HDF5 net description manipulation utilities.

class lava.lib.dl.netx.utils.NetDict(filename: Optional[str] = None, mode: str = 'r', f: Optional[Union[h5py._hl.files.File, h5py._hl.group.Group]] = None)

Provides dictionary like access to h5py object without the h5py quirks

  • filename (str or None, optional) – filename of h5py file to be loaded. It is only invoked if hdf5 file handle f is None. Default is None.

  • mode (str, optional) – file open mode, by default ‘r’.

  • f (h5py.File or h5py.Group, optional) – hdf5 file object handle. Overwrites the function of filename if it is not None. Default is None.

class lava.lib.dl.netx.utils.SYNAPSE_SIGN_MODE(value)

Enum for synapse sign mode. Options are {MIXED : 1, EXCITATORY : 2 and INHIBITORY : 2}.

lava.lib.dl.netx.utils.num_delay_bits(delays: numpy.ndarray) int

Calculates the number of delay bits required.


delays (np.ndarray) – delay vector


number of delay bits.

Return type


lava.lib.dl.netx.utils.optimize_weight_bits(weight: numpy.ndarray) Tuple[numpy.ndarray, int, int, lava.lib.dl.netx.utils.SYNAPSE_SIGN_MODE]

Optimizes the weight matrix to best fit in Loihi’s synapse.


weight (np.ndarray) – standard 8 bit signed weight matrix.


  • np.ndarray – optimized weight matrix

  • int – weight bits

  • int – weight_exponent

  • SYNAPSE_SIGN_MODE – synapse sign mode