Meetings | Nexus workshop (2001)

canSAS and NeXus can empathise!

Summary of the NeXus workshop at the PSI, Villigen, 20/21 March 2001

unofficial version: R. Ghosh, 26th March 2001

The NeXus project to simplify use of the Hierachical Data Format (HDF) of the National Center for Supercomputing Applications (NCSA), Urbana for Neutron and X-ray scattering measurements has lead to the development of the NeXus library (API) to write and read data files.

The meeting at the PSI, Villigen, brought together the developers and groups of scientists interested in using this format for raw data storage, treated data storage, and storage of the results of simulation experiments in a compatible fashion, aiding interchange between sites and groups.

In separate working groups the developers debated the future of the API, while the Users, typically instruments scientists, were posed the problem of nomenclature of instrument parameters, and establishing a mechanism for converging rapidly on an adequate description for basic instrument types. This was becoming of increasing importance with new facilities like the FRM-II reactor at Munich about to start, the SLS at the PSI, and the SNS in the US also advancing with large simulation activities.

For SAS activities it became evident that the developments by the NeXus teams could be used directly for treated data, and that, in the longer term, it would be possible to modify SAS instruments to store raw data in NeXus form.

Results from the Developers

One of the major hurdles posed by NCSA is the evolution to HDF5, which is a file format which is no longer directly compatible with previous versions of HDF, but which is itself a complete rewrite, simplifying, improving, and resolving a number of technical problems of earlier versions (text storage, maximum file size etc).

The NeXus developers were keen to advance to use this new version, and Users were encouraged to leave the problems of compatibility to the experts; some laboratories like the LLB, and PSI which have been storing raw data in NeXus/HDF4 form for some time would not be able to transform existing files to the new HDF5.

Most important for the success of NeXus was the need to develop applications; even linking existing display tools (Origen, IGOR etc) for easy use with these data was of great importance, as was the need to be able to import data from these toolsets.

Users requests

Whilst the aim of completeness for instrument datasets was a good thing, it was also important to envisage identifying a minimal set of parameters to allow manipulation of data. There was a general wish to be able to add extra data, with the possibility of some limited editing facility, and allowing, notably, some subsequent annotation.

Implications for canSAS

The proposal here for storing complex data can use all the technology developed by the NeXus team, and use their ongoing support simply by placing the complex data in a section with a class name (NXdata) which has been designated as the standard placement for spectral information. Axis information and other attributes can also be stored here, together, we hope, with annotations.

A number of HDF browsers can already display such multi-axis data with cuts and surface plots. One can then hope progressively to add the other classes describing the instrument (NXinstrument) and samples (NXsample) etc., conforming to the complete NeXus file. Starting with this limited form, but shared across the large SAS community, should give great encouragement to the NeXus developers.