[Ifeffit] Calculation of SPOIL value for the reconstruction of standard spectra

[LANDROT Gautier]

Hello Emmanuelle,

The PCA module featured in the current version of Athena does not calculate SPOIL values. This is actually one of the things that Bruce Ravel has listed in his current "Demeter/Athena/Artemis to do list" of his website, see section 4.5.3.

For SPOIL calculations I recommend you to use the PCA/TT module of Sixpack. According to Sam Webb, the latest Sixpack (version 1.4) employs a new/different PCA algorithm from the scikit-learn Python package. Using personal data set, I found that the PCA/target transformation results obtained from the new Sixpack v 1.4 are different from those obtained from an older version (v 1.2.10). It would be interesting to know whether other Sixpack users have come up to the same conclusion.

The SPOIL values obtained from a homemade code, following the method detailed in the Malinowski 1978 reference article, are very similar to SPOIL values reported in the Malinowski 1978 paper using mass spec data set from Ritter et al. 1976, and also SPOIL values obtained from an old Sixpack version (v 1.2.10) using personal xafs data set. Therefore, if you have installed on your PC the latest Sixpack version (v 1.4), you may want to select, in the menu bar on top of PCA/TT Sixpack module, "Rotation" then "Use Old PCA" to calculate SPOIL values. The calculated SPOIL values will be then equal to those obtained from the previous Sixpack versions.

Good luck,

Gautier


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20170126/6aa711c6/attachment.html>