cuquantum.OptimizerInfo¶

class cuquantum.OptimizerInfo(largest_intermediate: float, opt_cost: float, path: Iterable[Tuple[int, int]], slices: Iterable[Tuple[Hashable, int]], num_slices: int, intermediate_modes: Iterable[Iterable[Hashable]])[source]¶

A data class for capturing optimizer information.

largest_intermediate¶

The number of elements in the largest intermediate tensor. See CUTENSORNET_CONTRACTION_OPTIMIZER_INFO_LARGEST_TENSOR.

Type: float

opt_cost¶

The FLOP count of the optimized contraction path. See CUTENSORNET_CONTRACTION_OPTIMIZER_INFO_FLOP_COUNT.

Type: float

path¶

The contraction path as a sequence of pairs in the numpy.einsum_path() format.

Type: Iterable[Tuple[int, int]]

slices¶

A sequence of (sliced mode, sliced extent) pairs.

Type: Iterable[Tuple[Hashable, int]]

num_slices¶

The number of slices.

Type: int

intermediate_modes¶

A sequence of mode labels for all intermediate tensors

Type: Iterable[Iterable[Hashable]]