Simulating noisy cirq circuits

[lucas-slattery]

Is there a way to use the cuda-q framework to directly simulate noisy cirq circuits or to translate noisy cirq circuits into cudaq kernels?

If I wanted to simulate a noisy cirq circuit using the cuquantum 'NetworkState' using the tensor network simulator. Would I need to:

-sample a trajectory for a noisy circuit
-compute and save a contraction path
-for each trajectory
update the network to the new trajectory
sample a measurement outcome/observable/reduced density matrix using the saved contraction path
-combine results

or would the NetowrkState.sample() handle the trajectories.

[1tnguyen]

Is there a way to use the cuda-q framework to directly simulate noisy cirq circuits or to translate noisy cirq circuits into cudaq kernels?

In general, I don't think that there is a way to directly simulate a noisy Cirq circuit with CUDA-Q.

Cirq noisy simulation can be translated to CUDA-Q using the apply_noise method (see the doc here)

For example,

This Cirq snippet from its doc:

depo = cirq.asymmetric_depolarize(p_x=0.10, p_y=0.05, p_z=0.15)

circuit = cirq.Circuit(depo(cirq.LineQubit(0)))

can be translated to CUDA-Q as:

cudaq.apply_noise(cudaq.Pauli1, 0.10, 0.05, 0.15, q[0])

Regarding the second question about Cirq - NetworkState interoperability, I'd recommend that you post it to the cuQuantum GitHub repository.