C++ & Python API for Scientific I/O with openPMD

openPMD is an open meta-data schema that provides meaning and self-description for data sets in science and engineering. See the openPMD standard for details of this schema.

This library provides a reference API for openPMD data handling. Since openPMD is a schema (or markup) on top of portable, hierarchical file formats, this library implements various backends such as HDF5, ADIOS2 and JSON. Writing & reading through those backends and their associated files are supported for serial and MPI-parallel workflows.

Usage

C++

#include <openPMD/openPMD.hpp>
#include <iostream>

// ...

auto s = openPMD::Series("samples/git-sample/data%T.h5", openPMD::Access::READ_ONLY);

for( auto const & [step, it] : s.iterations ) {
    std::cout << "Iteration: " << step << "\n";

    for( auto const & [name, mesh] : it.meshes ) {
        std::cout << "  Mesh '" << name << "' attributes:\n";
        for( auto const& val : mesh.attributes() )
            std::cout << "    " << val << '\n';
    }

    for( auto const & [name, species] : it.particles ) {
        std::cout << "  Particle species '" << name << "' attributes:\n";
        for( auto const& val : species.attributes() )
            std::cout << "    " << val << '\n';
    }
}

Python

import openpmd_api as io

# ...

series = io.Series("samples/git-sample/data%T.h5", io.Access.read_only)

for k_i, i in series.iterations.items():
    print("Iteration: {0}".format(k_i))

    for k_m, m in i.meshes.items():
        print("  Mesh '{0}' attributes:".format(k_m))
        for a in m.attributes:
            print("    {0}".format(a))

    for k_p, p in i.particles.items():
        print("  Particle species '{0}' attributes:".format(k_p))
        for a in p.attributes:
            print("    {0}".format(a))

More!

Curious? Our manual shows full read & write examples, both serial and MPI-parallel!

Dependencies

Required:

CMake 3.15.0+
C++17 capable compiler, e.g., g++ 7+, clang 7+, MSVC 19.15+, icpc 19+, icpx

Shipped internally in share/openPMD/thirdParty/:

Catch2 2.13.10+ (BSL-1.0)
pybind11 2.10.1+ (new BSD)
NLohmann-JSON 3.9.1+ (MIT)
toml11 3.7.1+ (MIT)

I/O backends:

JSON
HDF5 1.8.13+ (optional)
ADIOS2 2.7.0+ (optional)

while those can be built either with or without:

MPI 2.1+, e.g. OpenMPI 1.6.5+ or MPICH2

Optional language bindings:

Python:
- Python 3.7 - 3.11
- pybind11 2.10.1+
- numpy 1.15+
- mpi4py 2.1+ (optional, for MPI)
- pandas 1.0+ (optional, for dataframes)
- dask 2021+ (optional, for dask dataframes)
CUDA C++ (optional, currently used only in tests)

Installation

Our community loves to help each other. Please report installation problems in case you should get stuck.

Choose one of the install methods below to get started:

Spack

# optional:               +python -adios2 -hdf5 -mpi
spack install openpmd-api
spack load openpmd-api

Conda

# optional:                      OpenMPI support  =*=mpi_openmpi*
# optional:                        MPICH support  =*=mpi_mpich*
conda create -n openpmd -c conda-forge openpmd-api
conda activate openpmd

Brew

brew tap openpmd/openpmd
brew install openpmd-api

PyPI

On very old macOS versions (<10.9) or on exotic processor architectures, this install method compiles from source against the found installations of HDF5, ADIOS2, and/or MPI (in system paths, from other package managers, or loaded via a module system, ...).

# we need pip 19 or newer
# optional:                   --user
python3 -m pip install -U pip

# optional:                        --user
python3 -m pip install openpmd-api

If MPI-support shall be enabled, we always have to recompile:

# optional:                                    --user
python3 -m pip install -U pip setuptools wheel
python3 -m pip install -U cmake

# optional:                                                                   --user
openPMD_USE_MPI=ON python3 -m pip install openpmd-api --no-binary openpmd-api

For some exotic architectures and compilers, you might need to disable a compiler feature called link-time/interprocedural optimization if you encounter linking problems:

export CMAKE_INTERPROCEDURAL_OPTIMIZATION=OFF
# optional:                                                --user
python3 -m pip install openpmd-api --no-binary openpmd-api

Additional CMake options can be passed via individual environment variables, which need to be prefixed with openPMD_CMAKE_.

From Source

openPMD-api can also be built and installed from source using CMake:

git clone https://github.com/openPMD/openPMD-api.git

mkdir openPMD-api-build
cd openPMD-api-build

# optional: for full tests, with unzip
../openPMD-api/share/openPMD/download_samples.sh

# for own install prefix append:
#   -DCMAKE_INSTALL_PREFIX=$HOME/somepath
# for options append:
#   -DopenPMD_USE_...=...
# e.g. for python support add:
#   -DopenPMD_USE_PYTHON=ON -DPython_EXECUTABLE=$(which python3)
cmake ../openPMD-api

cmake --build .

