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

R-factor and reduced chi square
by OWEN N LI 05 Oct '05

05 Oct '05

Hi, I understand R-factor and reduced chi square are statistical ways to see if the model is reasonable (R-factor) and if one model is better than the others (reduced chi square). But how much should I read into it? I have model A and B, I believe A is more theoretical sound than B. but when I fit them using IFEFFIT, the R-factor and reduced chi square of B is slightly better than A (B: 0.03, 28; A: 0.04, 33). What does that tell me? Is my assumption incorrect? Or are they more or less the same? Owen

5 4

RE: [Ifeffit] New Method for Normalization?
by Barton, David (DG) 04 Oct '05

04 Oct '05

Bruce, You mentioned: > Isn't Ifeffit's Cromer-Liberman normalization scheme > substantially similar to the method presented in that paper? > When I use the Cromer-Liberman background subtraction routine in Athena the Normalized data is not flattened. Normalized, but not *flattened* data results in a skewed XANES region which is affected by the amount of curvature in the background. If I choose "Flatten" in Athena, it is my understanding that it applies a linear pre-edge and 2nd order polynomial post-edge regardless of which background was chosen (Autobk or CLNorm). The Clnorm background does appear to "flatten" the chi(k) data which can be useful for EXAFS analysis, but it doesn't flatten the mu(E) data. When one attempts to analyze XANES data it is preferred to have the entire post-edge "flattened" in mu(E) space so that the background curvature is eliminated from the dataset. Ultimately, as Matt stated, it would be nice to be able to combine a Victoreen and CLNorm background to get a reasonably "flattened" mu(E) function. I would also suggest adding the ability to independently set the pre-edge and post-edge functions (n'th order polynomial, Victoreen, CLNorm, or MBACK??) for Normalization and Flattening. These choices would be separate from the choice of background removal method to get chi(k) (Autobk or CLNorm). Dave > -----Original Message----- > From: ifeffit-bounces(a)millenia.cars.aps.anl.gov > [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov] On Behalf > Of Bruce Ravel > Sent: Thursday, September 29, 2005 5:20 PM > To: XAFS Analysis using Ifeffit > Subject: Re: [Ifeffit] New Method for Normalization? > > > On Thursday 29 September 2005 15:48, Barton, David (DG) wrote: > > Has anyone considered using the methods of Penner-Hahn to normalize > > their data to get reliable intensities of near-edge features? The > > reference to their method of normalization is below: > > > > J. Synchrotron Rad. (2005). 12, 506-510 > [doi:10.1107/S0909049504034193] > > "A method for normalization of X-ray absorption spectra" > T.-C. Weng, > > G. S. Waldo and J. E. Penner-Hahn > > > > It is my understanding the Ifeffit uses a linear function to remove > > the pre-edge and a quadratic for the post-edge and that the > Athena's > > flatten is a subtraction of these functions. This is an excellent > > method under most scenarios since the background is usually fairly > > smooth and can be well approximated by a quadratic equation. > > Personally, I have had only a few rare cases where the quadratic > > function on the post-edge was not sufficient to reliably > normalize the > > data and in those cases a third-order polynomial was > sufficient. Does > > anyone else have an opinion on using alternative normalization > > routines? > > Isn't Ifeffit's Cromer-Liberman normalization scheme > substantially similar to the method presented in that paper? > > B > > > -- > Bruce Ravel ---------------------------------------------- > bravel(a)anl.gov > > Molecular Environmental Science Group, Building 203, Room > E-165 MRCAT, Sector 10, Advance Photon Source, Building 433, > Room B007 > > Argonne National Laboratory phone and voice mail: > (1) 630 252 5033 > Argonne IL 60439, USA fax: > (1) 630 252 9793 > > My homepage: http://cars9.uchicago.edu/~ravel > EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/ > > > _______________________________________________ > Ifeffit mailing list > Ifeffit(a)millenia.cars.aps.anl.gov > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit >

