THRILL decided to adopt the living standard for a meta-data rich structure data storage developed within the HELPMI project for the acquisition and exchange of laser operations data.
The HELPMI format was validated in 2024, within a pilot implementation at GSI and confirmed by machine learning (ML) practitioners within THRILL as suitable for exchanging data of laser operations, such as beam-profile images or spectrums, accompanied by machine-readable meta-data.

The HELPMI standard is compatible with and extends both the NeXus standard, used primarily by the accelerator community, and the openPMD standard, used in plasma and laser-plasma simulations. HELPMI defines an ontology for labeling types of measurements and various measurement and device parameters, that describe in the process by which the stored data was acquired. Consistent labeling of provided information and the inclusion of meta-data, such das device settings, is crucial for scientists to be able to compare the operation of high-power lasers at different facilities and for analysis pipelines to function data from different sources or to be adapted quickly at different facilites.

Adoption of the standard at facilities is an ongoing effort. The amount of actually stored meta data in a HELPMI file depends on the concrete data acquisition setup and can increased over time as more parameters are made available for readout from the control system. The HELPMI ontology is a living standard, which allows THRILL to extend it with terms required to more accurately describe details of high-power laser system operation, where terms do not exist in the current HELPMI ontology.

Compatibility with the NeXus standard, means that tools developed in the accelerator community to handle NeXus files can be used with HELPMI files, if the [HDF5 format]( ), which NeXus is based on, is used as backend. The openPMD API supports multiple storage backends, including HDF5. Using adios2 as the backend, allows direct streaming of data a high data rate experiments into analysis pipelines, in such a way, that the analysis pipelines does not need to adapt to whether a data from stream or from a stored file are ingested.

Overall, adoption of the HELMPI standard is expected to lead to more portable data pipelines with less programming overhead when ingesting data; and those meta-data rich files can more readily be published in a Findable, Accessible, Interoperable, and Reuseable (FAIR) manner.

HELPMI: https://laser-plasma-metadata.org/
openPMD: https://www.openpmd.org/
NeXus: https://www.nexusformat.org/
ADIOS: https://github.com/ornladios/ADIOS2
HDF5: https://www.hdfgroup.org/solutions/hdf5/