arxiv: 2604.17369 · v1 · submitted 2026-04-19 · 🪐 quant-ph · cs.IT· math-ph· math.IT· math.MP

Quantum channel tomography: optimal bounds and a Heisenberg-to-classical phase transition

Kean Chen , Filippo Girardi , Aadil Oufkir , Nengkun Yu , Zhicheng Zhang This is my paper

Pith reviewed 2026-05-10 06:04 UTC · model grok-4.3

classification 🪐 quant-ph cs.ITmath-phmath.ITmath.MP

keywords quantum channel tomographyquery complexityHeisenberg scalingdilation ratephase transitiondiamond normChoi norm

0 comments p. Extension

The pith

The optimal query complexity for quantum channel tomography exhibits a sharp transition from Heisenberg scaling 1/ε to classical scaling 1/ε² at dilation rate τ = r d₂ / d₁.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper determines how many uses of an unknown quantum channel are required to reconstruct its full description to accuracy ε. The authors introduce the dilation rate τ = r d₂ / d₁, which is always at least 1 for trace-preserving channels, and show that query complexity follows different scaling laws in three regimes of this parameter. When τ exactly equals 1 the complexity scales linearly with 1/ε; when τ is bounded away from 1 it scales quadratically as 1/ε²; near the boundary the scaling interpolates between the two. These results give a precise characterization of the resources needed for quantum process tomography under standard error measures.

Core claim

For a quantum channel with input dimension d₁, output dimension d₂ and Kraus rank at most r, the minimal number of queries needed to achieve error ε in the Choi trace norm is Θ(r d₁ d₂ / ε) when τ = 1 and Θ(r d₁ d₂ / ε²) when τ ≥ 1 + Ω(1). The same thresholds hold for diamond-norm error except that the upper bound at τ = 1 improves to O(min{r d₁^{1.5} d₂ / ε, r d₁ d₂ / ε²}). In the near-boundary regime 1 < τ < 1 + o(1) the complexity mixes the two scalings.

What carries the argument

The dilation rate τ = r d₂ / d₁, which partitions the space of channels into regimes that obey Heisenberg versus classical query-complexity scaling.

Load-bearing premise

The unknown channel is trace-preserving and has Kraus rank at most r, with error measured in the Choi trace or diamond norm under ideal quantum information models.

What would settle it

An explicit construction or numerical experiment showing that a channel with τ = 1 requires Ω(r d₁ d₂ / ε²) queries in the Choi norm would falsify the claimed Heisenberg scaling.

Figures

Figures reproduced from arXiv: 2604.17369 by Aadil Oufkir, Filippo Girardi, Kean Chen, Nengkun Yu, Zhicheng Zhang.

**Figure 2.** Figure 2: A pictorial representation of our constructions of channel packing nets. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗

**Figure 3.** Figure 3: The combination of a 4-comb X with a 3-comb Y , yielding a 1-comb X ⋆ Y on (H0, H7). Definition 2.6 (Probabilistic comb [CDP09]). Let n ≥ 1 be an integer. A probabilistic n-comb, defined on a sequence of 2n Hilbert spaces (H0, H1, . . . , H2n−1), is a positive semidefinite operator X on N2n−1 j=0 Hj such that X ≤ Y for some deterministic n-comb Y on (H0, H1, . . . , H2n−1). Remark 2.7. In this paper, quant… view at source ↗

**Figure 4.** Figure 4: Illustration of our construction. We define linear operators [PITH_FULL_IMAGE:figures/full_fig_p038_4.png] view at source ↗

**Figure 5.** Figure 5: Illustration of our construction. We define linear operators [PITH_FULL_IMAGE:figures/full_fig_p045_5.png] view at source ↗

**Figure 6.** Figure 6: Illustration of our construction. There are two cases depending on whether Case 1 : [PITH_FULL_IMAGE:figures/full_fig_p052_6.png] view at source ↗

