Ideal versus noisy circuit comparison

The same Bell circuit and seed run through the real N/M Worker runtime. Identity gate pairs increase physical gate count while preserving the ideal state, making accumulated model behavior visible.

Preview labComparison ready

This is a seeded, finite-shot local simulator estimate. It is not a hardware run, device calibration, decoder benchmark, threshold result, or fault-tolerance evidence.

Noise model

Depth-scaled global basis replacement used for a compact aggregate model.

Channel probability5%

Bounded to 0–25% in this lab; the runtime annotation itself remains visible in the generated source.

Depth multiplier3

Each extra level adds two H and two CNOT identity gates, preserving the ideal Bell state.

Shots

Seed

The same unsigned seed makes comparisons reproducible; it does not remove stochastic uncertainty.

Apply compatible mitigationReadout inversion for readout noise; bounded two-point ZNE for other nonzero models.

2 qubits · depth 1–8 · 128–4096 shots · probability 0–25%

Backend: Local statevector
Executed gates per shot: 10
Noisy Bell-support mass: 79.9%
Ideal/noisy TV distance: 20.1%

Outcome distribution

Every outcome shows ideal, noisy, and—when enabled—mitigated probability. Exact percentages remain available in the table.

IdealNoisyMitigated

|00>

Ideal48.4%

Noisy37.9%

Mitigated41.2%

|01>

Ideal0.0%

Noisy9.2%

Mitigated2.9%

|10>

Ideal0.0%

Noisy10.9%

Mitigated7.4%

|11>

Ideal51.6%

Noisy42.0%

Mitigated48.4%

Exact ideal, noisy, and mitigated probabilities for each two-qubit basis state.
State	Ideal	Noisy	Mitigated
\|00>	48.4%	37.9%	41.2%
\|01>	0.0%	9.2%	2.9%
\|10>	0.0%	10.9%	7.4%
\|11>	51.6%	42.0%	48.4%

Execution semantics

Channel placement: Depth-scaled global basis replacement
Trajectory simulation: no
Apply compatible mitigation: Two-point zero-noise extrapolation
Shots: 512
Seed: 42

Evidence boundary

The histogram is a finite-shot sample, not an exhaustive probability proof.
The provider is the local N/M simulator; no QPU executed this circuit.
The selected probability is a model input, not measured device calibration.
This comparison does not estimate a QEC threshold, logical error rate, or fault tolerance.

Inspect generated N/M source

module noise_lab_depolarizing;
@seed(42);
@noise("depolarizing", 0.05);
@mitigation("zne");

fn main() {
    let q: QReg<2> = qreg[2];

    sample 512 {
        H(q[0]);
        CNOT(q[0], q[1]);
        H(q[0]);
        H(q[0]);
        CNOT(q[0], q[1]);
        CNOT(q[0], q[1]);
        H(q[0]);
        H(q[0]);
        CNOT(q[0], q[1]);
        CNOT(q[0], q[1]);
        let a: Bit = measure(q[0]);
        let b: Bit = measure(q[1]);
        return (a, b);
    }
}

Saved Noise Lab runs

Keep reproducible comparisons in this browser and sync the same verified artifact to your account when Supabase is available.

Checking cloud

Loading saved runs…

Up to 50 integrity-checked runs are kept locally. Signed-in cloud history is account-scoped, stores no password or provider token, and supports user-controlled deletion.

Open this run in Playground Compose scoped and correlated noise