CoChat
Sign Up
Cited by 466Quantum Computing Algorithms and ArchitectureQuantum Information and CryptographyQuantum many-body systems

Characterizing quantum gates via randomized benchmarking

Open access
Magesan, Easwar, Gambetta, Jay M., Emerson, Joseph·2012·Physical Review A
DOI 10.1103/physreva.85.042311
Open this paper in CoChat — free accountOpen PDF

Abstract

We describe and expand upon the scalable randomized benchmarking protocol proposed in Phys. Rev. Lett. 106, 180504 (2011) which provides a method for benchmarking quantum gates and estimating the gate dependence of the noise. The protocol allows the noise to have weak time and gate dependence, and we provide a sufficient condition for the applicability of the protocol in terms of the average variation of the noise. We discuss how state-preparation and measurement errors are taken into account and provide a complete proof of the scalability of the protocol. We establish a connection in special cases between the error rate provided by this protocol and the error strength measured using the diamond norm distance.

AI snapshot

Written from the paper’s full text.

TL;DR

This paper analyzes and expands a scalable randomized benchmarking protocol for Clifford gates, providing zeroth- and first-order fitting models that tolerate weak time- and gate-dependent noise and account for state-preparation and measurement errors. It proves scalability and establishes an exact relationship between the benchmarking error rate and diamond-norm error in special Pauli cases.

Key findings

  • The protocol requires at most \(O(n^2)\) quantum gates, \(O(n^4)\) classical pre-processing, and a number of single-shot repetitions independent of \(n\); random Clifford sampling and decomposition total \(O(n^4)\) classical operations.
  • The zeroth-order model gives exponential fidelity decay when noise is gate- and time-independent; the first-order model adds a term \(C_1(m-1)(q(m)-p^2)p^{m-2}\) that measures gate-dependence of the noise.
  • State-preparation and measurement errors are absorbed into fit parameters \(A_0,B_0\) or \(A_1,B_1,C_1\); the protocol is robust against them unless the fidelity curve is constant, which occurs only for \(p=0\), \(p=1\), or \(A_0=0\).
  • For generalized Pauli channels, the paper establishes an exact relationship between the benchmarking error rate \(r=1-F_{\Lambda,I}\) and the diamond norm \(\|\Lambda-I\|_\diamond\).

Methods

  • For each sequence length \(m\), the protocol generates \(K_m\) sequences of \(m+1\) Clifford operations: the first \(m\) are uniformly random, and the final operation is the inverse of their composition. It measures the survival probability, averages over sequences, repeats for different \(m\), and fits the fidelity decay.
  • The analysis uses a perturbative expansion of gate- and time-dependent errors about the average error operator, Clifford twirling, and zeroth- and first-order fitting models. A sufficient condition for neglecting higher orders is derived using the \(\|\cdot\|_{H^{1\to1}}\) norm.
  • Scalability is proven using the symplectic representation of the Clifford group, Hoeffding’s inequality for the number of repetitions, and decomposition of Clifford operations into one- and two-qubit generators. No new experimental data are reported.

Limitations

  • The protocol assumes weak time- and gate-dependence of the noise; the paper states a sufficient condition in terms of the average variation of the noise, and higher-order terms are not negligible when this condition fails.
  • The paper notes a pathological failure case: if the noise is gate-dependent and equal to the exact inverse of the target gate, benchmarking can report zero error despite substantial error on each gate.
  • The exact relationship between the benchmarking error rate and the diamond norm is established only for generalized Pauli channels; the paper states that a general explicit relationship is impossible and leaves broader connections for future work.
  • The authors state that how the sufficient condition depends on the number of qubits \(n\) is left for further research, and that noise correlations between qubits may create large errors on Clifford elements even when errors on a generating set are controlled.
  • Evident from the text: this is a theoretical and protocol-analysis paper; it does not provide new experimental validation.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1103/physreva.85.042311

Cites

51 of 51 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 27 cited sources checked · 0 supported · 0 flagged

  • Findingnone given
    § I. INTRODUCTION · p. 1

    The average number of generators in such a decomposition is 1.875.

    “The average number of generators in such a decomposition is 1.875 which implies that even for the single-qubit case our protocol takes no more time to implement than that of [26].”
    Quote in the paperSubstantial claim
  • Findingcitation
    § I. INTRODUCTION · p. 1

    In the single-qubit case, the RB scheme can be implemented by explicitly writing down the 24 elements of the Clifford group decomposed into a sequence of the same generators randomly applied in [26].

