177.html 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/

2 2

Double dipping
by Kelly, Shelly D. 22 Nov '02

22 Nov '02

Hello All, I am new to this mailing list, so I hope my question is relevant. I have several EXAFS data sets that I want to simultaneously fit with one model. One of the data sets was modeled first and the results will be published separately from the rest. I would like to take a bunch of parameters (all of them that do not vary significantly from the known, as determined by individual fits to each data set) from this "known" data set and then fit the series including the "original" to see how a particluar parameter (the number of Ca atoms) changes as the ratio of U to Ca changes. I would like to hear a discussion on the fairness of this approach and how would you calculate the number of independent points? Shelly Kelly

2 1

Athena 0.8.007 and Artemis 0.5.003
by Bruce Ravel 22 Nov '02

22 Nov '02

Hi Folks, I just posted new versions of my two programs on my web site. The source tarball for Unix and OSX users as well as newly built windows executables are in their standard places. Most of the changes this time fix bugs that I discovered last week when I was visiting the wonderful folks at the Alberta Synchrotron Institute. The two most important changes fix the problems displaying the normalized background function and reading chi(k) data that is not on a uniform k-grid. Windows user who are new to Ifeffit: You just need to download the new executable files and drop them on top of the old ones in C:\Program Files\Ifeffit (or wherever you installed Ifeffit on your computer). The big installer file has *not* yet been updated to include these new versions. Unix, linux, or Mac OSX users: The source tarball now includes the ifeffit wrapper that comes with the main ifeffit distribution. Now, when you update ifeffit, it is no longer necessary to cd into the "wrappers/perl/" directory and build the perl wrapper. Instead the wrapper is built at the same time as Athena and Artemis. I am fairly confident that the wrapper gets compiled correctly. It works fine on my linux computer and on Mac OSX. Please let me know if you have any troubles with the compilation so I can fix the build scripts. As always, feel free to post questions, bug reports, or feature suggestions to this mailing list. Special thanks this time to Paul Fons and to all the folks at the ASI. Regards, 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

Re: pgplot question
by Matt Newville 20 Nov '02

20 Nov '02

Hi Paul, On Thu, 21 Nov 2002, Paul Fons wrote: > I noticed (from early on) that although pgplot generated graphics show > all spectra labels on screen, when I save them to a file, that the > lowermost scan of a multiple scan plot always has the label missing. > Is this something that is Mac specific or is everyone suffering from > this problem? I want to try tracking it down a little (perhaps > including sending a note to the pgplot author if helpful). This is my fault -- a bug that's been on the to-do list forever, and never done. I believe this will also cause the same behavior: my.x = range(1,100,1)/10 my.y = sin(my.x) my.z = cos(my.x) newplot my.x, my.y, key='sin' plot my.x, my.z, key='cos' #everything is fine... plot xmax = 8 and the most recent 'key' gets lost. Making a hardcopy simply does a plot() with a different device, and so often tickles this error. Workarounds include adding the last trace twice, as in newplot my.x, my.y, key='sin' plot my.x, my.z, key='cos' #everything is fine... plot my.x, my.z, key='cos', color='red' which will then trick a second plot(). Anyway, it's all my fault, not PGPLOT's. Probably should fix it huh?? --Matt

1 0

Ifeffit 1.0075
by Matt Newville 13 Nov '02

13 Nov '02

Hi Everyone, Ifeffit 1.0075 is now available. There are some bug fixes related to erasing paths, which caused serious problems with artemis. There was also a change in the guts of how the Perl wrapper was built, as the older one failed to compile with gcc3.2 on RedHat 8.0. I have not tried this on Mac OS X, and don't foresee any problems, but if someone is brave enough to try it, please let me know if it fails. The most significant change in 1.0075 is the inclusion of Feff6 in the source tree and a Windows executable of Feff6 in the Windows installer. That right, there is now a free version of Feff6 included with Ifeffit!! The plan is to turn this version of Feff into a library that can be called directly from the perl and python scripts. That is, the 'run feff' button on TkAtoms might actually work someday!! If you're as thrilled about this as I am, don't hesitate to thank John Rehr repeatedly. There are new versions of sixpack, artemis, and athena in the Windows installer. The latest source code for these should be downloaded from Sam's and Bruce's web pages. Special Thanks to John, Sam, and of course Bruce, --Matt