# optional
ctest

# sudo might be required for system paths
cmake --build . --target install

The following options can be added to the cmake call to control features. CMake controls options with prefixed -D, e.g. -DopenPMD_USE_MPI=OFF:

CMake Option	Values	Description
`openPMD_USE_MPI`	AUTO/ON/OFF	Parallel, Multi-Node I/O for clusters
`openPMD_USE_HDF5`	AUTO/ON/OFF	HDF5 backend (`.h5` files)
`openPMD_USE_ADIOS2`	AUTO/ON/OFF	ADIOS2 backend (`.bp` files in BP3, BP4 or higher)
`openPMD_USE_PYTHON`	AUTO/ON/OFF	Enable Python bindings
`openPMD_USE_INVASIVE_TESTS`	ON/OFF	Enable unit tests that modify source code ¹
`openPMD_USE_VERIFY`	ON/OFF	Enable internal VERIFY (assert) macro independent of build type ²
`openPMD_INSTALL`	ON/OFF	Add installation targets
`openPMD_INSTALL_RPATH`	ON/OFF	Add RPATHs to installed binaries
`Python_EXECUTABLE`	(newest found)	Path to Python executable

¹ e.g. changes C++ visibility keywords, breaks MSVC ² this includes most pre-/post-condition checks, disabling without specific cause is highly discouraged

Additionally, the following libraries are shipped internally. The following options allow to switch to external installs:

CMake Option	Values	Library	Version
`openPMD_USE_INTERNAL_CATCH`	ON/OFF	Catch2	2.13.10+
`openPMD_USE_INTERNAL_PYBIND11`	ON/OFF	pybind11	2.10.1+
`openPMD_USE_INTERNAL_JSON`	ON/OFF	NLohmann-JSON	3.9.1+
`openPMD_USE_INTERNAL_TOML11`	ON/OFF	toml11	3.7.1+

By default, this will build as a shared library (libopenPMD.[so|dylib|dll]) and installs also its headers. In order to build a static library, append -DBUILD_SHARED_LIBS=OFF to the cmake command. You can only build a static or a shared library at a time.

By default, the Release version is built. In order to build with debug symbols, pass -DCMAKE_BUILD_TYPE=Debug to your cmake command.

By default, tests, examples and command line tools are built. In order to skip building those, pass OFF to these cmake options:

CMake Option	Values	Description
`openPMD_BUILD_TESTING`	ON/OFF	Build tests
`openPMD_BUILD_EXAMPLES`	ON/OFF	Build examples
`openPMD_BUILD_CLI_TOOLS`	ON/OFF	Build command-line tools
`openPMD_USE_CUDA_EXAMPLES`	ON/OFF	Use CUDA in examples

Linking to your project

The install will contain header files and libraries in the path set with -DCMAKE_INSTALL_PREFIX.

CMake

If your project is using CMake for its build, one can conveniently use our provided openPMDConfig.cmake package, which is installed alongside the library.

First set the following environment hint if openPMD-api was not installed in a system path:

# optional: only needed if installed outside of system paths
export CMAKE_PREFIX_PATH=$HOME/somepath:$CMAKE_PREFIX_PATH

Use the following lines in your project's CMakeLists.txt:

# supports:                       COMPONENTS MPI NOMPI HDF5 ADIOS2
find_package(openPMD 0.15.0 CONFIG)

if(openPMD_FOUND)
    target_link_libraries(YourTarget PRIVATE openPMD::openPMD)
endif()

Alternatively, add the openPMD-api repository source directly to your project and use it via:

add_subdirectory("path/to/source/of/openPMD-api")

target_link_libraries(YourTarget PRIVATE openPMD::openPMD)

For development workflows, you can even automatically download and build openPMD-api from within a depending CMake project. Just replace the add_subdirectory call with:

include(FetchContent)
set(CMAKE_POLICY_DEFAULT_CMP0077 NEW)
set(openPMD_BUILD_CLI_TOOLS OFF)
set(openPMD_BUILD_EXAMPLES OFF)
set(openPMD_BUILD_TESTING OFF)
set(openPMD_BUILD_SHARED_LIBS OFF)  # precedence over BUILD_SHARED_LIBS if needed
set(openPMD_INSTALL OFF)            # or instead use:
# set(openPMD_INSTALL ${BUILD_SHARED_LIBS})  # only install if used as a shared library
set(openPMD_USE_PYTHON OFF)
FetchContent_Declare(openPMD
  GIT_REPOSITORY "https://github.com/openPMD/openPMD-api.git"
  GIT_TAG        "0.15.0")
FetchContent_MakeAvailable(openPMD)

Manually

If your (Linux/OSX) project is build by calling the compiler directly or uses a manually written Makefile, consider using our openPMD.pc helper file for pkg-config, which are installed alongside the library.

