mindchemistry.e3.o3.FullyConnectedTensorProduct

View Source On Gitee
class mindchemistry.e3.o3.FullyConnectedTensorProduct(irreps_in1, irreps_in2, irreps_out, ncon_dtype=float32, **kwargs)[source]

Fully-connected weighted tensor product. All the possible path allowed by \(|l_1 - l_2| \leq l_{out} \leq l_1 + l_2\) are made. Equivalent to TensorProduct with instructions='connect'. For details, see mindchemistry.e3.o3.TensorProduct.

Parameters

  • irreps_in1 (Union[str, Irrep, Irreps]) – Irreps for the first input.

  • irreps_in2 (Union[str, Irrep, Irreps]) – Irreps for the second input.

  • irreps_out (Union[str, Irrep, Irreps]) – Irreps for the output.

  • irrep_norm (str) – {'component', 'norm'}, the assumed normalization of the input and output representations. Default: 'component'. Default: 'component'.

  • path_norm (str) – {'element', 'path'}, the normalization method of path weights. Default: 'element'.

  • weight_init (str) – {'zeros', 'ones', 'truncatedNormal', 'normal', 'uniform', 'he_uniform', 'he_normal', 'xavier_uniform'}, the initial method of weights. Default: 'normal'.

  • ncon_dtype (mindspore.dtype) – The type of input tensors of ncon computation module. Default: mindspore.float32 .

Supported Platforms:

Ascend

Examples

>>> from mindchemistry.e3.o3 import FullyConnectedTensorProduct
>>> FullyConnectedTensorProduct('2x1o', '1x1o+3x0e', '5x2e+4x1o')
TensorProduct [connect] (2x1o x 1x1o+3x0e -> 5x2e+4x1o)