4 6

ifeffit and feff
by Bruce Ravel 06 Nov '02

06 Nov '02

Ahmed, I trust you do not mind that I answer your question on the ifeffit mailing list. It seems of sufficiently broad interest that an answer in that forum is appropriate. You asked: AS> I received a note with the latest version of your software. In my AS> understanding that the latest version includes the Feff6L without AS> a license fee. Does this mean that by downloading the latest AS> version of ifeffit, so that I do not need to buy a license of AS> feff? Well, my answer is ether a qualified yes or a qualified no, depending on what your interests are ;-) *IF* you are interested only in the extended EXAFS features of Feff and *IF* you do not think that you need any of the more recent developments of Feff (most notably, self consistent potentials), then you probably do not need a license for Feff. Feff6L is free in both senses of the word -- it doesn't cost you any money and you may freely use, modify, and redistribute the code under the terms of its license. See this post from John Rehr to the mailing list about the Feff6L license: http://millenia.cars.aps.anl.gov/pipermail/ifeffit/2002-November/000169.html However, if you are interested in computing XANES, partial electronic densities of state, charge transfer or ab initio Fermi energies, you will need to license a copy of the latest version of Feff. If you are interested in the latest developments of such theoretical topics as self-consistent potentials, improved self-energies, S02 calculations, resonant scattering corrections, or a whole host of other features available in the latest Feff and not available in Feff6L, you will need to license a copy of the latest version of Feff. Feff6L is useful code for the extended exafs practitioner. If that were not true than Matt would not have been so effusive in his post yesterday announcing Ifeffit 1.0075. While Feff6L is useful and usable for exafs analysis, you must be aware that it is largely limitted to exafs analysis. If you need XANES calculations, you need a Feff8 license. All clear? 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

FEFF6L - Lite EXAFS only version of FEFF6 source code
by John J. Rehr 05 Nov '02

05 Nov '02

Dear Feff developers, In an effort to facilitate the development of EXAFS analysis software, the FEFF Project has made available the source code for a "lite" subset of the FEFF6 distribution sufficient for EXAFS analysis, named FEFF6L. Thus cross-sections and XANES capabilities of FEFF6 are disabled in FEFF6L. A copy of the FEFF6L distribution can be found (shortly) on the FEFF Developer download pages. The subroutines are distributed with a License that permits redistribution without a license fee; see the LICENSE in the distribution for details. c/////////////////////////////////////////////////////////////////////// c Distribution: FEFF6L c Copyright (c) [2002] University of Washington c c This software was prepared in part with US Government Funding under c DOE contract DE-FG03-97ER45623. c c Redistribution and use of this Distribution in source and binary c formats, with or without modification is permitted, provided the c following conditions are met: c c Redistributions must retain the above notices and the following list c of conditions and disclaimer; c c Modified versions carry the marking c "Based on or developed using Distribution: FEFF6L c Copyright (c) [2002] University of Washington" etc. The distribution FEFF6L should not be confused with the full distribution of FEFF6. Also the FEFF Project will continue to develop advanced versions of the FEFF codes which will be distributed with the UWOTT License and license fee. Comments on this EXAFS version are welcome. Sincerely, FEFF Project PIs J. J. Rehr A. L. Ankudinov

1 0

Re: [Ifeffit] Debye factor and S02 correlation
by Bruce Ravel 05 Nov '02

05 Nov '02

jjr> I don't understand Bruce's remark. Incorporating prior knowledge jjr> into statistics is completely proper according to modern jjr> statistical theory. If two quantities are correlated and one jjr> knows one, then naturally the other is better determined. Well, leaving aside the question of whether one can call a Levenberg- Marquardt fit "modern", I agree that one can set a value and call it part of the model. One does that all the time in any fitting problem. But in that case you are asserting a value with no error bar. There is no situation where you can assert a value AND an error bar. My concern was your statement that you somehow knew the value of the error bar on S02. If it floats, you have to use the error bar the fitting algorithm reports. But the most important thing in my comment and the most important thing in the discussion on the topic of amplitude terms in exafs that we have seen in the last week on this list is that, in a difficult fitting problem (i.e. not copper), one rarely has reliable a priori knowledge of the *product* of S02, coordination, and the systematic uncertainties. Thus one rarely has a reliable prior in the Baysian sense. 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