Fourier machinery for the uniform periodic axes of PICurv's Cartesian grids. More...
Go to the source code of this file.
Namespaces | |
| namespace | picurv_cli |
| namespace | picurv_cli.periodic_spectral |
Functions | |
| picurv_cli.periodic_spectral.drop_imported_package (str package_name) | |
| Remove a failed/partial import package tree from sys.modules. | |
| picurv_cli.periodic_spectral.prune_incompatible_python_site_paths (paths) | |
| Remove site-package paths for a different Python major/minor version. | |
| picurv_cli.periodic_spectral.require_numpy () | |
| Import NumPy on first use, retrying without other Pythons' site-packages. | |
| picurv_cli.periodic_spectral.validate_spectral_grid (nodes, transform_axes=(0, 1, 2), subject="spectra require") | |
| Validate the uniform axis-aligned Cartesian grid an FFT on periodic axes requires. | |
| picurv_cli.periodic_spectral.spectral_symbols (cells_kji, lengths_xyz) | |
| Build continuum and centered-discrete Fourier symbols. | |
| picurv_cli.periodic_spectral._resample_axis (values, storage_axis, target_count, length, shift) | |
| Move one periodic axis of an array of samples to another uniform resolution. | |
| picurv_cli.periodic_spectral.filter_transfer (continuum_symbols, filter_spec) | |
| Transfer function of a resampling filter on the target grid's wavenumbers. | |
| picurv_cli.periodic_spectral.fourier_resample (values, target_cells_kji, lengths_xyz, source_first_xyz, target_first_xyz, filter_spec=None) | |
| Resample a triply periodic vector field onto another uniform grid, optionally filtered. | |
Variables | |
| picurv_cli.periodic_spectral._SITE_PYTHON_VERSION = re.compile(r"python(?:-)?(\d+)\.(\d+)", re.IGNORECASE) | |
| picurv_cli.periodic_spectral._NUMPY_MODULE = None | |
| tuple | picurv_cli.periodic_spectral.RESAMPLE_FILTERS = ("none", "sharp", "gaussian", "box") |
Fourier machinery for the uniform periodic axes of PICurv's Cartesian grids.
Shared by ic.gen, which synthesizes and resamples initial conditions, and spectra.gen, which measures spectra; both load it from picurv_cli/. Both need the same answer to three questions: which axes may be transformed, what the continuum and PICurv-discrete wavenumbers are, and how a periodic field is moved between resolutions. Arrays are in PICurv storage order, [k, j, i, component]; lengths and wavenumbers are in Cartesian x, y, z order. Every function imports NumPy through require_numpy(), so importing this module never imports NumPy itself; that import, with its retry, is defined here once for the conductor and every generator.
Definition in file periodic_spectral.py.