mpi4py_fft.utilities package

Submodules

mpi4py_fft.utilities.generate_xdmf module

mpi4py_fft.utilities.generate_xdmf.generate_xdmf(h5filename, periodic=True, order='paraview')

Generate XDMF-files

Parameters:

• h5filename (str) – Name of hdf5-file that we want to decorate with xdmf
• periodic (bool or dim-sequence of bools, optional) – If true along axis i, assume data is periodic. Only affects the calculation of the domain size and only if the domain is given as 2-tuple of origin+dx.
• order (str) – paraview or visit For some reason Paraview and Visit requires the mesh stored in opposite order in the XDMF-file for 2D slices. Make choice of order here.

mpi4py_fft.utilities.h5py_file module

class mpi4py_fft.utilities.h5py_file.HDF5File(h5name, T, domain=None, mode='a', **kw)

Bases: mpi4py_fft.utilities.file_base.FileBase

Class for reading/writing data to HDF5 format

Parameters:

• h5name (str) – Name of hdf5 file to be created.

• T (PFFT) – Instance of a PFFT class. Must be the same as the space used for storing with HDF5Writer.write.

• domain (Sequence, optional) –

  The spatial domain. Sequence of either

  • 2-tuples, where each 2-tuple contains the (origin, length) of each dimension, e.g., (0, 2*pi).
  • Arrays of coordinates, e.g., np.linspace(0, 2*pi, N). One array per dimension.

• mode (str, optional) – r, w or a for read, write or append. Default is a.

static backend()

open()

read(u, name, **kw)

Read into array u

Parameters:

• u (array) – The array to read into.
• name (str) – Name of array to be read.
• forward_output (bool, optional) – Whether the array to be read is the output of a forward transform or not. Default is False.
• step (int, optional) – Index of field to be read. Default is 0.

write(step, fields, **kw)

Write snapshot step of fields to HDF5 file

Parameters:

• step (int) – Index of snapshot.
• fields (dict) – The fields to be dumped to file. (key, value) pairs are group name and either arrays or 2-tuples, respectively. The arrays are complete arrays to be stored, whereas 2-tuples are arrays with associated global slices.
• forward_output (bool, optional) – Whether fields to be stored are shaped as the output of a forward transform or not. Default is False.

Example

>>> from mpi4py import MPI
>>> from mpi4py_fft import PFFT, HDF5File, newDistArray
>>> comm = MPI.COMM_WORLD
>>> T = PFFT(comm, (15, 16, 17))
>>> u = newDistArray(T, forward_output=False, val=1)
>>> v = newDistArray(T, forward_output=False, val=2)
>>> f = HDF5File('h5filename.h5', T)
>>> f.write(0, {'u': [u, (u, [slice(None), 4, slice(None)])],
...             'v': [v, (v, [slice(None), 5, 5])]})
>>> f.write(1, {'u': [u, (u, [slice(None), 4, slice(None)])],
...             'v': [v, (v, [slice(None), 5, 5])]})

This stores data within two main groups u and v. The HDF5 file will in the end contain groups:

/u/3D/{0, 1}
/u/2D/slice_4_slice/{0, 1}
/v/3D/{0, 1}
/v/1D/slice_5_5/{0, 1}

Note

The list of slices used in storing only parts of the arrays are views of the global arrays.

mpi4py_fft.utilities.nc_file module

class mpi4py_fft.utilities.nc_file.NCFile(ncname, T, domain=None, clobber=True, mode='r', **kw)

Bases: mpi4py_fft.utilities.file_base.FileBase

Class for writing data to NetCDF4 format

Parameters:

• ncname (str) – Name of netcdf file to be created

• T (PFFT) – Instance of a PFFT class. Must be the same as the space used for storing with NCWriter.write

• domain (Sequence) –

  The spatial domain. Sequence of either

  • 2-tuples, where each 2-tuple contains the (origin, length) of each dimension, e.g., (0, 2*pi).
  • Arrays of coordinates, e.g., np.linspace(0, 2*pi, N). One array per dimension.

• clobber (bool, optional)

• mode (str) – r or w for read or write. Default is ‘r’.

Note

Each class instance creates one unique NetCDF4-file, with one step-counter. It is possible to store multiple fields in each file, but all snapshots of the fields must be taken at the same time. If you want one field stored every 10th timestep and another every 20th timestep, then use two different class instances and as such two NetCDF4-files.

static backend()

open()

read(u, name, **kw)

Read into array u

Parameters:

• u (array) – The array to read into.
• name (str) – Name of array to be read.
• step (int, optional) – Index of field to be read. Default is 0.

write(step, fields, **kw)

Write snapshot step of fields to NetCDF4 file

Parameters:

• step (int) – Index of snapshot
• fields (dict) – The fields to be dumped to file. (key, value) pairs are group name and either arrays or 2-tuples, respectively. The arrays are complete arrays to be stored, whereas 2-tuples are arrays with associated global slices.

mpi4py_fft.utilities.file_base module

class mpi4py_fft.utilities.file_base.FileBase(T, domain=None, **kw)

Bases: object

Base class for reading/writing structured arrays

Parameters:

• T (PFFT) – Instance of a PFFT class.

• domain (sequence, optional) –

  The spatial domain. Sequence of either

  • 2-tuples, where each 2-tuple contains the (origin, length) of each dimension, e.g., (0, 2*pi).
  • Arrays of coordinates, e.g., np.linspace(0, 2*pi, N). One array per dimension.

static backend()

close()

open()

read(u, name, **kw)

Read into array u

Parameters:

• u (array) – The array to read into.
• name (str) – Name of array to be read.

write(step, fields, **kw)

Write snapshot step of fields to file

Parameters:

• step (int) – Index of snapshot.
• fields (dict) – The fields to be dumped to file. (key, value) pairs are group name and either arrays or 2-tuples, respectively. The arrays are complete arrays to be stored, whereas 2-tuples are arrays with associated global slices.

Module contents