[Ifeffit] XASLIB (was Re: data availability)

Matt Newville newville at cars.uchicago.edu
Wed Apr 27 10:52:13 CDT 2016


Hi Nino,

Thanks for bringing this up and for the kind words.  I agree that not only
is access to measured XAFS spectra a problem now, it's been a problem for a
long time, as evidenced by the Farrel Lytle database.  In fact, it has long
been a topic of discussion in the XAFS community of how to provide such
databases, and developments over the past few years on data formats have
been driven at least partly by the desire to have better spectral
databases.

For everyone: please try out XASLIB at  http://cars.uchicago.edu/xaslib
and let us know if you have any questions, comments, or suggestions for
improving this database.   Right now, there is not a lot of data here, but
it is enough to give you an idea of how it works.

Comments can either come here or as github Issues at the source repository
for this code:
   https://github.com/XraySpectroscopy/XASDataLibrary

There is a bit of documentation and wiki pages at
   https://github.com/XraySpectroscopy/XASDataLibrary/wiki


A longer explanation:

The standard XDI format for XAFS data (largely the work of Bruce)  was
presented at an IXAS workshop (Q2XAFS) in Tsukuba, Japan in 2011.  At the
most recent Q2XAFS meeting in Hamburg last August (as a satellite meeting
of the XAFS conference), I presented an initial version of an on-line
database based on this format, and intended for general use in the
community.    With feedback from that meeting, the version linked above is
now ready for beta testing and comments.

The plan is to move this to the IXAS Portal  http://www.ixasportal.net/ixas/,
but we'd like to get community suggestions and features and testing before
we make that move.

The database uses proper relational database to store the actual data, and
tries to use modern web technologies for the interface.

Some of the main features of this database are:
  -  spectra can be uploaded from XDI files through the web interface.
  -  multiple spectra can be tagged as belonging to the same Suite.
  -  individual spectra or Suites can be rated and/or commented on by
     anyone, giving "Amazon-like" reviews.
  -  in order to upload or comment, one has to log in by email address.

The ratings and reviews are meant to provide some mechanism for data
quality assurance, but to avoid having one person "police" the data before
it can be added.  This is trying to recognize that no one person can
actually curate all the spectra, that truly bad data can be best identified
by experts on that system, and also that there is actually some value to
imperfect data.

The requirements for data to be uploaded are fairly minimal except that the
data must be in XDI format (basically so that metadata can be parsed easily
and the columns are well-labeled), and that the monochromator lattice
constant must be provided (so that energy re-calibration can be done if
needed).    We can help translate existing data into XDI so it can be
uploaded into this database.

There are several features to add or improve including:
   - adding some information about energy resolution.
   - being able to export a Suite of spectra as a single zip file or Athena
project.
   - better web page interface to setting citation information.
   - better web page interface to setting sample information.

But suggestions from anyone would be greatly appreciated.

Finally, although the database is meant to provide a public, on-line
databases useful to anyone, and uses modern relational database engines as
the backend, the software can actually run on an individual machine to
provide a local, private database of XAS spectra, with the same web-based
interface.  This is not very well documented at the moment, but if anyone
is interested in trying to set such a database at their home institution,
let me know.

--Matt Newville
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20160427/c6e0a716/attachment-0001.html>


More information about the Ifeffit mailing list