[Ifeffit] Amplitude in Artemis

ynliu at eps.mcgill.ca ynliu at eps.mcgill.ca
Thu Mar 2 18:43:00 CST 2006


Hello Bruce,

I have an opposite question about the standard selection in the linear fitting
feature of Athena. Instead of when to exclude one standard, when should I add
in more standards? For example, a given sample spectrum, I can fit it with 2
standards, or 3 standards or 4 standards without seeing that warning message.
The 4 standards fitting is the best one according to the statistical
parameters. But does this mean I have so many types of species in my sample or
just because similar to other fittings, the more variables you add, the better
fittings you'll get? In short, how can we assess a linear fitting, how much
confidence do we have for the fitting results to be close to the reality? Or
the linear fitting is actually giving us a first guess? Thank you very much!

Yanan



Quoting Bruce Ravel <bravel at anl.gov>:

>
> An additional US$0.02 worth: The linear combination fits tend to be
> pretty simple, well defined, and well constrained in a statistical
> sense.  If you get one of those scary sounding messages (with the
> scary colored letters ;-) it is usually telling you something about
> your fit.  Hopefully, that something is not that the program is
> broken, that your standards are broken, or that you aren't using the
> program correctly.  Rather, it should be telling you something about
> your choice of standards.
>
> If you have taken care to normalize your data well and consistently,
> then a fit that pins a weight at zero probably means that that
> standard does not exist in the sample.  Thus, the warning is intended
> to be useful information.
>
> Athena's LCF utility has a lot of bells and whistles, some of which
> Dave mentions.  The point of having all those knobs to turn is to be
> able to poke corners of the fit and to try to understand why a fit is
> turning out the way it is turning out.  You should not hesitate to
> turn those knobs -- they aften provide useful information.  Everything
> is explained on the document page for the LCF utility, albeit
> tersely.  If the effect of any of the knobs is unclear, don't hesitate
> to ask the list.
>
> B
>
> On Thursday 02 March 2006 16:21, bsin1033 at mail.usyd.edu.au wrote:
> > Dave,
> >
> > Thank you for your comments. By doing your way it works but one of the
> > standard is not quite right, as you have also suggested. We will rerun the
> > standards and try it again.
> >
> > Cheers,
> >
> > Balwant
> >
> > Quoting "Barton, David (DG)" <DGBarton at dow.com>:
> > > Balwant,
> > >
> > > I have done a number of LCF using Athena and have found that this error
> > > message is typically associated with the rejection of one of your
> > > standards as a possible component in the fit.  This error message may be
> > > avoided in most cases if you uncheck the options of --weights between 0
> > > &1 and --Force weights to sum to 1 AND you put in an appropriate initial
> > > guess for the fit.  When you rerun the fit with these options unchecked,
> > > it should be obvious what caused the error since one or more of your
> > > standard spectra will get a negative weight.  When this occurs it is
> > > probably a good sign that the standard is not appropriate in the fit and
> > > some other standard should be used.
> > >
> > > Hope this helps,
> > > Dave
> > >
> > > > -----Original Message-----
> > > > From: ifeffit-bounces at millenia.cars.aps.anl.gov
> > > > [mailto:ifeffit-bounces at millenia.cars.aps.anl.gov] On Behalf
> > > > Of bsin1033 at mail.usyd.edu.au
> > > > Sent: Thursday, March 02, 2006 3:41 PM
> > > > To: XAFS Analysis using Ifeffit; Kelly, Shelly D.
> > > > Cc: XAFS Analysis using Ifeffit
> > > > Subject: RE: [Ifeffit] Amplitude in Artemis
> > > >
> > > >
> > > > Hi Ifeffitters,
> > > >
> > > > I was trying to use the linear combination fitting function
> > > > in Athena to quantifying Cr(VI) and Cr(III) species. I used
> > > > two endmembers of Cr species as the two standards but the
> > > > program fails and gives error message "The fit flagged a
> > > > warning--probably that error bars could not be calculated.
> > > > That is probably an indication that one or more of your
> > > > standards are inappropriate for this fit." It appears I am
> > > > either using the wrong settings or even wrong function to do
> > > > the job, any suggestions. We have additional 3 spectra with
> > > > variable combinations of Cr(VI) and
> > > > Cr(III) as well.
> > > >
> > > > Cheers,
> > > >
> > > > Balwant Singh
> > > >
> > > > ----------------------------------------------------------------
> > > > This message was sent using IMP, the Internet Messaging
> > > > Program. _______________________________________________
> > > > Ifeffit mailing list
> > > > Ifeffit at millenia.cars.aps.anl.gov
> > > > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
> > >
> > > _______________________________________________
> > > Ifeffit mailing list
> > > Ifeffit at millenia.cars.aps.anl.gov
> > > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
> >
> > ----------------------------------------------------------------
> > This message was sent using IMP, the Internet Messaging Program.
> > _______________________________________________
> > Ifeffit mailing list
> > Ifeffit at millenia.cars.aps.anl.gov
> > http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>
> --
>  Bruce Ravel  ---------------------------------------------- bravel at anl.gov
>
>  Molecular Environmental Science Group, Building 203, Room E-165
>  MRCAT, Sector 10, Advanced 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 at millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>




----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.



More information about the Ifeffit mailing list