How to exactly compute observable of 1 qubit in Braket TN1 and don't measure the rest?

Question

I am aware that exact simulation (on the local simulator) to get the state vector can be done by setting `shots=0`. However, TN1 only allows shots range of [1, 1000]. I tried to reduce the computation cost of my QFT circuit by doing
```python
qft = make_qft_circuit(10)
qft.expectation(observable=Observable.Z(), target=0)
result = device.run(qc, shots=100).result()
```
However, upon checking the result, I found that all the 10 qubits are being measured, with the counts stats, whereas I only need the expectation of the Pauli Z on the 0th qubit. Is there a way to prevent the wasteful computation of the remainder 9 qubits (since this is tensor network simulation after all)? Even better if it is possible to do an exact simulation along the line of `shots=0` for the statevector.

Accepted Answer

For the first question, despite all the qubits are measured, we can extract the expectation value for a subset of qubits via `result.values`. Suppose we want to measure the expectation of the Pauli Z on the 0th qubit, then we can do the following [In the given example, `qc` is not defined, and I changed that]

```
qc = make_qft_circuit(10)
qc.expectation(observable=Observable.Z(), target=0)
result = device.run(qc, shots=100).result()
Z_0 = result.values[0] # This is the expectation value desired.
```

For the second question, no, TN1 does not support `shots=0` result type.

How to exactly compute observable of 1 qubit in Braket TN1 and don't measure the rest?

관련 콘텐츠