An 80-qubit result is easy to make sound dramatic. The harder and more useful task is to say exactly what was run, what was measured, and where the claim stops.
This project executed a tracker-compatible Operator Loschmidt Echo, or OLE, experiment on the 156-qubit IBM Kingston processor through Q-CTRL Fire Opal. The active circuit used 80 qubits connected by 88 heavy-hex edges. It was derived from the released 70-qubit OLE circuit family and kept the released observable, perturbation strength, fixed core, and three-layer edge schedule.
The important word is compatible. This is an extension of a released tracker family, not a newly released official tracker instance.
The run in numbers
- Backend:
ibm_kingston - Active qubits: 80
- Active heavy-hex edges: 88
- Trotter parameter:
L = 3 - Perturbation:
delta = 0.15 - Observable:
Z52 Z59 Z72in the released declared indexing - Uniform basis samples:
N_init = 8 - Shots per circuit: 8,000
- Circuits: 8 perturbed plus 8 delta-zero controls
- Logical CZ gates per circuit: 1,056
Fire Opal compiled the perturbed circuits to a representative depth of 105 with 674 two-qubit gates. The delta-zero controls had depth 88 with 574 two-qubit gates. All 16 circuits were submitted together under Fire Opal action 2333919 and IBM Runtime job d98kkdkqp3as739t35mg.
The measured result
The eight sampled basis states gave a perturbed weighted mean of 0.32385603 +/- 0.01008261 and a delta-zero weighted mean of 0.43850987 +/- 0.01479688.
Their global ratio was
\[R_{\mathrm{OLE}}=0.74028847\pm0.01663657,
\qquad 95\%\ \mathrm{interval}=[0.70768080,0.77289615].
\]
Every individual sample ratio was positive, with values from 0.68759 to 0.83013.
That is a clean hardware result. It is not yet the same thing as a fully converged estimate over 500 initial states. The statistical error should decrease roughly as 1/sqrt(N_init), but hardware drift and batch-to-batch systematics do not have to follow that simple law.
What this is not
This experiment is not full quantum-state tomography. It does not reconstruct a vector with 2^80 amplitudes, and it does not invert a correlated 2^80 readout matrix. All 80 qubits were measured, but the reported OLE term uses the three-qubit ZZZ marginal selected by the tracker observable.
It is also not a black-hole simulation. The later connection to black holes concerns information scrambling and operator growth, not curved spacetime, an event horizon, or Hawking radiation generated inside the processor.
Why the result matters
The achievement is narrower and stronger than a vague 80-qubit headline. A precise observable was defined before execution. The exact submitted QASM, sample seed, raw payload, job identifiers, analysis, and classical benchmark are archived together in the associated version-controlled project.
That makes the result inspectable. The next article explains the OLE quantity itself and why a forward evolution, a perturbation, and an echo can reveal how information spreads.