1 0

RE: [Ifeffit] R-factor and reduced chi square
by Kelly, Shelly D. 04 Oct '05

04 Oct '05

Hi Owen, A test for a significant change in the reduced chi square (RCS). [(RCS1)^2/(RCS2)^2] - 1 greater than 2 * sqrt ( 2 / degrees of freedom) If you can not read that, I have this published in J. Phys Chem B, V105 no27 year (2001). You can down load that paper from my web page www.mesg.anl.gov. I guess it is a reasonable test:). For most model fits a change by a factor of 2 is needed to be statistically significant. Good luck Shelly Shelly Kelly Bldg 203 RM E113 Skelly(a)anl.gov Argonne National Laboratory 630-252-7376 9700 S Cass Ave www.mesg.anl.gov Argonne, IL 60440 > -----Original Message----- > From: ifeffit-bounces(a)millenia.cars.aps.anl.gov [mailto:ifeffit- > bounces(a)millenia.cars.aps.anl.gov] On Behalf Of OWEN N LI > Sent: Tuesday, October 04, 2005 1:08 PM > To: ifeffit(a)millenia.cars.aps.anl.gov > Subject: [Ifeffit] R-factor and reduced chi square > > Hi, > I understand R-factor and reduced chi square are statistical ways to > see if the model is reasonable (R-factor) and if one model is better > than the others (reduced chi square). But how much should I read into it? > > I have model A and B, I believe A is more theoretical sound than B. > but when I fit them using IFEFFIT, the R-factor and reduced chi square > of B is slightly better than A (B: 0.03, 28; A: 0.04, 33). What does > that tell me? Is my assumption incorrect? Or are they more or less the > same? > > Owen > > > _______________________________________________ > Ifeffit mailing list > Ifeffit(a)millenia.cars.aps.anl.gov > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

1 0

Muffin tin radius adjustment?
by Paul Fons 04 Oct '05

04 Oct '05

I am trying to analyze some XAFS data on a Sb-Te compound and when I run feff(6) I get errors regarding the coordination number being too small implying the muffin tin radius is too big. In this material the first nearest neighbor bond length is about 3 angstroms (slightly more or less depending on the site). I assume this is related to the cause of the problem. Any suggestions as to how to fix/avoid this sort of problem? feff6 output is pasted below. Interestingly enough, feff8 does not output any error messages and the sum of the paths looks identical (at least at first glance). Is this a false error message or (more likely) has feff8 become more intelligent with regards to defining the muffin tin radius? Feff6 output Calculating potentials and phases... free atom potential and density for atom type 0 free atom potential and density for atom type 1 free atom potential and density for atom type 2 overlapped potential and density for unique potential 0 overlapped potential and density for unique potential 1 overlapped potential and density for unique potential 2 muffin tin radii and interstitial parameters phase shifts for unique potential 0 Hard tests failed in fovrg. Muffin-tin radius may be too large; coordination number too small. phase shifts for unique potential 1 Hard tests failed in fovrg. Muffin-tin radius may be too large; coordination number too small. phase shifts for unique potential 2 Hard tests failed in fovrg. Muffin-tin radius may be too large; coordination number too small. Preparing plane wave scattering amplitudes... nncrit in prcrit 9 Searching for paths... WARNING in PATHS Module: rmax > distance to most distant atom. Some paths may be missing. rmax, ratx 15.50000 15.40291 Rmax 15.5000 keep and heap limits 0.0000000 0.0000000 Preparing neighbor table Feff8 output Calculating potentials ... free atom potential and density for atom type 0 free atom potential and density for atom type 1 free atom potential and density for atom type 2 initial state energy overlapped potential and density for unique potential 0 overlapped potential and density for unique potential 1 overlapped potential and density for unique potential 2 muffin tin radii and interstitial parameters iph, rnrm(iph)*bohr, rmt(iph)*bohr, folp(iph) 0 1.79879E+00 1.71274E+00 1.13279E+00 1 1.81342E+00 1.72495E+00 1.13595E+00 2 1.83703E+00 1.74437E+00 1.14149E+00 mu_old= -0.508 Done with module 1: potentials. Calculating LDOS ... LDOS calculation for specified grid Calculating energy and space dependent l-DOS. It takes time ... potential type 0 potential type 1 potential type 2 Calculating chi and rho... 0 Calculating chi and rho... 1 Calculating chi and rho... 2 Done with LDOS. Calculating cross-section and phases... absorption cross section phase shifts for unique potential 0 phase shifts for unique potential 1 phase shifts for unique potential 2 Done with module 2: cross-section and phases... Done with module 3: FMS. Preparing plane wave scattering amplitudes... Searching for paths... WARNING: rmax > distance to most distant atom. Some paths may be missing. rmax, ratx 1.55000E+01 0.00000E+00 Rmax 15.5000 keep and heap limits 0.0000000 0.0000000 Preparing neighbor table Dr. Paul Fons Senior Researcher National Institute for Advanced Industrial Science & Technology METI Center for Applied Near-Field Optics Research (CANFOR) AIST Central 4, Higashi 1-1-1 Tsukuba, Ibaraki JAPAN 305-8568 tel. +81-298-61-5636 fax. +81-298-61-2939 email: paul-fons(a)aist.go.jp The lines below are in a Japanese font 〒３０５−８５６８茨城県つくば市東１−１−1 つくば中央第４近接場光応用工学センターポール・フォンス主任研究官

