Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsA new arXiv preprint reports that amplitude damping, a common model of energy loss in qubits, stayed ahead of its Pauli-twirled counterpart by at most about three percentage points in accuracy once a trainable output scale was used. That is a simulation result from a version-1 preprint. It is not a hardware demonstration or a general guarantee that quantum noise is capped at three points (Nguyen, Truong, Nguyen and Le, arXiv preprint, 2026; abstract).
What the paper actually compared
The paper is titled “Zero- Versus Infinite-Temperature Damping in Variational Quantum Circuits: Feature Scale, Sampling Cost, and Frame Gauge.” The abstract was submitted on 1 October 2026, and the manuscript is dated 2 October 2026. As of 5 October 2026 it was listed as version 1 (arXiv record, manuscript PDF).
As an Amazon Associate I earn from qualifying purchases.
The authors compare two noise models in variational circuits:
- Amplitude damping (AD): a model of energy relaxation, often called T1 decay. It shrinks the Bloch-vector components and also shifts them, which is its non-unital bias.
- The Pauli twirl of AD: it keeps the contraction and removes the non-unital term. The authors describe it as generalized amplitude damping at infinite temperature.
Comparing the two isolates the effect of AD’s zero-temperature bias. In the authors’ words: “The Pauli twirl of amplitude damping (AD) is generalized amplitude damping at infinite temperature: it keeps the contraction of AD and removes its non-unital term, so comparing the two in variational circuits isolates the zero-temperature bias, which acts mainly through the scale of the features.” This is a comparison between two noise models. It is not a proposal that real devices can choose the temperature of their noise.
#1 Best Overall
The numbers, with their conditions
| Figure | What it refers to | Source |
|---|---|---|
| About 3 percentage points | The most that AD stayed ahead of its Pauli twirls once a trainable output scale was used, in the authors’ simulations | Nguyen et al., arXiv preprint, 2026 (abstract) |
| Within 1.5 percentage points | Gap to noiseless accuracy for a four-qubit classifier under AD at p=0.3, with 1,000 measurement shots per image in training and testing | Nguyen et al., arXiv preprint, 2026 |
| Up to 33 percentage points | Accuracy lost by the twirled classifiers against noiseless conditions in the same comparison | Nguyen et al., arXiv preprint, 2026 |
| Up to eight qubits | Widths over which feature behavior is reported. The four-qubit classifier result is separate. | Nguyen et al., arXiv preprint, 2026 (manuscript) |
The “three points” is therefore a gap between two noise models. It is not a cap on how much accuracy noise can cost. In the headline classifier example, AD stayed close to noiseless accuracy while the twirled versions fell much further.
Why a bias in the noise can help
The authors argue that AD’s non-unital bias acts mainly by changing the scale of the features. A trainable output scale can compensate for much of that change, which is why it removes most of the accuracy differences. That compensation has a price in measurement resources. Similar fitted accuracy does not mean equal shot cost, so the central result links noise model, feature scale, accuracy and sampling (manuscript).
Rank #2
Where the noise happens matters
The damping direction can behave as a gauge, meaning a change of representation. This holds when damping follows complete entangling layers and trainable circuit boundaries can absorb the change. In the authors’ eigensolver setup, damping inside a decomposed two-qubit gate makes the direction physically relevant. The manuscript says the effect vanishes when the same damping follows the gate. Circuit layout is therefore part of the result, not a side detail.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →On depth, the authors report that at weaker damping the separation depth grows roughly as (np)−1 ln(1/p), for n qubits and damping strength p. They say that at damping levels relevant to current hardware, this separation can concern very deep circuits.
What was tested
- Single-qubit re-uploading fits.
- A four-qubit re-uploading classifier on MNIST and Fashion-MNIST.
- A three-qubit eigensolver.
Limits on how far to trust the headline
- Simulation only. Nothing here shows an improved quantum processor or a hardware guarantee.
- Narrow setup. The classifier is four qubits, on specific datasets, at a specific shot count. Its accuracy numbers should not be carried over to other circuits, datasets, qubit counts or noise conditions.
- Exact versus finite-shot. The manuscript reports that some gains seen in exact simulation do not survive finite-shot training and testing.
- Conditions matter. The results depend on trainable output scaling, damping probability, circuit depth and where damping sits relative to gates.
- Unreviewed. The arXiv record establishes the claims and setup. It does not establish peer review or independent replication.
Secondary coverage is available from Quantum Zeitgeist (4 October 2026). For background on the noise models, the paper cites Nielsen and Chuang’s Quantum Computation and Quantum Information.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.The Bottom Line
Read the three-point figure as a finding about noise-model structure in simulated variational circuits. It is not a hardware milestone. With a trainable output scale, amplitude damping’s bias costs little accuracy, but the cost shifts to measurement shots. Independent replication would be needed before treating it as settled.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
