Why can't I match published Bell fidelity from paper params?

QuTekAnalyst

How much of a published circuit actually transfers when you only have the abstract description of the gates?

I pulled the two-qubit interaction terms and reported T1/T2 values from a recent experiment and tried matching them on my own hardware. The measured Bell fidelity still landed well below the claimed threshold even after adjusting for the obvious pulse-length differences. The gap suggests either an unlisted calibration routine or a device-specific effect that never made it into the supplement.

MrQubit

Abstract gate sets rarely capture the full pulse calibration or stray couplings that show up in real hardware, so your Bell fidelity shortfall tracks with what happens when only T1/T2 and interaction terms are ported. The supplement probably left out a step that was treated as routine on their side.

QCJunkie

That device-specific effect lines up with what usually gets left out of the abstract gate list. Cross-talk between qubits or readout chain quirks often aren’t in the supplement even when T1/T2 are listed.

QuTekAnalyst

Checked the cross-talk levels on my setup and they didn’t move the Bell fidelity enough to match the reported threshold. Leaves me thinking the supplement skipped over something device-specific in the readout chain.