**Figure 7.** Figure 7: Illustration of our construction. We define linear operators [PITH_FULL_IMAGE:figures/full_fig_p059_7.png] view at source ↗

read the original abstract

How many black-box queries to a quantum channel are needed to learn its full classical description? This question lies at the heart of quantum channel tomography (also known as quantum process tomography), a fundamental task in the characterization and validation of quantum hardware. Despite extensive prior work, the optimal query complexity for quantum channel tomography is far from fully understood. In this paper, we study tomography of an unknown quantum channel with input dimension $d_1$, output dimension $d_2$, and Kraus rank at most $r$, to within error $\varepsilon$. We identify the dilation rate $\tau = r d_2 / d_1$ (which always satisfies $\tau\geq 1$ due to the trace preservation of quantum channels) as a key parameter, and establish that the optimal query complexity of channel tomography exhibits distinct scaling laws across three regimes of $\tau$. - In the boundary regime ($\tau = 1$): we show that the query complexity is $\Theta(r d_1 d_2/\varepsilon)$ for Choi trace norm error $\varepsilon$, and is upper bounded by $O(\min\{r d_1^{1.5} d_2/\varepsilon, r d_1 d_2/\varepsilon^2\})$ and lower bounded by $\Omega(r d_1 d_2/\varepsilon)$ for diamond norm error $\varepsilon$. - In the away-from-boundary regime ($\tau \geq 1+\Omega(1)$): we show that the query complexity is $\Theta(r d_1 d_2/\varepsilon^2)$ for both Choi trace norm and diamond norm errors $\varepsilon$. Our results uncover a sharp Heisenberg-to-classical phase transition in the query complexity of quantum channel tomography: at $\tau=1$, the optimal query complexity exhibits Heisenberg scaling $1/\varepsilon$, whereas for $\tau\geq 1+\Omega(1)$, it exhibits classical scaling $1/\varepsilon^2$. In addition, we show that in the near-boundary regime ($1< \tau < 1+o(1)$), the query complexity exhibits a mixture of Heisenberg and classical scaling behaviors.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper organizes channel tomography query complexity around the dilation rate τ with a clear phase transition, though diamond-norm bounds at τ=1 have a sqrt(d1) gap.

read the letter

The one thing to know is that the authors parameterize the query complexity of quantum channel tomography using the dilation rate τ = r d₂ / d₁, and they show it undergoes a Heisenberg-to-classical transition depending on whether τ sits at 1 or is larger. What the paper does well is organize the problem into three regimes and give explicit scaling laws. For the Choi trace norm, they achieve tight Θ(r d1 d2 / ε) at τ=1 and Θ(r d1 d2 / ε²) when τ ≥ 1 + Ω(1). The away-from-boundary case also matches for diamond norm. This is a useful way to think about when you get quantum advantage in learning channels. The soft spots are concentrated at the boundary for diamond norm error. They have a lower bound of Ω(r d1 d2 / ε) but only an upper bound of O(r d1^{1.5} d2 / ε), so the constants don't match and there's a dimension-dependent gap. This is not minor because the claim of Heisenberg scaling 1/ε for the transition relies on both norms behaving similarly, and diamond norm is the more operationally relevant one. The near-boundary regime is described as mixed but without precise exponents. The work is for quantum information theorists focused on process tomography and its sample complexity. Anyone trying to understand optimal query bounds in this area will get something from the τ parameter and the phase transition result. The citation pattern seems standard, and the math is presented as derived bounds rather than assumptions. I would send this to peer review. It addresses an incompletely understood question with a new organizing idea, so it merits referee time even if revisions are needed to close the gap in the diamond-norm bounds.

