iSWAP maximises the second-moment spectral gap in random quantum circuits

Haoran Zhu; Yanying Liang

REVIEW 1 major objections 4 minor 25 references

iSWAP maximises the second-moment spectral gap in random quantum circuits

T0 review · 1 major / 4 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper proves that the iSWAP gate maximises the second-moment spectral gap among all Hermitian two-local circuit ensembles on every connected graph with at least three vertices, making it the fastest such circuit for forming unitary 2-d

desk verdict Resolves a real conjecture with a genuinely new invariant-cone argument; the core comparison is solid, but the final step leans on an imported localization result that a referee must check. read the letter →

arxiv 2607.29551 v1 pith:7PY6OXBV submitted 2026-07-31 quant-ph math-phmath.MP

Yanying Liang , Haoran Zhu This is my paper

classification quant-phmath-phmath.MP MSC 81P4505C5015A1860J10

keywords randomquantumcircuitsiSWAPgatespectralgapsecond-momentoperatorunitary2-designsPerron-Frobeniusfour-pointinequalitiesKAKparameters

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

Random quantum circuits are a practical way to generate approximate unitary 2-designs, but how quickly they converge depends on which two-qubit gate is applied. This paper proves that the iSWAP gate is the best possible choice in a precise sense: on every connected graph with three or more qubits, and among all two-local circuit ensembles whose second-moment operator is Hermitian, the iSWAP circuit has the largest spectral gap. A larger spectral gap means faster convergence toward the uniform Haar distribution over unitaries. The proof resolves a conjecture from the recent gate-design literature by reducing the infinite family of gates to a two-parameter affine boundary family and certifying the comparison through a Perron-Frobenius eigenvector and a cone of four-point inequalities. The result holds for mixtures of gates and for edge-dependent gate choices, not just for a single fixed gate.

What carries the argument

The local second-moment gadget of any two-qubit unitary reduces to h(a,c) = αP₊ + βP₋, where P₊ and P₋ are two fixed orthogonal rank-one projections and α,β are functions of the KAK coordinates, constrained to the affine polytope 0≤α≤10/9 and 0≤β≤2−3α/5. The proof works on the upper boundary β=2−3α/5, applies the similarity transform that maps the two product zero modes to the consensus states, and obtains a block M_α indexed by non-empty proper subsets of V; this block is a Z-matrix, meaning all off-diagonal entries are non-positive. The named device carrying the argument is the cone C of asymmetric four-point inequalities 4f(A∪{i}) + f(A∪{j}) ≥ 2f(A) + 2f(A∪{i,j}), plus the swapped version

What would settle it

Compute, for a small connected graph such as a triangle or a four-vertex path, the full second-moment spectrum of an edge-dependent Hermitian ensemble with parameters on the boundary β=2−3α/5. If any choice of αₑ in [0,10/9] gave min Spec(M_α) > γ₀, the certified comparison in Theorem 6.3 would fail; more directly, if any Hermitian two-local ensemble had ∆(Tᴱ) > ∆(T^iSWAP), Theorem 1.1 would be false. The explicit matrix formulas make this a finite numerical search.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.1: for any connected graph G with n≥3, and any prescribed two-qubit probability distribution on each edge of G, if the resulting second-moment operator is Hermitian, then its spectral gap is at most that of the homogeneous iSWAP ensemble. Since the spectral gap controls the asymptotic convergence rate of the random circuit to a unitary 2-design, this identifies iSWAP as the universal maximiser of second-moment mixing speed among Hermitian two-local circuits. The proof has three stages. First, every ironed two-qubit gadget is shown, through its KAK coordinates, to lie in a convex envelope whose upper boundary is a one-parameter family h_α. Second, a ground-space

Load-bearing premise

The proof leans on a previously established localisation result that for iSWAP circuits on n≥3 qubits the slowest mode of the full second-moment operator lies in the single-qubit Haar-invariant subspace, and that the full spectral gap equals the compressed gap; if that localisation failed, the comparison would only constrain a compressed operator and would not transfer to the true circuit gap.

