Looking for: API / Bindings / Software using HDF
Gerd Wellenreuther
first I would like to report to you that also the german PNI-institutes
are seriously considering to establish HDF5 + NeXus as their standard
data format. PNI stands for Photons, Neutrons and Ions, and part are all
german large scale facilities providing these beams (in german:
http://www.helmholtz.de/forschung/struktur_der_materie/forschung_mit_photonen_neutronen_und_ionen_pni/).
Second, partly as a result of the workshop, I would like to gather a
list of APIs / Bindings which exist for both HDF5 + NeXus, as well as a
list of applications which are already or in the future capable of
supporting HDF5 or NeXus. I started to add whatever came to my mind on
http://groups.google.com/group/mahid/web/important-web-pages . Please,
have a look, because I will have overseen some things, and I need input
from different people about their software projects:
@Tom: Axil will support HDF5 - can you tell us when?
@Matt + Bruce: What about FEFF / ATHENA / ARTEMIS? Will they at one
point support HDF5 or NeXus? Can you tell me when?
@Gunthard: What about DPDAC, when will the software be available, when
will it be able to read HDF5 / NeXus?
@Chris + Burkhard: Maybe you could write something about aXis2000, or
ask Adam Hitchcock to do that. Is HDF5-support planned?
@Stefan: What about MAPS?
@Majid + Alain: Maybe you could provide an overview which software from
SOLEIL is available, which is capable to read generic NeXus or
HDF5-files, and is of general interest?
Cheers, Gerd
Vicente Sole
I guess you can add to the list of applications supporting HDF5:
ViTables (made for)
PyMca (since the 4.4.0 release)
Armando
Quoting Gerd Wellenreuther <Gerd.Well...@desy.de>:
> --
> You received this message because you are subscribed to the Google
> Groups "Methods for the analysis of hyperspectral image data" group.
> To post to this group, send email to ma...@googlegroups.com.
> To unsubscribe from this group, send email to
> mahid+un...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/mahid?hl=en.
DUMAS Paul
Bon week end !
Paul
Depuis mon iPhone-From my iPhone
Vicente Sole
Anyways, PyTables does not build on top of h5py as the web page suggest.
PyTables and h5py are two different python bindings to HDF5.
Armando
Quoting Gerd Wellenreuther <Gerd.Well...@desy.de>:
Bruce Ravel
> @Matt + Bruce: What about FEFF / ATHENA / ARTEMIS? Will they at one
> point support HDF5 or NeXus? Can you tell me when?
>
Hi Gerd,
Matt and I have just begun talking about this and I need to bring myself up to
speed on how to use HDF5. That said, doing so is very attractive to me for
both Athena and Artemis. As for when, that I cannot yet say....
B
Gerd Wellenreuther
just knowing that you will most probably provide some kind of
HDF5-support in the future (whether near or distant) is an important
information for me.
Thanks, Gerd
Bruce Ravel schrieb:
Matt Newville
> @Matt + Bruce: What about FEFF / ATHENA / ARTEMIS? Will they at one point
> support HDF5 or NeXus? Can you tell me when?
To follow up on Bruce's message, and just to clarify for posterity,
there are several separate programs here:
FEFF will probably never support HDF5 -- it does not need to read
experimental data.
IFEFFIT will likely support HDF5 files in its next major version.
I'd guess that's at least
a year away from an alpha release, but it is in progress.
ATHENA / ARTEMIS (GUIs on top of IFEFFIT and FEFF) may support HDF5 sooner
than is available from the next major version of IFEFFIT.
Since "Support HDF5" is a bit vague, let me explain that Bruce and I
are thinking about using HDF5 for storing groups of XAFS spectra. So
"support HDF5" may only mean support for those HDF5 files that have
such an XAFS-specific (and/or MAHID-specific) layout and semantics.
Like, I doubt that ATHENA/ARTEMIS would support browsing and
extracting slices of data from arbitrary HDF5 files in the near-term.
I am less sure about NeXus support, as I don't know what exactly that
would mean. I doubt these programs would use the NeXus API (For one
thing, Athena and Artemis are in Perl -- I think bindings do not
exist. For Ifeffit2, I have not looked at the NeXus Python modules).
Having a NeXus Application Definition for XAFS is worth considering,
but I doubt it would ever get widespread (XAFS community) support.
OTOH, if NeXus is read as "HDF5 + semantics that sort of follow a
convention", then the HDF5 files these programs use might follow those
conventions.
Cheers,
--Matt
"V. Armando Solé"
Just my comments.
I do not see an issue about using or not the NeXus API. I do not use
their python API for instance because h5py and phynx give me very nice
access.
If Bruce and you decide not to have to browse for the XAFS information
into the HDF5 files it is even better: that will enforce the coming
"standard". On PyMca I have to browse for the data information because
of the more generic aspect of the program.
I do see NeXus as "HDF5 + semantics": if something I want is already in
NeXus I take it, if not I write it anyways without conflicting with it.
Let's put it this way, Bruce and you figure out what would you like to
have in the HDF5 file to perform the analysis with your codes. I do not
foresee major difficulties to arrive to write that information into an
"HDF5 + semantics" file that at the same time is described by a NeXus
application definition. Probably all what NeXus compliance would require
is to have the application definition information into an "HDF5 folder"
with an NX_CLASS attribute set to NXentry. That should not be a great
effort.
It is just my opinion, but if it is widespread in the XAFS community or
not will depend on if Bruce and you support it. By one side scientists
want to use your programs, by other side their raw data will be in HDF5
format, why not to write them in a way allowing direct access?
Best regards,
Armando