Referee Report

1 major / 0 minor

Summary. The paper studies the query complexity of quantum channel tomography for an unknown trace-preserving channel with input dimension d1, output dimension d2, and Kraus rank at most r, to error ε in either Choi trace norm or diamond norm. It introduces the dilation rate τ = r d2 / d1 (always ≥1) and proves that the optimal query complexity exhibits a phase transition: Θ(r d1 d2 / ε) (Heisenberg scaling) at τ=1 for Choi norm, with diamond-norm bounds Ω(r d1 d2 / ε) lower and O(min{r d1^{1.5} d2 / ε, r d1 d2 / ε²}) upper at τ=1; Θ(r d1 d2 / ε²) (classical scaling) for both norms when τ ≥ 1 + Ω(1); and mixed scaling in the near-boundary regime 1 < τ < 1 + o(1).

Significance. If the stated bounds hold and can be tightened where gaps exist, the work provides a sharp, parameter-dependent characterization of resources for quantum process tomography, identifying a clean Heisenberg-to-classical transition governed by τ. This is a substantive advance over prior dimension-dependent bounds, with potential implications for hardware characterization. The explicit separation into three τ regimes and the matching lower/upper bounds in the Choi-norm and away-from-boundary cases are particularly valuable.

major comments (1)

[Abstract] Abstract (diamond-norm case at τ=1): the lower bound Ω(r d1 d2 / ε) and upper bound O(r d1^{1.5} d2 / ε) (active for ε ≪ 1/√d1) differ by a √d1 factor. This gap is load-bearing for the claim of optimal Heisenberg scaling 1/ε and for the asserted sharpness of the phase transition under diamond norm, since the upper bound does not match the lower bound up to dimension-independent constants.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading, positive assessment of the significance of our results, and for identifying the gap in the diamond-norm bounds at τ=1. We address this point below and will make corresponding revisions.

read point-by-point responses

Referee: [Abstract] Abstract (diamond-norm case at τ=1): the lower bound Ω(r d1 d2 / ε) and upper bound O(r d1^{1.5} d2 / ε) (active for ε ≪ 1/√d1) differ by a √d1 factor. This gap is load-bearing for the claim of optimal Heisenberg scaling 1/ε and for the asserted sharpness of the phase transition under diamond norm, since the upper bound does not match the lower bound up to dimension-independent constants.

Authors: We agree that a √d1 gap exists between the Ω(r d1 d2 / ε) lower bound and the O(r d1^{1.5} d2 / ε) upper bound (active for small ε) for diamond-norm error at τ=1. Note that the abstract claims matching Θ(r d1 d2 / ε) only for Choi trace norm at τ=1; for diamond norm we state the bounds without asserting tightness up to dimension-independent factors. The gap originates from our upper-bound construction, which relies on a reduction to state tomography on a dilated space and incurs the extra √d1 factor in the worst case. The lower bound is information-theoretic and shows that Heisenberg scaling is necessary at the boundary. This does not undermine the phase transition: the lower bound already establishes a change from 1/ε at τ=1 to 1/ε² for τ ≥ 1 + Ω(1), where we obtain matching upper and lower bounds for both norms. We will revise the abstract and introduction to explicitly note the remaining gap for diamond norm at τ=1, clarify that optimality is claimed only where bounds match, and emphasize that the transition is witnessed by the lower bound at τ=1 versus the classical scaling away from the boundary. Closing the gap may require new algorithmic techniques and is left as future work. revision: partial

Circularity Check

0 steps flagged

Derivation self-contained; no circular reductions identified

full rationale

The paper defines the dilation rate τ = r d2 / d1 as a parameter (with τ ≥ 1 from trace preservation) and derives query complexity bounds Θ(r d1 d2 / ε) or O(min{...}) in different regimes of τ. These are presented as analysis results from standard quantum information models, not as redefinitions or statistical fits of the inputs. No load-bearing self-citations, ansatzes smuggled via prior work, or uniqueness theorems from the same authors appear in the claims. The phase transition follows from the distinct scaling behaviors established by the bounds, without tautological equivalence. The noted gap between lower and upper bounds for diamond norm at τ=1 is a sharpness issue, not a circularity reduction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based solely on the abstract, no explicit free parameters, axioms, or invented entities are detailed; the work relies on standard quantum channel assumptions such as trace preservation and Kraus rank bounds.

pith-pipeline@v0.9.0 · 5720 in / 1223 out tokens · 37322 ms · 2026-05-10T06:04:12.484535+00:00 · methodology

discussion (0)

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Chaos Emerge with Exceptional Points in Reset-Driven Floquet Dynamics
quant-ph 2026-05 unverdicted novelty 7.0

Tuning a chaos parameter drives an exceptional-point transition in reset-driven Floquet channel spectra from real eigenvalues in an ergodic regime to complex pairs in a chaotic regime, distinguishing multiple dynamica...

