Use cases · Aerospace · sovereign programmes · 2026-11

Characterise the quantum machine you already own

From a processor’s own calibration data we read its clock to 0.0029%, locate the qubit that is losing information, and route around it at submission. A national programme learns what its installed machine can run this quarter, measured on the machine itself.

Impact

Your machine’s clock, read to three parts in a hundred thousand
Baseline: the vendor calibration sheet
Condition: 0.0029% from the device’s own response, eight time rounds, superconducting processor
The weak qubit found and avoided
Baseline: average error rates that hide it
Condition: the lossy qubit located and routed around at submission
Full-size runs on the machine you have
Baseline: waiting for a hundred-qubit method
Condition: 42 working units on a 256-atom machine with coupling set by placement; 26 working units carrying 1,448 entangling operations on a 156-qubit chip

The problem

Programmes that own a small quantum machine wait for methods that need a hundred qubits, while the machine sits idle and its vendor figures go unchecked. The question is what the installed device can run now.

Our approach

We read the device as a physical system: its clock from the frequencies in its own response, its loss from where information leaves during each operation. The read is made on the machine, the lossy qubit is named, and the first circuits are placed around it.

The evidence

  • Clock read to 0.0029% on a superconducting processor from its own response; the lossy qubit identified.
  • Placement on a 256-atom neutral-atom machine: 42 working units, coupling set by atom placement.
  • Full-processor runs: 26 working units carrying 1,448 entangling operations on a 156-qubit processor.
  • Not claimed: that the vendor’s figures are wrong, or any application result from the characterisation alone.

Status: Quantum hardware. Measured on quantum hardware; job identifiers on file for every figure.