HDF5 Event File Format
HDF5 is a binary data format, organised similarly to a data base.
Some modern matrix element generators support the output of
parton-level events in the HDF5 format.
The information is stored in the Les Houches Accord in close
analogy to (default) ASCII Les Houches Event Files.
The HDF5 Les Houches event reader derives from the
class and acts as a replacement of the Pythia Les Houches reader.
Versions As the LHAHDF5 format is relatively new, no clear
version numbering standard has solidified. The reader attempts at
guessing the format, but there is no guarantee that the correct
version can be determined. For full control, the version can be set
LHAHDF5 version number of the event file in the format X.Y.Z.
The Pythia-internal version numbering scheme is:
- The event file contains an
index group, in which the indices
of the particles in a single event are stored. The indices refer to the
- No weight variations are supported. Event weights are a single
floating point number per event stored in the
index group is removed.
- Weight variations are supported. Event weights are a stored in a,
possibly one-dimensional, array per event in the
HDF5 Event File Usage
A number of external libraries are required to read HDF5 LHA event
files. The MPICH implementation of
MPI (message passing interface) is required, and its location can be
specified with the
--with-mpich flags passed to the
configure script. Similarly, the HDF5 library and development
headers are required and can be configured with the
--with-hdf5 options. Finally, the
header library is also required which can be configured with the
--with-highfive options. Currently, version 2.2 is
supported and can be downloaded as follows.
curl -OL https://github.com/BlueBrain/HighFive/archive/v2.2.zip \
&& unzip v2.2.zip && mv HighFive-2.2/include/highfive . \
&& rm -r HighFive-2.2/ && rm v2.2.zip