Reference graph

Works this paper leans on

26 extracted references · 23 canonical work pages · cited by 1 Pith paper · 3 internal anchors

[1]

Anshu and S

[AA23] Anurag Anshu and Srinivasan Arunachalam. A survey on the complexity of learning quantum states (2023).arXiv preprint arXiv:2305.20069,

work page arXiv 2023
[2]

A quantum speed-up for approxi- mating the top eigenvectors of a matrix

[CGdW25] Yanlin Chen, Andr´ as Gily´ en, and Ronald de Wolf. A quantum speed-up for approxi- mating the top eigenvectors of a matrix. InProceedings of the 2025 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA), pages 994–1036. SIAM,

2025
[3]

Adaptivity can help exponentially for shadow tomography.arXiv preprint arXiv:2412.19022,

[CGZ24] Sitan Chen, Weiyuan Gong, and Zhihan Zhang. Adaptivity can help exponentially for shadow tomography.arXiv preprint arXiv:2412.19022,

work page arXiv
[4]

Inverse-free quantum state estimation with heisenberg scaling.arXiv preprint arXiv:2510.25750,

[Che25] Kean Chen. Inverse-free quantum state estimation with heisenberg scaling.arXiv preprint arXiv:2510.25750,

work page arXiv
[5]

Local test for unitarily invariant properties of bipartite quantum states

[CWZ24] Kean Chen, Qisheng Wang, and Zhicheng Zhang. Local test for unitarily invariant properties of bipartite quantum states.arXiv preprint arXiv:2404.04599,

work page arXiv
[6]

Quantum channel tomography and estimation by local test

[CYZ25] Kean Chen, Nengkun Yu, and Zhicheng Zhang. Quantum channel tomography and estimation by local test.arXiv preprint arXiv:2512.13614,

work page arXiv
[7]

Optimal lower bound for quantum channel tomography in away-from-boundary regime.arXiv preprint arXiv:2601.10683,

[CZY26] Kean Chen, Zhicheng Zhang, and Nengkun Yu. Optimal lower bound for quantum channel tomography in away-from-boundary regime.arXiv preprint arXiv:2601.10683,

work page arXiv
[8]

Efficient Learning of Structured Quantum Circuits via Pauli Dimensionality and Sparsity

[GL25] Sabee Grewal and Daniel Liang. Efficient learning of structured quantum circuits via Pauli dimensionality and sparsity.arXiv preprint arXiv:2510.00168,

work page internal anchor Pith review Pith/arXiv arXiv
[9]

Girardi, F

[GML25] Filippo Girardi, Francesco Anna Mele, and Ludovico Lami. Random purification chan- nel made simple.arXiv preprint arXiv:2511.23451,

work page arXiv
[10]

and Mele F

78 [GMZ+25] Filippo Girardi, Francesco Anna Mele, Haimeng Zhao, Marco Fanizza, and Ludovico Lami. Random Stinespring superchannel: converting channel queries into dilation isometry queries.arXiv preprint arXiv:2512.20599,

work page arXiv
[11]

The Church of the Symmetric Subspace

[Har13] Aram W Harrow. The church of the symmetric subspace.arXiv preprint arXiv:1308.6595,

work page Pith review arXiv
[12]

Optimal learning of quantum channels in diamond distance

[MB25] Antonio Anna Mele and Lennart Bittel. Optimal learning of quantum channels in diamond distance.arXiv preprint arXiv:2512.10214,

work page arXiv
[13]