Editorial extensions

If this is right

Any Hermitian two-local circuit ensemble on a connected graph with n≥3 converges to a unitary 2-design no faster than the iSWAP circuit; iSWAP sets a universal speed limit for second-moment mixing.
The optimality survives mixtures and edge-dependent gate distributions, so practical circuit implementations may vary gates from edge to edge without exceeding the iSWAP bound.
The theorem covers Hermitian ensembles that may have negative eigenvalues; even allowing such negative spectrum does not improve the spectral gap beyond iSWAP.
The proof reduces the infinite gate comparison to a finite algebraic certificate per graph, so the claimed optimality can in principle be verified mechanically on any fixed graph.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural testable extension is to non-uniform edge selection: the theorem assumes each edge is chosen uniformly at each step, and the edge-dependent comparison uses that uniform weight; reweighting edges would require a modified derivative certificate and is not covered by the paper.
The proof is specific to the second moment; a similar question for third- or higher-order moments would need a new local compression, since the two-projector reduction and the four-point cone rely on the t=2 structure, and nothing in the paper suggests the same gate would automatically be optimal at higher order.
Because the four-point cone is finite and explicit, one could independently verify the spectral comparison on any small fixed graph by solving the cone inequalities numerically—an easy check for n=3 or n=4 that would exercise the proof without full circuit simulation.
If the imported localisation step for iSWAP were shown to hold for other gate families with a less negative minimum eigenvalue, the same compression argument could yield optimality statements for broader families of gates; this is an editorial speculation, not a claim of the paper.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Here's my take on Liang-Zhu. The main result is exactly what it claims: for any connected graph with n≥3 and any Hermitian two-local ensemble with uniform edge selection, the iSWAP gate maximizes the second-moment spectral gap. This resolves Kong-Li-Liu's conjecture and generalizes their complete-graph result. The paper does something real: it builds a four-point cone, shows it's invariant under the transpose semigroup, and uses the Perron–Frobenius eigenvector as a componentwise certificate. That's a new technique, and the proof is carefully written. I checked the linear algebra in Section 6 and the cone identities in Appendix B; they hold up. The reduction to the boundary family is clean, and the Collatz–Wielandt step is legitimate.

The soft spot is the bridge at the very end. Proposition 7.1—that for n≥3 the iSWAP full spectral gap equals the compressed gap γ0—is imported from [17] without proof. The paper states it as a specialization and gives the condition, but the actual verification is in the earlier paper. That's not fatal if [17] is correct, and the authors are transparent about it. But it means the headline theorem is only as solid as that external result. A referee should check it separately, and the authors should either reproduce the argument or point to exactly where in [17] it is proved. Also, the abstract says 'all two-local unitary circuit ensembles' but the theorem requires uniform edge selection; that's an overstatement in the abstract, not in the math.

There's a smaller issue: [19] is listed in the references but never cited in the body. Either cite it or remove it. The reader flagged this; I agree.

The n≥3 restriction is genuinely structural (see Remark 7.2), so that's not a flaw. And the edge-dependent ensemble statement is a real strengthening, not a restatement.

Overall: this is a serious paper that resolves a conjecture with a new method. I'd send it to a referee, and the referee should spend most time on Proposition 7.1 and on the edge-dependent compression to make sure no eigenvalue outside W breaks the gap. If those confirm, it's a strong result. I'd take it to reading group.

Referee Report

1 major / 4 minor

Summary. The paper proves Theorem 1.1: on any connected graph with n≥3, for any edge-dependent two-local unitary ensemble whose second-moment operator is Hermitian, the spectral gap is at most that of the iSWAP circuit. The proof proceeds by reducing every ironed gadget to a two-parameter family (α,β), bounding the allowed parameters by an affine envelope, replacing the Hamiltonian by a boundary family, applying a ground-state similarity transform to obtain a Z-matrix block, and then proving a cone-invariance result for asymmetric four-point inequalities. The Perron–Frobenius eigenvector of the iSWAP block is placed in this cone, yielding a componentwise comparison certificate. Finally, a local Haar compression step and an imported localization result from Kong–Li–Liu identify the compressed iSWAP eigenvalue with the full spectral gap, completing the proof of the conjecture.