    “In the single-qubit case, our RB scheme can be implemented by explicitly writing down the 24 elements of the Clifford group decomposed into a sequence of the same generators that are randomly applied in [26].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [26] E. Knill, D. Leibfried, R. Reichle, J. Britton, R. B. Blakestad, J. D. Jost, C. Langer, R. Ozeri, S. Seidelin, and D. J. Wineland, Physical Review A 77 , 012307 (2008).

      Not checked in the free run.

  • Findingnone given
    § A. Protocol · p. 6

    For each m, in the limit K_m → ∞, F_seq(m,ψ) converges to the exact uniform average F_g(m,ψ) over all sequences.

    “We note that for each m , in the limit of K m → ∞ , F seq ( m,ψ ) converges to the exact (uniform) average, F g ( m,ψ ), over all sequences,”
    Quote in the paperSubstantial claim
  • Findingnone given
    § Definition 1. Unitary t-Design · p. 5

    χ_{0,0} for Λ_k^dep decreases to 1/d^2 exponentially in k.

    “Moreover χ 0 , 0 for Λ k dep decreases to 1 d 2 exponentially in k.”
    Quote in the paperNot a substantial claim
  • Findingreasoning
    § 1. Zeroth Order Model · p. 8

    The zeroth order fitting model is exact when the noise is independent of both the gate chosen and time.

    “First, we look at the zeroth order fitting model F (0) g ( m, | ψ 〉 ) and note that F (0) g ( m, | ψ 〉 ) is exact in the case that the noise is independent of both the gate chosen and time, ie. Λ i j ,j = Λ.”
    Quote in the paperSubstantial claim
  • May combine two claims. 30
  • The quote only partly states this claim. 2
  • Classified as probably not a substantial claim. 15

  • May combine two claims.

    Findingreasoning
    § I. INTRODUCTION · p. 1

    When the error is gate-dependent and equal to the exact inverse of the target gate, the error rate given by the protocol is always zero while there is substantial error on each gate.

    “An unphysical but intuitively simple example is when the error is gate-dependent and equal to the exact inverse of the target gate. The error rate given by the protocol is always equal to zero however in actuality there is substantial error on each gate (see Sec. IV B).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § I. INTRODUCTION · p. 2

    The protocol requires at most O(n^2) quantum gates, O(n^4) cost in classical pre-processing, and a number of single-shot repetitions that is independent of n.

    “We provide a detailed proof that our protocol requires at most O ( n 2 ) quantum gates, O ( n 4 ) cost in classical pre-processing (to select each gate-sequence), and a number of single-shot repetitions that is independent of n .”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning
    § B. Case Where Benchmarking Fails · p. 12

    In a time-independent noise model where each gate's noise is the inverse of the gate (Λ_i = C_i†), the benchmarking fidelity F_g(m,ψ) equals 1 for every m despite substantial error on each C_i, so benchmarking fails.

    “Suppose the noise is timeindependent and for each i, Λ i = C † i . Then F g ( m,ψ ) = 1 for every m even though there is substantial error on each C i and so benchmarking clearly fails.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning
    § VI. SCALABILITY OF THE PROTOCOL · p. 15

    The size of the Clifford group scales as 2^{O(n^2)} and the number of sequences of length m scales as 2^{mO(n^2)}; hence full averaging over the Clifford group for each sequence length does not scale well in either n or m.

    “First, we note that the size of the Clifford group scales as 2 O ( n 2 ) and so the number of sequences of length m scales as 2 mO ( n 2 ) . Hence if full averaging over the Clifford group is required for each sequence length, our protocol does not scale well in either of n or m .”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning
    § C. State Preparation and Measurement Errors · p. 13

    p=1 corresponds to Λ being the identity channel, meaning all gates are perfect; in this case the fidelity equals A0+B0 = tr(Λ(ρψ)Eψ) = tr(ρψEψ).

    “p = 1: This case corresponds to Λ being the identity channel which means all gates are perfect. Again, in practice this situation is unlikely as the implementation of any gate will have some associated error. Note that in this case the fidelity is equal to A 0 + B 0 which is just tr(Λ( ρ ψ ) E ψ )) = tr( ρ ψ E ψ ).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingnone given
    § C. State Preparation and Measurement Errors · p. 13

    p=0 occurs if and only if Λ is the totally depolarizing channel, and in this case the fidelity is constant at B0 = tr(Eψ)/d ≤ 1/d.

