Skip to content

flowTorch 1.11.0

Latest
Latest

Choose a tag to compare

View all tags
@AndreWeiner AndreWeiner released this 03 Nov 17:05
v1.11.0
edafcdc
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Highlights

  • Native Tecplot TDV71/TDV112 finite-element readers with direct spatial field and connectivity slices.
  • PLT-to-HDF5/XDMF conversion through PLT2HDF5 and flowtorch-plt-to-hdf5, preserving separate zones and reusable Float64 geometry with independently selected Float32/Float64 fields.
  • Lossless gzip or optional Zstandard compression, configurable levels and shuffle, separately for geometry and fields.
  • Validated append of new snapshots to complete split-file datasets, preserving old chunks and geometry and refreshing the XDMF time series.
  • Physical-field, complex-mode and SPOD animation export against existing geometry.
  • Expanded native OpenFOAM ASCII/binary/gzip reading, architecture-aware types, uniform fields and spatial slicing.
  • Updated workflow guides, examples, glossary and navigation.

Compatibility

Existing unversioned HDF5 files remain readable. Direct TecplotDataloader calls retain their ParaView default; use backend="native" or "auto" for the supported native formats. Zstd requires the compression extra and compatible native filters in ParaView/pvserver. Validated append requires split geometry/field output, rejects duplicate times and incompatible data, and does not provide live readers, concurrent writers or power-loss recovery.

The native Tecplot backend supports selected FE-block layouts; ordered zones, polyhedral face maps, bit-packed variables and face-neighbour records remain unsupported. ASCII/gzip OpenFOAM slices scan or decompress preceding values, so they are not constant-time random access. Default OpenFOAM conversion exports supported volume fields and skips face fields such as phi.

Validation

The complete available suite passed locally on Python 3.10, 3.12 and 3.14: 1,045 passed and five CUDA-dependent skips per environment, including integration tests on disposable dataset copies. Formatting, lint, mypy, documentation and package validation passed. The installed 1.11.0 wheel passed conversion, gzip/Zstd, append, exact readback, unchanged-geometry, complex-export and CLI smoke tests outside the checkout.

Separate application checks cover surface and volume PLT data and representative ParaView readback. On the tested complete Skander volume conversion, Zstd level 3 plus shuffle reduced HDF5 storage from 29.02 GB to 14.06 GB with exact readback. See the repository's validation records for the scope and limitations of each check.

Documentation · Detailed release notes

Assets 2
Loading