e3j.utils.options.TensorProduct

class e3j.utils.options.TensorProduct

Evaluation strategy for TensorProduct.

Values:

• UNFUSED – Pull-back inputs by CG indices, multiply, then aggregate.

• DENSE – Evaluate via einsum on dense CG coefficient array.

• FUSED_CUDA – Dispatch to custom CUDA kernel via XLA-FFI (requires e3j_ops).

• FUSED_MOSAIC_TPU – Dispatch to the Pallas Mosaic TPU kernel.

• FUSED_MOSAIC_GPU – Dispatch to the Pallas Mosaic GPU kernel, TRAILING_CHANNELS layout, OUTER and MAP modes. Targets Hopper (sm_90) and newer.

Methods

__getitem__(name)	Return the member matching name.
__iter__()	Return members in definition order.
__len__()	Return the number of members (no aliases)

Attributes

UNFUSED
DENSE
FUSED_CUDA
FUSED_MOSAIC_TPU
FUSED_MOSAIC_GPU