    “p = 0: This occurs if and only if Λ is the totally depolarizing channel and in this case the fidelity is constant at B 0 = tr( E ψ ) d ≤ 1 d .”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning
    § C. State Preparation and Measurement Errors · p. 13

    From Eq. (3.26), an exponential decay occurs if and only if A0 is non-zero and p lies in (0,1); hence no decay occurs if and only if one of p=0, p=1, or A0=0 occurs.

    “From Eq. (3.26) an exponential decay occurs if and only if A 0 is non-zero and p lies in (0 , 1). Hence no decay occurs if and only if one of p = 0, p = 1 or A 0 = 0 occurs.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning
    § B. Relating the Diamond Norm and Error Rate in Benchmarking · p. 15

    If E2=I in Eq. (5.4), then r0=1 and for every 1≤i≤d^2-1, ri=0.

    “Now suppose that E 2 = I in Eq. (5.4). Then, r 0 = 1 and for every 1 ≤ i ≤ d 2 -1, r i = 0.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning
    § C. State Preparation and Measurement Errors · p. 13

    In the A0=0 case, the fidelity equals B0 and can take any value in [0,1].

    “Note that the fidelity will be equal to B 0 in this case and so can take any value in [0 , 1].”
    Quote in the paperNot a substantial claim

  • May combine two claims. The quote only partly states this claim.

    Findingreasoning
    § VI. SCALABILITY OF THE PROTOCOL · p. 16

    The number of trials k needed to estimate F_g(m,ψ) to accuracy ε with confidence 1-δ is given by Eq. (6.4) and is explicitly independent of m and n.

    “Note that k is explicitly independent of m and n which provides a solution to 1.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Findingnone given
    § VI. SCALABILITY OF THE PROTOCOL · p. 16

    Performing process tomography on a 4-qubit system already requires 65536 measurements.

    “As a direct comparison, performing process tomography on a 4 qubit system already requires 65536 measurements.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning
    § VI. SCALABILITY OF THE PROTOCOL · p. 16

    The expected value of the normalized k-fold sum S_k(m,|ψ〉) equals F_g(m,ψ), and Hoeffding's inequality gives a probabilistic bound on |S_k(m,|ψ〉) - F_g(m,ψ)|.

    “Let S k ( m, | ψ 〉 ) = F ⃗ i m g ( m, | ψ 〉 )+ .... + F ⃗ i m g ( m, | ψ 〉 ) k be the normalized k-fold sum of the random variable F i ⃗ m g ( m, | ψ 〉 ) and note that BX [ S k ( m, | ψ 〉 )] = F g ( m,ψ ). A probablistic bound on | S k ( m, | ψ 〉 ) -F g ( m,ψ ) | is given by H¨ oeffding's inequality,”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingreasoning
    § Solution to 2: · p. 16

    A random Clifford element can be produced in O(n^4) classical operations.

    “Since randomly choosing 2 n elements of the Pauli group that satisfy the required commutation relations is equivalent to inductively choosing random solutions to 2 n sets of linear equations (which requires O ( n 3 ) operations), we can produce a random Clifford element in O ( n 4 ) (classical) operations.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation
    § I. INTRODUCTION · p. 1

    The protocol of [26] is limited to the single-qubit case and fits observed fidelity decay to an exponential.

    “The protocol of [26] is limited to the single-qubit case and fits the observed fidelity decay averaged over sequences of single-qubit gates (where each gate consists of a random generator of the Clifford group composed with a random Pauli operator) to an exponential.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [26] E. Knill, D. Leibfried, R. Reichle, J. Britton, R. B. Blakestad, J. D. Jost, C. Langer, R. Ozeri, S. Seidelin, and D. J. Wineland, Physical Review A 77 , 012307 (2008).

      Not checked in the free run.

  • May combine two claims.

    Premisecitation
    § I. INTRODUCTION · p. 1

    The protocol of [18] assumes noise independent of the chosen gate, in which case the fidelity decay curve averaged over randomly chosen unitaries is exponential in sequence length.

