Qiskit / qiskit / 27292754037

88%

Build Type

push

github

Committed by web-flow

Commit Message

Add QuantumCircuit method to compute estimated fidelity (#16292)

* Add QuantumCircuit method to compute estimated fidelity

This commit adds a new method to the quantum circuit (in both Python and
C) to easily compute the estimated fidelity of a physical circuit for a
given target. This uses the fidelity estimation that Qiskit's internal passes
such as VF2Layout, VF2PostLayout, and TwoQubitPeepholeOptimization use
for estimating fidelity of a circuit or subcircuit. This is a common operation
when working with circuit compilation, especially when trying different
pass manager constructions, or pass configurations to try and optimize the circuit for
execution on a given backend. Having a quick metric for comparison of how the
transpiler reasons about different compilations of the same circuit is
useful for debugging compilation workflows.

While writing this function is fairly straightforward to do in Python
with a simple for loop like:

```python
error = 1.0
for inst in circuit.data:
    qargs = tuple(circuit.find_bits(x).index for x in inst.qubits)
    error *= target[inst.name][qargs].error
1.0 - error
```
(it's a bit more involved in C but fairly trivial too) given the general
utility of this and avoiding Python iteration for large circuits writing this
in Rust and exposing it to both C and Python for a consistent method is worth
the addition.

The rust implementation of this new function is put in
`qiskit_transpiler::target::estimate_fidelity` mostly because that was
the lowest part of the internal dependency tree that it could be
implemented. Typically this kind of function we would put in the
`qiskit_accelerate` crate as a random accelerator, however because this
function is being exposed to C too that was not a viable destination for
it. Since we need both `CircuitData` and `Target` available to implement
the function this made the most sense. The other alternative is probably
to create a standalone module in the transpiler crate but this funct... (continued)

Coverage Stats

41 of 41 new or added lines in 4 files covered. (100.0%)

23 existing lines in 4 files now uncovered.

110866 of 126646 relevant lines covered (87.54%)

955306.06 hits per line