Significance. If correct, the paper resolves a conjecture of Kong, Li, and Liu and provides a strong, parameter-free universality result: iSWAP is the optimal Hermitian two-local gate ensemble for second-moment convergence on every connected graph with at least three vertices. The main technical novelty—the invariant polyhedral cone of four-point inequalities and the Perron–Frobenius certificate—is original and likely to be useful for other spectral comparison problems. The proof is detailed and structurally coherent, and the comparison is genuinely parameter-free. The central caveat is that the final identification of the full iSWAP spectral gap with the compressed eigenvalue is imported from [17] rather than proved here, making the main theorem conditional on that external result.

major comments (1)

[Section 7, Proposition 7.1] The proof of Theorem 1.1 relies on the imported localization result that the largest non-trivial eigenvalue of the full iSWAP moment operator is attained in the compressed space W and that the full spectral gap equals γ0. Without this statement, the argument only yields ∆(T^E) ≤ γ0, and since ∆(T^iSWAP) is not independently established, the claimed inequality (1.1) would not follow. The condition n ≥ 2/(1+λ_min) and the value λ_min(T^iSWAP_2) = -1/3 are asserted but not verified in this manuscript. Please either provide a self-contained proof of Proposition 7.1 (or a precise quotation of [17, Lemma 4.13 and Corollary 4.14] with all hypotheses checked), or explicitly state that Theorem 1.1 is conditional on that result. As it stands, the main theorem's sharpness rests entirely on an external, unproved-in-text statement.

minor comments (4)

[Lemma 5.2, proof of identity (5.10)] The displayed expansion contains an arithmetic error: the coefficient of f(A_j) should be -7, not -10, in both occurrences. The final identity is correct, but the intermediate expression is wrong.
[References] Reference [19] (the authors' companion paper) is listed in the bibliography but never cited in the text. It should be cited where relevant or removed from the reference list.
[Section 7, Proposition 7.1] The computation of λ_min(T^iSWAP_2) = -1/3 is stated without proof or reference. Please provide a citation to the known result or a short derivation.
[General] The paper would be easier to evaluate if the imported localization result were summarized more explicitly, including the definition of the full moment operator and the sense in which the compressed eigenvalue determines the gap. Currently the reader must consult [17] to verify the key step.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the proof is a parameter-free comparison reduced to an external localization lemma, not to its own inputs.

full rationale

The paper's central comparison is not circular. The KAK envelope (Prop. 3.4) is derived from elementary bounds on cosines; the parameters alpha,beta range over all admissible two-qubit ensembles and are not fitted to any target. The boundary reduction (Prop. 3.5), the Z-matrix construction (Prop. 4.1, Lemmas 4.2-4.5), and the Perron-Frobenius certificate (Thm. 6.3-6.4) are proved inside the paper. The cone in Sec. 5 is intentionally matched to the derivative columns (Sec. 5.1), but its invariance is established by explicit local identities (Lemma 5.2) and Perron-Frobenius asymptotics, not by assuming the desired inequality. The only load-bearing external input is Proposition 7.1, imported from Kong-Li-Liu [17, Lemma 4.13 and Corollary 4.14]; that is an independent prior result, not authored by the present authors, and it does not contain the target comparison. A failure of that localization would be a correctness risk, not a circularity. The self-citation [19] appears only in the reference list and is not used in the derivation, so it is not load-bearing. No fitted quantity is renamed as a prediction, and no definition is smuggled in via a self-citation chain. Score 0.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The proof is analytical and contains no fitted parameters. Its axioms are standard matrix theory (Perron-Frobenius, Collatz-Wielandt, Schur-Weyl/Haar integration) plus two external results taken from [17]: the explicit KAK form of the ironed second-moment gadget and the localisation of the iSWAP spectral gap to the compressed W-space for n≥3. No new physical entities are introduced.

assumptions (5)

standard math Dim K_2(d)=2 via Schur-Weyl duality / Haar integration (Section 2.1).
Used to identify the two global fixed modes of every second-moment operator; standard representation theory.
standard math Perron-Frobenius theorem and Collatz-Wielandt bounds for irreducible nonnegative matrices / M-matrices (Cor 4.6, Lemma 6.2).
Provides the positive eigenvector ℓ and the spectral comparison certificate.
domain assumption KAK parametrisation of two-qubit unitaries and the explicit ironed-gadget matrix T(a,c) from Kong-Li-Liu [17, Sec.4.4] (Prop 3.1).
The paper records rather than derives this formula; all local KAK dependence flows from it, so correctness of Theorem 1.1 depends on [17]'s computation.
domain assumption iSWAP localisation theorem: for n≥3, the largest non-trivial eigenvalue of the full iSWAP moment operator is attained in W and equals the compressed eigenvalue (Prop 7.1, from [17, Lemma 4.13 & Cor 4.14]).
This bridges the compressed comparison certificate (γ0) to the full iSWAP spectral gap; if false, Theorem 1.1's conclusion would not follow.
standard math Irreducibility of k-token graphs of connected graphs / finite exclusion process (Lemma 4.3).
Used to prove irreducibility of the iSWAP transient block M0.

