Function _array_numpy_to_dask

pyxrf/core/map_processing.py:292–328

Convert an array (e.g. XRF map) from numpy array to chunked Dask array. Select chunk size based on the desired number of pixels `chunk_pixels`. The array is considered as an image with pixels along axes 0 and 1. The array is chunked only along axes 0 and 1. Parameters ---------

(data, chunk_pixels, n_chunks_min=4)

def _array_numpy_to_dask(data, chunk_pixels, n_chunks_min=4):
    """
    Convert an array (e.g. XRF map) from numpy array to chunked Dask array. Select chunk
    size based on the desired number of pixels `chunk_pixels`. The array is considered
    as an image with pixels along axes 0 and 1. The array is chunked only along axes 0 and 1.

    Parameters
    ----------
    data: ndarray(float), 3D
        Numpy array of the shape `(ny, nx, ...)` with at least 2 dimensions. If `data` is
        an image, then `ny` and `nx` represent the image dimensions.
    chunk_pixels: int
        Desired number of pixels in a chunk. The actual number of pixels may differ from
        the desired number to accommodate minimum requirements on the number of chunks or
        limited size of the dataset.
    n_chunks_min: int
        minimum number of chunks, which should be selected based on the minimum number of
        workers that should be used to process the map. Each chunk will contain at least
        one pixel: if there is not enough pixels, then the number of chunks will be reduced.

    Results
    -------
    Dask array of the same shape as `data` with chunks selected based on the desired number
    of pixels `chunk_pixels`.
    """

    if not isinstance(data, np.ndarray) or (data.ndim < 2):
        raise ValueError(f"Parameter 'data' must numpy array with at least 2 dimensions: type(data)={type(data)}")

    ny, nx = data.shape[0:2]
    # Since numpy array is not chunked by default, set the original chunk size to (1,1)
    #   because here we are performing 'original' chunking
    chunk_y, chunk_x = _compute_optimal_chunk_size(
        chunk_pixels=chunk_pixels, data_chunksize=(1, 1), data_shape=(ny, nx), n_chunks_min=n_chunks_min
    )

    return _chunk_numpy_array(data, (chunk_y, chunk_x))


def prepare_xrf_map(data, chunk_pixels=5000, n_chunks_min=4):