First set the following environment hint if openPMD-api was not installed in a system path:

# optional: only needed if installed outside of system paths
export PKG_CONFIG_PATH=$HOME/somepath/lib/pkgconfig:$PKG_CONFIG_PATH

Additional linker and compiler flags for your project are available via:

# switch to check if openPMD-api was build as static library
# (via BUILD_SHARED_LIBS=OFF) or as shared library (default)
if [ "$(pkg-config --variable=static openPMD)" == "true" ]
then
    pkg-config --libs --static openPMD
    # -L/usr/local/lib -L/usr/lib/x86_64-linux-gnu/openmpi/lib -lopenPMD -pthread /usr/lib/libmpi.so -pthread /usr/lib/x86_64-linux-gnu/openmpi/lib/libmpi_cxx.so /usr/lib/libmpi.so /usr/lib/x86_64-linux-gnu/hdf5/openmpi/libhdf5.so /usr/lib/x86_64-linux-gnu/libsz.so /usr/lib/x86_64-linux-gnu/libz.so /usr/lib/x86_64-linux-gnu/libdl.so /usr/lib/x86_64-linux-gnu/libm.so -pthread /usr/lib/libmpi.so -pthread /usr/lib/x86_64-linux-gnu/openmpi/lib/libmpi_cxx.so /usr/lib/libmpi.so
else
    pkg-config --libs openPMD
    # -L${HOME}/somepath/lib -lopenPMD
fi

pkg-config --cflags openPMD
# -I${HOME}/somepath/include

Author Contributions

openPMD-api is developed by many people. It was initially started by the Computational Radiation Physics Group at HZDR as successor to libSplash, generalizing the successful HDF5 & ADIOS1 implementations in PIConGPU. The following people and institutions contributed to openPMD-api:

Axel Huebl (HZDR, now LBNL): project lead, releases, documentation, automated CI/CD, Python bindings, Dask, installation & packaging, prior reference implementations
Franz Poeschel (CASUS): JSON & ADIOS2 backend, data staging/streaming, reworked class design
Fabian Koller (HZDR): initial library design and implementation with HDF5 & ADIOS1 backend
Junmin Gu (LBNL): non-collective parallel I/O fixes, ADIOS improvements, benchmarks

Further thanks go to improvements and contributions from:

Carsten Fortmann-Grote (EU XFEL GmbH, now MPI-EvolBio): draft of our Python unit tests
Dominik Stańczak (Warsaw University of Technology): documentation improvements
Ray Donnelly (Anaconda, Inc.): support on conda packaging and libc++ quirks
James Amundson (FNAL): compile fix for newer compilers
René Widera (HZDR): design improvements for initial API design
Erik Zenker (HZDR): design improvements for initial API design
Sergei Bastrakov (HZDR): documentation improvements (windows)
Rémi Lehe (LBNL): package integration testing on macOS and Linux
Lígia Diana Amorim (LBNL): package integration testing on macOS
Kseniia Bastrakova (HZDR): compatibility testing
Richard Pausch (HZDR): compatibility testing, documentation improvements
Paweł Ordyna (HZDR): report on NVCC warnings
Dmitry Ganyushin (ORNL): Dask prototyping & ADIOS2 benchmarking
John Kirkham (NVIDIA): Dask guidance & reviews
Erik Schnetter (PITP): C++ API bug fixes
Jean Luca Bez (LBNL): HDF5 performance tuning
Bernhard Manfred Gruber (CERN): CMake fix for parallel HDF5
Nils Schild (IPP): CMake improvements for subprojects

Grants

The openPMD-api authors acknowledge support via the following programs. Supported by the CAMPA collaboration, a project of the U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research and Office of High Energy Physics, Scientific Discovery through Advanced Computing (SciDAC) program. Previously supported by the Consortium for Advanced Modeling of Particles Accelerators (CAMPA), funded by the U.S. DOE Office of Science under Contract No. DE-AC02-05CH11231. Supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of two U.S. Department of Energy organizations (Office of Science and the National Nuclear Security Administration). This project has received funding from the European Unions Horizon 2020 research and innovation programme under grant agreement No 654220. This work was partially funded by the Center of Advanced Systems Understanding (CASUS), which is financed by Germany's Federal Ministry of Education and Research (BMBF) and by the Saxon Ministry for Science, Culture and Tourism (SMWK) with tax funds on the basis of the budget approved by the Saxon State Parliament.

Transitive Contributions

openPMD-api stands on the shoulders of giants and we are grateful for the following projects included as direct dependencies:

ADIOS2 by S. Klasky, N. Podhorszki, W.F. Godoy (ORNL), team, collaborators and contributors
Catch2 by Phil Nash, Martin Hořeňovský and contributors
HDF5 by the HDF group and community
json by Niels Lohmann and contributors
toml11 by Toru Niina and contributors
pybind11 by Wenzel Jakob (EPFL) and contributors
all contributors to the evolution of modern C++ and early library preview developers, e.g. Michael Park (Facebook)
the CMake build system and contributors
packaging support by the conda-forge, PyPI and Spack communities, among others
the openPMD-standard by Axel Huebl (HZDR, now LBNL) and contributors

lucafedeli88/openPMD-api