    “For instance [18] assumes the highly idealized situation of the noise being independent of the chosen gate, in which case the fidelity decay curve averaged over randomly chosen unitaries takes the form of an exponential (in the sequence length).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [18] J. Emerson, R. Alicki, and K. Zyczkowski, Journal of Optics B: Quantum and Semiclassical Optics 7 , S347 (2005). source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisenone given
    § I. INTRODUCTION · p. 1

    Conditions for when the assumption of an exponential decay is valid in the realistic case of gate-dependent and time-dependent noise were not given by previous protocols.

    “However, conditions for when the assumption of an exponential decay is valid, specifically in the realistic case of gatedependent and time-dependent noise, were not given.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data
    § I. INTRODUCTION · p. 2

    The zeroth order model shows that for time-independent and gate-independent errors the fidelity decay is modeled by an exponential decay, and the decay rate produces an estimate for the average error rate of the noise.

    “Our zeroth order model directly shows that for time-independent and gate-independent errors the fidelity decay is indeed modeled by an exponential decay, and the decay rate produces an estimate for the average error rate of the noise.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation
    § I. INTRODUCTION · p. 2

    Gate-dependent errors can lead to a deviation from the exponential decay, defining a partial test for such effects in the noise.

    “The fitting formula shows that gatedependent errors can lead to a deviation from the exponential decay (defining a partial test for such effects in the noise), which was illustrated via numerical examples in [25].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [25] E. Magesan, J. M. Gambetta, and J. Emerson, Phys. Rev. Lett. 106 , 180504 (2011). source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation
    § A. Diamond Norm, Average Gate Fidelity and Minimum Gate Fidelity · p. 4

    Twirling a quantum operation over the Clifford group produces a depolarizing channel and the average fidelity is invariant under the twirling operation.

    “Thus twirling a quantum operation over the Clifford group produces a depolarizing channel and the average fidelity is invariant under the twirling operation.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [18] J. Emerson, R. Alicki, and K. Zyczkowski, Journal of Optics B: Quantum and Semiclassical Optics 7 , S347 (2005). source

      Not checked in the free run.

    • cites [40] M. Nielsen, Physics Letters A 303 , 249 (2002). source

      Not checked in the free run.

  • May combine two claims.

    Premisenone given
    § I. INTRODUCTION · p. 1

    Previous RB protocols have shortcomings: extensions to multi-qubit systems are either not scalable or not well understood, and it is unclear how to explicitly account for state preparation and measurement errors.

    “Other important shortcomings of these previous RB protocols are that extensions to multi-qubit systems are either not scalable or not well understood, and it is unclear how to explicitly account for state preparation and measurement errors.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationcitation
    § I. INTRODUCTION · p. 1

    The paper gives a full analysis of the scalable multi-qubit randomized benchmarking protocol for Clifford gates proposed in [25] which overcomes the shortcomings described above.

    “In this paper we give a full analysis of the scalable multi-qubit randomized benchmarking protocol for Clifford gates we proposed in [25] which overcomes the shortcomings described above.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [25] E. Magesan, J. M. Gambetta, and J. Emerson, Phys. Rev. Lett. 106 , 180504 (2011). source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisenone given
    § Definition 3. Average Error Rate · p. 5

    Monte-Carlo methods have the drawback of assuming negligible state-preparation and measurement errors.

    “The Monte-Carlo methods also have the drawback of assuming negligible state-preparation and measurement errors.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given
    § Definition 3. Average Error Rate · p. 5

    The average error rate r and the Pauli channel error rate r_P are related via r_P = (d+1)r/d.

    “One can show that r and r P are related via r P = ( d +1) r d .”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § B. Case Where Benchmarking Fails · p. 12

    In that pathological example, the noise is highly gate-dependent, so the sufficient condition for ignoring higher-order terms is not satisfied (γ is far from 0).

    “The key point to note here is that the noise is highly dependent on the gate chosen and so we expect that the sufficient condition derived above for ignoring higher order terms will not be satisfied (ie. γ in this example will be far from 0).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning
    § 2. First Order Model · p. 11

    The difference between Eq. (3.44) and Eq. (3.26) is the C1(m-1)(q(m)-p^2)p^{m-2} term, which can be thought of as a measure of the gate-dependence of the noise.

