Ideas behind
the loop.

Control, physical compute, and the cost of keeping hardware true.

Abstract gradient field

Hardware20266 min

96 qubits, nine hours: the contract holds native drift

Fifty of fifty cycles on ibm_kingston through the public cloud, decision SUPPORTED under the registered rule. Drifting qubits gained 25.7 points of usable output; residual held at 0.27x drift; downtime was zero.

Hardware20264 min

Third vendor, 156 qubits: the advantage widens as the problem grows

IBM Heron is the third commercial quantum processor to validate the loop live. The headline is 45.8% lower RMS error. The deeper result is the fixed-budget trend.

Evidence20264 min

Hardware validation: what two quantum processors settled

Two processors, one retained-versus-reset ablation, one 107-channel closed loop, and the boundary of the framework drawn from both sides.

Framework20264 min

Three axes of advantage: what wins, when, and where

A method can win on measurement efficiency, optimization quality, or regulation under drift. Confusing the axes hides the real result.

Methods20264 min

The starved-budget regime

When the budget is tight enough, the method, not the math, decides whether the loop can finish.

Open work20264 min

Reproduce it yourself

The useful artifact is not a chart. It is the script that makes the chart on a fresh problem, with disclosed seeds and baselines.

Measurement20264 min

When every measurement costs you

Some problems are slow because every objective evaluation is a physical event. The optimizer must be priced in measurements, not iterations.

Architecture20264 min

The configuration is the state

Retaining the physical configuration changes the shape of the loop and the economics of the next decision.

Knowing when not to use it

Applicability follows from structure. Performance is measured on the customer plant, never forecast.