dvnoble
commented
4 months ago There is the qc translate command to make sure that the result circuit obeys the gateset.
Edit: qc translate has only three options which I don't know if they represent the gateset in the layout. Here is the layout I used (note that I've removed the h from gateset):
NAME: perth_7 QUBITNUM: 7 GATESET: {x, rz, id, sx, cnot} COUPLINGMAP: [[1], [0, 2, 3], [1], [1, 5], [5], [3, 4, 6], [5]] SGERROR: [0, 0, 0, 0, 0, 0, 0] SGTIME: [0, 0, 0, 0, 0, 0, 0] CNOTERROR: [[1], [0, 2, 3], [1], [1, 5], [5], [3, 4, 6], [5]] CNOTTIME: [[1], [0, 2, 3], [1], [1, 5], [5], [3, 4, 6], [5]]
Greetings!
The Perth backennd in Qiskit has no H gate. You can check in https://github.com/Qiskit/qiskit-ibm-runtime/blob/stable/0.24/qiskit_ibm_runtime/fake_provider/backends/perth/conf_perth.json. Search for "basis_gates": ["cx", "id", "rz", "sx", "x"] When I use the Perth layout according to the Qiskit specification (without H gates in the layout file) and I use the following dof script file:
qcir read benchmark/qft/qft_7.qasm qcir print --stat qc2zx zx print --stat zx copy zx optimize --full zx print --stat zx2qc qcir print --stat qcir optimize qcir print --stat
device read benchmark/topology/perth_7.layout duostra -c qcir print --stat
qcir list
qcir write examples/qft_7-opt.qasm
quit -f
The resulting circuit still includes H gates which I think is not the expected. I am using qsyn version 0.7.0.