ARCADE Readout Zoo
Home Learn Classifiers Results Quickstart Docs FTQC and NISQ goals

Home · Results · Goals

System endings

From classifier metrics to the goal you optimize

The same frozen predictions support two stories. For fault tolerance: logical error versus readout duration, idle noise, and wall-clock cycle time. For NISQ: accuracy versus latency under a coherence budget.

Primary

Fault-tolerant QEC

Assignment error becomes physical readout error. Idle grows with acquisition plus post-measurement control. Wall-clock cycle = readout + control + decode.

Also in scope

NISQ operating point

Pick an accuracy–latency knee that fits your coherence budget. That point can later constrain readout-pulse design.

Logical error versus duration

LER sweeps · fixed physical noise
LER versus duration

No idle scaling: physical noise is held fixed, so LER tracks assignment error across the full acquire window (to 2000 ns).

Interactive LER

Cycle time

Readout + control + decode
Cycle timing

Wall-clock stack

NISQ · accuracy versus latency

Operating-point view
Accuracy versus latency

Same comparison, different ending: choose the knee your budget allows.

Accuracy vs readout error

Reference results Learn about errors Quickstart