3 2

Runner.exe Network Error
by Barton, David (DG) 28 Sep '05

28 Sep '05

Bruce and Matt, I am running the Ifeffit 1.2.8 on a Win2000 box and I having several issues with the startup of Athena/Artemis when I am not connected to a network. First thing is that all programs run fine when I am connected to the internet. So the obvious workaround for me is to just make sure that I have an open network connection before running Athena/Artemis. However, it would be nice if the auto-update feature in "Runner.exe" would still let me run Athena/Artemis when I wasn't connected to the network, basically just ignore looking for an update when I don't have a network connection. The errors that I get in the Runner.exe.log are below. There are two different cases shown. The first set of errors are when I have a connection to my local network, but have not yet logged onto the firewall that lets me out onto the internet. The second set of errors comes up when I unplug the network connection completely. Hope these help. NETWORK CONNECTION NO INTERNET CONNECTION ----------------------------------------- Traceback (most recent call last): File "runner.py", line 753, in ? File "runner.py", line 184, in __init__ File "runner.py", line 275, in LookForUpdates File "runner.py", line 299, in read_web_version File "urllib.pyc", line 76, in urlopen File "urllib.pyc", line 181, in open File "urllib.pyc", line 306, in open_http File "urllib.pyc", line 319, in http_error File "urllib.pyc", line 565, in http_error_302 File "urllib.pyc", line 580, in redirect_internal File "urllib.pyc", line 181, in open File "urllib.pyc", line 306, in open_http File "urllib.pyc", line 319, in http_error File "urllib.pyc", line 616, in http_error_401 File "urllib.pyc", line 624, in retry_http_basic_auth File "urllib.pyc", line 650, in get_user_passwd File "urllib.pyc", line 658, in prompt_user_passwd NO NETWORK CONNECTION -------------------------------------- File "runner.py", line 753, in ? File "runner.py", line 184, in __init__ File "runner.py", line 275, in LookForUpdates File "runner.py", line 302, in read_web_version Dave _________________________________ David Barton The Dow Chemical Company

2 1

getting chi(r) information out of command line IFEFFIT
by Sean X. Luo 28 Sep '05

28 Sep '05

Hello all. It's been a while since I've written on the mailing list about this project, but it's come up that I need everyone's general input. Just a brief overview of the project: I am working with Scott Calvin to simultaneously fit 12 data sets in the command line of IFEFFIT, and I need to somehow get the chi(r) information for the fit out of the command line. I've tried around with the "show" command and the different permutations possible with that, but whenever I type in "show(fit.r, fit.chir_real)" (which is our guess right now on how to get the information I need out of the command line), it gives me the error that the arrays fit.r and fit.chir_real does not exist, or something along those lines. Does anyone else have any idea on how to run this? Thanks in advance, - Sean.

3 3

RE: [Ifeffit] Fe oxide
by Kelly, Shelly D. 26 Sep '05

26 Sep '05

Hi Juro, I had some fun playing with this project file. I hope you don't mind. Here is a list of suggestions. 1st: Don't use the paths that are less than 0.1 Angstroms apart. 1st-b: If you don't know the CN set it to ONE and then vary just the amplitude. There is a degen box multiplied by a amp box. Degen should be one. 2st: set the values for CN, and s2, while letting delr vary 3rd. Let the CN and delr vary. You will find that the Fe path around 3 Ansgroms isn't needed. 4th. Make sure your fit range and data ranges are appropriate. 5th. Add as much constraints as you think is reasonable. First Oxygen coordination number should be 6 or 4 total. I thought 6 worked well. 6th. Fit the data with k-weight of 1,2, and 3, to get the uncertainties smaller. Take a look at the project file. I think that this model or one like it might work. You need a separate S02 value from the Amp value. Amp should be S02*Amp. I didn't add a S02 part. You should get that from a known standard measured on the same beamline. S02 should be between 0.70 and 1.0. So the CN reported are likely a bit too big. Good luck, Shelly Shelly Kelly Bldg 203 RM E113 Skelly(a)anl.gov Argonne National Laboratory 630-252-7376 9700 S Cass Ave www.mesg.anl.gov Argonne, IL 60440 > -----Original Message----- > From: ifeffit-bounces(a)millenia.cars.aps.anl.gov [mailto:ifeffit- > bounces(a)millenia.cars.aps.anl.gov] On Behalf Of Juraj Majzlan > Sent: Monday, September 26, 2005 8:46 AM > To: ifeffit(a)millenia.cars.aps.anl.gov > Subject: [Ifeffit] Fe oxide > > Hello, > > I am trying to fit an EXAFS spectrum of an X-ray amorphous > Fe precipitate. In a model (goethite), there are Fe atoms > at 3.02, 3.29, and 3.45 A. Fitting the two most distant > shells of iron gives strange results. Fitting them > simultaneously gives very high amplitudes (32, -66) which > cancel each other (fig. 1) and produce a beautiful and > unreasonable fit (fig. 2). > > Fitting them separate gives small ('reasonable') > amplitudes, but the distance (dr) have very high > uncertainties, e.g., dr = 0.12 +- 20.3. > > Can I expect to resolve these three Fe shells or is it > asking too much? > > I am also attaching the Artemis file, but will appreciate > any comments and suggestions. > > Juro Majzlan > > > ---------------------------------- > Juraj Majzlan > Institute of Mineralogy and Petrology > Albert-Ludwig-University of Freiburg > Albertstrasse 23b > Freiburg, D-79104, Germany > telephone +49-761-203-6416 > fax +49-761-203-6407 > http://www.minpet.uni-freiburg.de/sites/personen/majzlan.html