    “The difference between Eq.'s (3.44) and (3.26) is the C 1 ( m -1)( q ( m ) -p 2 ) p m -2 term contained in Eq. (3.44), which can be thought of as a measure of the gate-dependence of the noise.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given
    § Definition 3. Average Error Rate · p. 5

    The advantages of tomography and Monte-Carlo methods are that the average fidelity of each gate can be estimated and the scheme is efficient in n.

    “The advantages of these methods are that the average fidelity of each gate can be estimated and the scheme is efficient in n .”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § A. Bounding Higher Order Perturbation Terms · p. 12

    The diamond norm is much stronger than the H1→1 norm, so the γ associated with H1→1 will be much smaller than γ associated with the diamond norm, providing a tighter bound on the fidelity difference.

    “The diamond norm ‖ · ‖ ⋄ is a candidate however by Eq. (2.11) ‖ ‖ H 1 → 1 is much weaker than ‖ · ‖ ⋄ . Therefore γ associated with ‖ ‖ H 1 → 1 will be much smaller than γ associated with ‖ · ‖ ⋄ , providing a tighter bound on the fidelity difference.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationnone given
    § V. AVERAGE ERROR RATE AND THE DIAMOND NORM · p. 14

    In general an explicit relationship between average error rate r and diamond norm is impossible to obtain, but in certain cases relevant to fault-tolerant noise models such a relationship can be obtained.

    “In general an explicit relationship will be impossible to obtain, however we show that in certain cases that are relevant in various fault-tolerant noise models we can obtain such a relationship.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning
    § C. State Preparation and Measurement Errors · p. 13

    The protocol is robust against any state preparation or measurement errors unless these errors create a constant fidelity curve.

    “Thus the protocol is robust against any state preparation or measurement errors unless these errors create a constant fidelity curve.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning
    § C. State Preparation and Measurement Errors · p. 13

    A constant fidelity decay curve can only occur in extreme cases, so it is safe to assume the protocol is independent of state preparation and measurement errors.

    “This discussion shows that a constant fidelity decay curve can only occur in extreme cases and so it is safe to assume the protocol is independent of state preparation and measurement errors.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: In practice, gate operations are reasonably precise, so substantial state-preparation or measurement errors are unlikely.

  • May combine two claims. The quote only partly states this claim.

    Interpretationreasoning
    § VI. SCALABILITY OF THE PROTOCOL · p. 16

    For realistic parameter values δ=0.05, ε=10^-3, b-a=0.2, the required number of trials k is large but independent of n, and this compares favorably with quantum process tomography which scales as 16^n.

    “While this number is large it is independent of n and thus compares favourably with quantum process tomography which scales as 16 n .”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: The physical gates have errors on the order of 10^-4.
    • The authors state this as an assumption: The accuracy ε can be relaxed as m grows large.
    • The authors state this as an assumption: The range b-a can be assumed relatively small for small m but converges to 1 - 1/d as m grows large.
    • The authors state this as an assumption: m=100 and the fidelity decay curve is well-approximated by an exponential.

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given
    § VII. DISCUSSION · p. 17

    The unitary group can be generated by adding just one single-qubit rotation not in the Clifford group (for instance the π/8-gate).

    “Second, the unitary group can be generated by adding just one single-qubit rotation not in the Clifford group (for instance the π 8 -gate).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § VII. DISCUSSION · p. 17

    Benchmarking the Clifford group provides significant information for both fault-tolerant quantum computation and obtaining a benchmark for a generating set of the full unitary group.

    “It is not difficult to see that benchmarking the Clifford group provides significant information for both fault-tolerant quantum computation as well as obtaining a benchmark for a generating set of the full unitary group.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Any realistic implementation of a quantum computer will have to take advantage of error-correction codes to perform fault-tolerant quantum computation.
    • Rests on an assumption: Most of the codes used in fault-tolerant theory are stabilizer codes.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given
    § VII. DISCUSSION · p. 18

    The noise associated to a multi-qubit Clifford element is given by the noise associated to the sequence of generators comprising the Clifford.

    “As previously noted, the noise associated to a multi-qubit Clifford element is given by the noise associated to the sequence of generators comprising the Clifford.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Premisenone given
    § VII. DISCUSSION · p. 17

    Most of the codes used in fault-tolerant theory are stabilizer codes, which implies that encoding and decoding operations can be chosen to be Clifford operations.