how reviews work

0 comments

Cite this review

Pith. "Pith review of iSWAP maximises the second-moment spectral gap in random quantum circuits." pith.science (2026). https://pith.science/paper/7PY6OXBV

@misc{pith2026260729551,
  author       = {Pith},
  title        = {Pith review of: iSWAP maximises the second-moment spectral gap in random quantum circuits},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7PY6OXBV}},
  note         = {Machine review of arXiv:2607.29551}
}

abstract

We prove that the $\mathrm{iSWAP}$ gate maximises the spectral gap of the Hermitian second-moment operator on every connected graph with at least three vertices, among all two-local unitary circuit ensembles. We further prove that the polyhedral cone defined by asymmetric four-point inequalities is invariant under the transpose of the $\mathrm{iSWAP}$ semigroup, yielding a componentwise comparison certificate for its positive Perron--Frobenius eigenvector. These results resolve a conjecture of Kong, Li, and Liu.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

25 extracted references · 4 linked inside Pith

[17]

L. Kong, Z. Li and Z.-W. Liu,Convergence efficiency of quantum gates and circuits, arXiv:2411.04898, 2024

arXiv 2024
[19]

Liang and H

Y. Liang and H. Zhu,Spectral gaps of ironed two-qubit gadgets matching the iSWAP gap, preprint, arXiv:2607.28521, 2026

arXiv 2026
[1]

Belkin, J

D. Belkin, J. Allen, S. Ghosh, C. Kang, S. Lin, J. Sud, F.T. Chong, B. Fefferman and B.K. Clark, Approximatet-designs in generic circuit architectures, PRX Quantum5(2024), 040344

2024
[2]

Berman and R.J

A. Berman and R.J. Plemmons,Nonnegative Matrices in the Mathematical Sciences, Classics in Ap- plied Mathematics, vol. 9, Society for Industrial and Applied Mathematics, Philadelphia, 1994

1994
[3]

Brand˜ ao, A.W

F.G.S.L. Brand˜ ao, A.W. Harrow and M. Horodecki,Local random quantum circuits are approximate polynomial-designs, Commun. Math. Phys.346(2016), 397–434

2016
[4]

Brown and L

W.G. Brown and L. Viola,Convergence rates for arbitrary statistical moments of random quantum circuits, Phys. Rev. Lett.104(2010), 250501

2010
[5]

Collins and P

B. Collins and P. ´Sniady,Integration with respect to the Haar measure on unitary, orthogonal and symplectic group, Commun. Math. Phys.264(2006), 773–795

2006
[6]

Dankert, R

C. Dankert, R. Cleve, J. Emerson and E. Livine,Exact and approximate unitary2-designs and their application to fidelity estimation, Phys. Rev. A80(2009), 012304. 23

2009

Show all 25 references

[7]

Emerson, R

J. Emerson, R. Alicki and K. ˙Zyczkowski,Scalable noise estimation with random unitary operators, J. Opt. B: Quantum Semiclass. Opt.7(2005), S347–S352

2005
[8]

Fabila-Monroy, D

R. Fabila-Monroy, D. Flores-Pe˜ naloza, C. Huemer, F. Hurtado, J. Urrutia and D.R. Wood,Token graphs, Graphs Combin.28(2012), 365–380

2012
[9]

Farina and S

L. Farina and S. Rinaldi,Positive Linear Systems: Theory and Applications, Pure and Applied Math- ematics, Wiley, New York, 2000

2000
[10]

Gross, K

D. Gross, K. Audenaert and J. Eisert,Evenly distributed unitaries: on the structure of unitary designs, J. Math. Phys.48(2007), 052104

2007
[11]

Haferkamp,Random quantum circuits are approximate unitaryt-designs in depthO(nt 5+o(1)), Quan- tum6(2022), 795