Optimal Counterfeiting Attacks and Gen- eralizations for Wiesner’s Quantum Money

[MGC+25] Francesco Anna Mele, Filippo Girardi, Senrui Chen, Marco Fanizza, and Ludovico Lami. Random purification channel for passive gaussian bosons.arXiv preprint arXiv:2512.16878,

work page arXiv
[14]

Improved lower bounds for learning quantum chan- nels in diamond distance.arXiv preprint arXiv:2601.04180,

[OG26] Aadil Oufkir and Filippo Girardi. Improved lower bounds for learning quantum chan- nels in diamond distance.arXiv preprint arXiv:2601.04180,

work page arXiv
[15]

Sample-optimal quantum process tomography with non-adaptive inco- herent measurements

[Ouf23b] Aadil Oufkir. Sample-optimal quantum process tomography with non-adaptive inco- herent measurements. In2023 IEEE International Symposium on Information Theory (ISIT), page 1919–1924. IEEE, June

1919
[16]

In:arXiv:2511.15806 [quant-ph](2025)

[PSTW25] Angelos Pelecanos, Jack Spilecki, Ewin Tang, and John Wright. Mixed state tomog- raphy reduces to pure state tomography.arXiv preprint arXiv:2511.15806,

work page arXiv
[17]

Pelecanos, J

[PSW25] Angelos Pelecanos, Jack Spilecki, and John Wright. The debiased keyl’s algorithm: a new unbiased estimator for full state tomography.arXiv preprint arXiv:2510.07788,

work page arXiv
[18]

Quantum channel testing in average-case distance.arXiv preprint arXiv:2409.12566, 2024

[RAS+24] Gregory Rosenthal, Hugo Aaronson, Sathyawageeswar Subramanian, Animesh Datta, and Tom Gur. Quantum channel testing in average-case distance.arXiv preprint arXiv:2409.12566,

work page arXiv
[19]

Online learning of quantum processes.arXiv preprint arXiv:2406.04250,

[RCEK24] Asad Raza, Matthias C Caro, Jens Eisert, and Sumeet Khatri. Online learning of quantum processes.arXiv preprint arXiv:2406.04250,

work page arXiv
[20]

Optimal lower bounds for quantum state tomography.arXiv preprint arXiv:2510.07699,

[SSW25] Thilo Scharnhorst, Jack Spilecki, and John Wright. Optimal lower bounds for quantum state tomography.arXiv preprint arXiv:2510.07699,

work page arXiv
[21]

Testing matrix product states

[SW22] Mehdi Soleimanifar and John Wright. Testing matrix product states. InProceedings of the 2022 Annual ACM-SIAM Symposium on Discrete Algorithms (SODA), pages 1679–1701. SIAM,

2022
[22]

Conjugate queries can help

[TWZ25] Ewin Tang, John Wright, and Mark Zhandry. Conjugate queries can help.arXiv preprint arXiv:2510.07622,

work page internal anchor Pith review arXiv
[23]

Walter and F

81 [WW25] Michael Walter and Freek Witteveen. A random purification channel for arbitrary sym- metries with applications to fermions and bosons.arXiv preprint arXiv:2512.15690,

work page arXiv
[24]

Quantum Advantage in Storage and Retrieval of Isometry Channels

[YMM25] Satoshi Yoshida, Jisho Miyazaki, and Mio Murao. Quantum advantage in storage and retrieval of isometry channels.arXiv preprint arXiv:2507.10784,

work page internal anchor Pith review Pith/arXiv arXiv
[25]

Yoshida, R

[YNM25] Satoshi Yoshida, Ryotaro Niwa, and Mio Murao. Random dilation superchannel.arXiv preprint arXiv:2512.21260,

work page arXiv
[26]

Zambrano, S

[ZRCK25] Leonardo Zambrano, Sergi Ramos-Calderer, and Richard Kueng. Fast quantum measurement tomography with dimension-optimal error bounds.arXiv preprint arXiv:2507.04500,

work page arXiv