    “The fact that most of the codes used in fault-tolerant theory are stabilizer codes implies that the encoding and decoding operations that have to be performed can be chosen to be Clifford operations.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Most of the codes used in fault-tolerant theory are stabilizer codes.

  • Classified as probably not a substantial claim.

    Premisenone given
    § VII. DISCUSSION · p. 18

    Rigorous fault-tolerant analyses sometimes invoke the diamond norm as a measure of error strength rather than the weaker characterization provided by average fidelity.

    “Rigorous fault-tolerant analyses sometimes invoke the diamond norm as a measure of the error strength rather than the weaker characterization provided by the average fidelity.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning
    § VII. DISCUSSION · p. 17

    Benchmarking the full unitary group is provably inefficient because generating a Haar-random unitary operator is inefficient in n; benchmarking the Clifford group is an efficient task.

    “While benchmarking the full unitary group would be ideal, this is a provably inefficient task since just generating a Haar-random unitary operator is inefficient in n . On the other hand as we have shown here benchmarking the Clifford group is an efficient task.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning
    § VII. DISCUSSION · p. 18

    The diamond norm is essentially a worst-case maximization over input (entangled) states.

    “From the expression given in Eq. (2.2) one can see that the diamond norm is essentially a 'worst-case' maximization over input (entangled) states.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning
    § VII. DISCUSSION · p. 18

    In that case the noise must have a strong gate-dependence because otherwise q - p^2 would be small, implying the two fitting functions would produce similar estimates for p.

    “In this case the noise must have a strong gate-dependence because otherwise q -p 2 would be small which implies the two fitting functions would produce similar estimates for p .”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationnone given
    § VII. DISCUSSION · p. 17

    The protocol allows for time- and gate-dependent noise, and the fitting models for the fidelity function take into account state preparation and measurement errors.

    “The protocol allows for time and gatedependent noise and the fitting models for the fidelity function take into account state preparation and measurement errors.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given
    § VII. DISCUSSION · p. 18

    There is a direct relationship between the minimum gate fidelity and the diamond norm given by Eq. (2.21).

    “Additionally, if one were able to obtain an estimate of the minimum gate fidelity from knowledge of the average fidelity they could use the direct relationship between the minimum gate fidelity and diamond norm given by Eq. (2.21) to obtain information about the error strength in terms of the diamond norm.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § VII. DISCUSSION · p. 18

    The concentration of measure effect of the gate fidelity implies that as n increases, the measure of the set of states which produce a fidelity close to the minimum yet far from the average is exponentially small in n.

    “A result that may be useful in this direction of research is the 'concentration of measure effect' of the gate fidelity which implies that as n increases, the measure of the set of states which produce a fidelity close to the minimum yet far from the average is exponentially small in n [41, 42].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [41] E. Magesan, R. Blume-Kohout, and J. Emerson, Phys. Rev. A 84 , 012309 (2011).

      Not checked in the free run.

    • cites [42] E. Magesan, Quant. Inf. Comp. 11 , 0466 (2011). source

      Not checked in the free run.

42 of 42

  • The authors state this as an assumption.

    The actual implementation of each Clifford gate C_i at time j can be represented as E_i,j = Λ_i,j ◦ C_i for some error map Λ_i,j.

    “Suppose that the actual implementation of C i at time j (1 ≤ j ≤ M ) results in the map E i,j with E i,j = Λ i,j ◦ C i for some error map Λ i,j .”

  • The authors state this as an assumption.

    Each operation C_i_j is allowed to have some error, represented by Λ_i_j,j.

    “By assumption each operation C i j is allowed to have some error, represented by Λ i j ,j , and each sequence can be modelled by the operation,”

  • The authors state this as an assumption.

    In the simplest ideal scenario, the noise operator at each step is independent of the applied gate and time-invariant.

    “assuming the simplest (ideal) scenario where the noise operator at each step is independent of the applied gate (and is also time-invariant)”

  • The authors state this as an assumption.

    The perturbation δΛ_i,j is small relative to Λ.

    “Our approach will be valid provided δ Λ i,j is a small perturbation from Λ in a sense to be made precise later.”

  • The authors state this as an assumption.

    The gate-dependence of the noise is weak enough that the zeroth-order expression in Eq. (3.26) is a valid model for the fidelity decay curve.

    “For simplicity of the discussion let us assume the gate-dependence of the noise is weak enough so that the zeroth order expression given in Eq. (3.26) is a valid model for the fidelity decay curve.”

