state_projection_mps_compute_tensor_env#

cuquantum.bindings.cutensornet.state_projection_mps_compute_tensor_env(
intptr_t handle,
intptr_t tensor_network_projection,
intptr_t env_spec,
strides_in,
intptr_t env_tensor_data_in,
strides_out,
intptr_t env_tensor_data_out,
int32_t apply_inv_metric,
int32_t re_resolve_channels,
intptr_t work_desc,
intptr_t cuda_stream,
)[source]#

Computes the projection for the specified environment.

Parameters:

  • handle (intptr_t) – cuTensorNet library handle.

  • tensor_network_projection (intptr_t) – Tensor network state MPS projection.

  • env_spec (intptr_t) – Specification of the requested environment. Note that currently only single site environments are supported.

  • strides_in (object) –

    Strides of the provided MPS representation tensor for the specified environment. Required to be a nullptr if MPS projection is symmetric. It can be:

    • an int as the pointer address to the array, or

    • a Python sequence of int64_t.

  • env_tensor_data_in (intptr_t) – Optional input value (GPU-accessible pointer) for tensor to replace part of the initial state in the environment. The extents of the tensor are indetical to the output tensor extents and it will replace the MPS tensors in between the lower and upper bounds of specified environment. Required to be a nullptr if MPS projection is not symmetric.

  • strides_out (object) –

    Strides of the output tensor environment for the specified environment. It can be:

    • an int as the pointer address to the array, or

    • a Python sequence of int64_t.

  • env_tensor_data_out (intptr_t) – Computed tensor environment for the specified contiguous subset of sites (GPU-accessible pointer). The provided buffer will be conjugated on-the-fly during the environment contraction.

  • apply_inv_metric (int32_t) – Whether or not to apply the inverse metric of the MPS state with respect to the environment to the computed tensor environment. Currently application of the inverse metric is not supported.

  • re_resolve_channels (int32_t) – Whether or not to reresolve the channels of the computed tensor environment on compute call. If true, the behaviour aligns with that of other State API properties computed on a State instance for which StateCompute has not yet been invoked. If false, the behaviour aligns with that of other State API properties computed on a State instance for which StateCompute has already been invoked.

  • work_desc (intptr_t) – Allocated workspace descriptor.

  • cuda_stream (intptr_t) – CUDA stream.

See also

cutensornetStateProjectionMPSComputeTensorEnv()