XDMF

XDMF
Name	XDMF
Caption	eXtensible Data Model and Format
Developer	Laboratory for Advanced Scientific Computing
Initial release	2006
Latest release	4.2
Operating system	Cross-platform
Genre	Scientific data format
License	BSD-like

XDMF

XDMF is a hybrid scientific data specification combining XML metadata with binary data descriptions to support large-scale simulation, visualization, and analysis. It enables interoperability among applications by describing topology, geometry, and attribute layouts while delegating heavy-weight array storage to external binary containers. The format is used across high-performance computing ecosystems and integrates with visualization engines, finite element solvers, and workflow managers.

Overview

XDMF was designed to bridge scientific codes and visualization stacks such as ParaView, VisIt, VTK (Visualization Toolkit), EnSight and to interoperate with storage systems like HDF5, NetCDF, POSIX-based filesystems and parallel I/O libraries such as MPI (Message Passing Interface). It formalizes relationships among meshes, fields, and time series for transient simulations produced by solvers like OpenFOAM, ANSYS, Abaqus, and research codes from institutions including Lawrence Berkeley National Laboratory, Argonne National Laboratory, Oak Ridge National Laboratory, and Los Alamos National Laboratory. XDMF is commonly referenced in workflows orchestrated by CMake, Jenkins (software), GitHub, and data portals hosted by National Center for Atmospheric Research or NASA-sponsored projects.

File Format and Structure

The format separates a human-readable XML "light" file from "heavy" binary arrays. The XML layer references binary containers managed by formats such as HDF5, NetCDF, or raw POSIX files and describes layout constructs compatible with visualization engines like ParaView and libraries such as VTK. XDMF uses constructs analogous to XML Schema conventions for element typing and leverages identifiers and URIs for linking to external resources similar to patterns used by W3C recommendations. Typical files include domain descriptions, topology elements, geometry entries, and attribute blocks that map to arrays stored in external datasets handled by parallel libraries like MPI and PHDF5. Tools parse XDMF files using parsing libraries from ecosystems including libxml2, Expat, xerces-c, and language bindings in C++, Python (programming language), Fortran (programming language), and Java (programming language). The format supports time-dependent datasets and multi-grid hierarchies comparable to concepts in Adaptive Mesh Refinement applications developed at Lawrence Livermore National Laboratory and Sandia National Laboratories.

Data Model and Semantics

XDMF's conceptual model distinguishes topology types (structured, unstructured, rectilinear) and geometry representations (point, node-centered, cell-centered), mapping them to memory layouts consumed by renderers like OpenGL backends in ParaView and compute kernels in CUDA and OpenCL. The semantic layer encodes field associations, centering, data precision, endianness, and sparse versus dense storage strategies familiar to authors of solvers such as PETSc, Trilinos, deal.II, and FEniCS Project. Temporal collections are expressed in sequences that tracking simulation time steps similar to frameworks employed by SimGrid and ASPECT. Metadata elements align with cataloguing practices from Digital Object Identifier systems and data management plans advocated by NSF and DOE policies, enabling provenance capture consistent with projects like CODATA and Research Data Alliance.

Implementations and Tools

Multiple open-source and commercial implementations read and write XDMF metadata and coordinate binary payloads. Visualization front-ends include ParaView, VisIt, and Tecplot; libraries and bindings are maintained in projects hosted on GitHub and mirrored through GitLab CI pipelines. Scientific toolkits integrate XDMF support in VTK readers, the HDF5 C API, Python packages such as h5py and NumPy, and language bindings in Boost C++ Libraries wrappers. Conversion utilities exist in repositories maintained by groups at CERN, European Centre for Medium-Range Weather Forecasts, and national labs; batch processing and deployment are often scripted with Bash (Unix shell), Python (programming language), Ansible, or Docker (software). Commercial vendors like Siemens (through Simcenter), ANSYS, and Hexagon AB provide import/export pipelines to integrate with engineering workflows.

Use Cases and Applications

XDMF is used for high-fidelity simulations in domains represented by institutions like European Organization for Nuclear Research, NASA Jet Propulsion Laboratory, National Oceanic and Atmospheric Administration, and research groups developing climate models at NOAA and Met Office. Common applications include computational fluid dynamics in automotive and aerospace projects by companies such as Boeing and Airbus, structural mechanics in civil engineering studies by Arup (engineering) and AECOM, and multiphysics coupling in fusion research at ITER and Princeton Plasma Physics Laboratory. It supports visualization pipelines for medical imaging collaborations with MGH (Massachusetts General Hospital), bioinformatics datasets coordinated with EMBL-EBI, and geoscience meshes used by USGS and Schlumberger. Data management for long-term curation often integrates with archives like Zenodo and institutional repositories coordinated through DataCite.

History and Development

XDMF originated within collaborations among national laboratories and academic centers to address interoperability between simulation codes and visualization tools. Early work involved contributors from Lawrence Livermore National Laboratory, Oak Ridge National Laboratory, and Argonne National Laboratory, while community adoption grew through integration with VTK and ParaView driven by developers associated with Kitware, Inc. and visualization researchers formerly at Los Alamos National Laboratory. The development trajectory mirrors trends in scientific data stewardship promoted by funding agencies such as DOE Office of Science and NSF, and benefited from contributions by open-source communities on platforms like SourceForge and later GitHub. Subsequent revisions refined schema semantics, extended support for parallel I/O with HDF5 and MPI-IO, and improved language bindings inspired by interoperability efforts in projects like HDFGroup and OpenFOAM Foundation.

Category:Scientific file formats