J. Haferkamp,Random quantum circuits are approximate unitaryt-designs in depthO(nt 5+o(1)), Quan- tum6(2022), 795

2022
[12]

Harrow and R.A

A.W. Harrow and R.A. Low,Random quantum circuits are approximate2-designs, Commun. Math. Phys.291(2009), 257–302

2009
[13]

Harrow and S

A.W. Harrow and S. Mehraban,Approximate unitaryt-designs by short random quantum circuits using nearest-neighbor and long-range gates, Commun. Math. Phys.401(2023), 1531–1626

2023
[14]

Horn and C.R

R.A. Horn and C.R. Johnson,Matrix Analysis, 2nd ed., Cambridge University Press, Cambridge, 2013

2013
[15]

Hunter-Jones,Unitary designs from statistical mechanics in random quantum circuits, arXiv:1905.12053, 2019

N. Hunter-Jones,Unitary designs from statistical mechanics in random quantum circuits, arXiv:1905.12053, 2019

1905 arXiv
[16]

Khaneja, R

N. Khaneja, R. Brockett and S.J. Glaser,Time optimal control in spin systems, Phys. Rev. A63 (2001), 032308

2001
[18]

Kraus and J.I

B. Kraus and J.I. Cirac,Optimal creation of entanglement using a two-qubit gate, Phys. Rev. A63 (2001), 062309

2001
[20]

Levin, Y

D.A. Levin, Y. Peres and E.L. Wilmer,Markov Chains and Mixing Times, 2nd ed., American Math- ematical Society, Providence, RI, 2017

2017
[21]

Mittal and N

S. Mittal and N. Hunter-Jones,Local random quantum circuits form approximate designs on arbitrary architectures, arXiv:2310.19355, 2023

2023 arXiv
[22]

Seneta,Non-negative Matrices and Markov Chains, 2nd ed., Springer, New York, 1981

E. Seneta,Non-negative Matrices and Markov Chains, 2nd ed., Springer, New York, 1981

1981
[23]

Zhang, J

J. Zhang, J. Vala, S. Sastry and K.B. Whaley,Geometric theory of nonlocal two-qubit operations, Phys. Rev. A67(2003), 042313

2003
[24]

Brown and O

W. Brown and O. Fawzi,Decoupling with random quantum circuits, Commun. Math. Phys.340(2015), 867–900. 24

2015
[25]

Ho and S

W.W. Ho and S. Choi,Exact emergent quantum state designs from quantum chaotic dynamics, Phys. Rev. Lett.128(2022), 060601. 25

2022

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [17]

L. Kong, Z. Li and Z.-W. Liu,Convergence efficiency of quantum gates and circuits, arXiv:2411.04898, 2024

arXiv 2024

[2] [19]

Liang and H

Y. Liang and H. Zhu,Spectral gaps of ironed two-qubit gadgets matching the iSWAP gap, preprint, arXiv:2607.28521, 2026

arXiv 2026

[3] [1]

Belkin, J

D. Belkin, J. Allen, S. Ghosh, C. Kang, S. Lin, J. Sud, F.T. Chong, B. Fefferman and B.K. Clark, Approximatet-designs in generic circuit architectures, PRX Quantum5(2024), 040344

2024

[4] [2]

Berman and R.J

A. Berman and R.J. Plemmons,Nonnegative Matrices in the Mathematical Sciences, Classics in Ap- plied Mathematics, vol. 9, Society for Industrial and Applied Mathematics, Philadelphia, 1994

1994

[5] [3]

Brand˜ ao, A.W

F.G.S.L. Brand˜ ao, A.W. Harrow and M. Horodecki,Local random quantum circuits are approximate polynomial-designs, Commun. Math. Phys.346(2016), 397–434

2016

[6] [4]

Brown and L

W.G. Brown and L. Viola,Convergence rates for arbitrary statistical moments of random quantum circuits, Phys. Rev. Lett.104(2010), 250501

2010

[7] [5]

Collins and P

B. Collins and P. ´Sniady,Integration with respect to the Haar measure on unitary, orthogonal and symplectic group, Commun. Math. Phys.264(2006), 773–795

2006

[8] [6]

Dankert, R

C. Dankert, R. Cleve, J. Emerson and E. Livine,Exact and approximate unitary2-designs and their application to fidelity estimation, Phys. Rev. A80(2009), 012304. 23

2009