1 0

Fe oxide
by Juraj Majzlan 26 Sep '05

26 Sep '05

Hello, I am trying to fit an EXAFS spectrum of an X-ray amorphous Fe precipitate. In a model (goethite), there are Fe atoms at 3.02, 3.29, and 3.45 A. Fitting the two most distant shells of iron gives strange results. Fitting them simultaneously gives very high amplitudes (32, -66) which cancel each other (fig. 1) and produce a beautiful and unreasonable fit (fig. 2). Fitting them separate gives small ('reasonable') amplitudes, but the distance (dr) have very high uncertainties, e.g., dr = 0.12 +- 20.3. Can I expect to resolve these three Fe shells or is it asking too much? I am also attaching the Artemis file, but will appreciate any comments and suggestions. Juro Majzlan ---------------------------------- Juraj Majzlan Institute of Mineralogy and Petrology Albert-Ludwig-University of Freiburg Albertstrasse 23b Freiburg, D-79104, Germany telephone +49-761-203-6416 fax +49-761-203-6407 http://www.minpet.uni-freiburg.de/sites/personen/majzlan.html

2 1

problems on running the windows 1.2.8 ifeffit version
by M Ceolin 23 Sep '05

23 Sep '05

Dear Bruce et al. I just installed the new 1.2.8 for windows version of ifeffit and I am experiencing problems to run any of the programs associated (arthemis, athena, etc.). I installed the programs in a Win98 machine where the previous versions of IFEFFIT were running. I took care of adding the suggested line to the config.sys file and installed the program in c:\ifeffit (as suggested). However, both, starting from the icons in the desktop, from the applications installed at c:\ifeffit\bin or from the menu at the start, the runner program (installed with the pakage) send an error message forcing to abort the run. The message says: (sorry, in spanish) Este programa ha efectuado una operacion no admitida y sera interrumpido (in the datails window says) RUNNER provoco un error de pagina no valida en el modulo MSVCRT.DLL de 017f:7801274b. etc. (registry numbers). Could you comment the point and, perhaps, suggest a solution ?. Yours. Marcelo Ceolin. *************************************** Dr. Marcelo Ceolin Centro Regional de Estudios Genomicos UNLP-AUGM Av. Calchaqui km 23,5 (4to piso) (1888) Florencio Varela Argentina Tel/Fax: +54-11-42758100 e-mail 1: mceolin(a)creg.org.ar e:mail 2: mrceolin(a)yahoo.com.ar WEB: www.creg.org.ar *************************************** PLEASE NOTE THAT THE E-MAIL ADDESSS mceolin(a)gwdg.de IS NO LONGER AVAILABLE. USE ONE OF THE ADDRESSES ABOVE. --------------------------------- 1GB gratis, Antivirus y Antispam Correo Yahoo!, el mejor correo web del mundo Abrí tu cuenta aquí

2 1

RE: [Ifeffit] writing beam time proposals
by Frenkel, Anatoly 23 Sep '05

23 Sep '05

I rate proposals on the merit but it a former XAFS school student is a research group member it makes me happier. Anatoly -----Original Message----- From: ifeffit-bounces(a)millenia.cars.aps.anl.gov on behalf of Matt Newville Sent: Fri 9/23/2005 12:06 PM To: Carlo Segre; XAFS Analysis using Ifeffit Cc: Subject: Re: [Ifeffit] writing beam time proposals In my opinion, it is better for the student or post-doc to write the beamtime proposal and be listed as PI, with the advisor listed somewhere on the proposal Many proposal foms have a checkbox for 'is this work part of a thesis', and that is definitely monitored (at least at the APS, but I'd imagine it to be true at other facilities), even if it is not formalized. There is usually an eagerness for bringing new students (and new post-docs and new faculty members) to the syncrotrons. Also, in my experience reading beamtime proposals now and then, students and post-docs do a better job at writing them <wink>. Even f the professor is listed first (and many groups do this!), I would suggest that the student/post-doc actually write the proposal and have the advisor look it over before submitting. Other advice for writing proposals: - communicate with the beamline scientist to make sure the work is feasible and appropriate. - write it to convince someone in a *different field* why the experiment is important. - include references to previous work to show that you know the field. --Matt _______________________________________________ Ifeffit mailing list Ifeffit(a)millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit

1 0