  • The authors state this as an assumption.

    The physical gates have errors on the order of 10^-4.

    “Let us assume that the physical gates have errors on the order of 10 -4 .”

  • The authors state this as an assumption.

    The accuracy ε can be relaxed as m grows large.

    “Intuitively, since the fidelity curve decays in sequence length it is reasonable to assume that ǫ can be relaxed as m grows large.”

  • The authors state this as an assumption.

    The range b-a can be assumed relatively small for small m but converges to 1 - 1/d as m grows large.

    “Similarly, b -a can be assumed to be relatively small for small values of m but will converge to 1 -1 d as m grows large.”

  • The authors state this as an assumption.

    m=100 and the fidelity decay curve is well-approximated by an exponential.

    “Let us assume m = 100 and a fidelity decay curve that is well-approximated by an exponential.”

  • The paper relies on this without support.

    In practice, gate operations are reasonably precise, so substantial state-preparation or measurement errors are unlikely.

    “Since gates are reasonably precise in practice, this situation occurs when at least one of state preparation or measurement has substantial error.”

  • The paper relies on this without support.

    In quantum computation, the measure of accessible states (states reachable in polynomial time using a generating set for the unitary group) is equal to 0.

    “In quantum computation it is the case that the measure of accessible states (states that can be reached in polynomial time using a generating set for the unitary group) is equal to 0.”

  • The paper relies on this without support.

    Any realistic implementation of a quantum computer will have to take advantage of error-correction codes to perform fault-tolerant quantum computation.

    “First, any realistic implementation of a quantum computer will have to take advantage of error-correction codes in order to perform fault-tolerant quantum computation.”

  • The paper relies on this without support.

    Most of the codes used in fault-tolerant theory are stabilizer codes.

    “The fact that most of the codes used in fault-tolerant theory are stabilizer codes implies that the encoding and decoding operations that have to be performed can be chosen to be Clifford operations.”
13 of 13

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Existing works have already proposed to adopt randomly generated benchmarks for evaluating the error rates of different gate operations regarding a particular NISQ device.

    “Existing works have already proposed to adopt randomly generated benchmarks for evaluating the error rates of diferent gate operations regarding a particular NISQ device [83, 84, 106].”

    From QASMBench: A Low-Level Quantum Benchmark Suite for NISQ Evaluation and Simulation (2023) · cites this paper as [84]

    The Introduction (p. 1) describes benchmarking Clifford gates using randomization, and the Discussion (p. 17) says the protocol estimates average fidelity across those gates. It does not establish separate error rates for different gate operations on a particular NISQ device; its reported estimate is an average across the Clifford gate set. (LLM, conf=0.84)

    Experimental — accuracy not yet measured

  • In noisy intermediate-scale quantum (NISQ) systems, performance of isolated gates may not predict the behavior of the system.

    “In these noisy intermediate-scale quantum (NISQ) systems [4], performance of isolated gates may not predict the behavior of the system.”

    From Validating quantum computers using randomized model circuits (2019) · cites this paper as [4]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Conflicting time complexities for Clifford decomposition

    Minor

    The paper says it describes a method that decomposes a Clifford element in O(n2) time, but later assigns O(n3) time to decomposing it into the described sequence of operations.

  • Inconsistent number of operators in the generalized Pauli basis

    Minor

    Indexing the proposed basis from 0 through d2 gives d2+1 operators. Elsewhere the paper indexes such a basis from 0 through d2−1, giving d2 operators.

2 of 2

Cites

Flagged first, then by citations

Research works printed but not linked: 9

  1. P. Shor, 1994, Proceedings of the 35th Annual Symposium on Foundations of Computer Science (FOCS)
  2. D. Aharonov, 1997, Proceedings of the 29th Annual ACM Symposium on Theory of Computing (STOC)
  3. M. J. Biercuk, 2009, Quantum Inf. Comput., 9, 0920
  4. D. Aharonov, 1998, Proceedings of the 30th Annual ACM Symposium on the Theory of Computing, Dallas, TX, 1998
  5. R. Horn, 1990, Matrix Analysis
  6. J. Watrous, 2005, Quantum Inf. Comput., 5, 058
  7. E. Magesan, 2011, Quantum Inf. Comput., 11, 0466
  8. P. Boykin, 1999, Proceedings of the 40th Annual Symposium on Foundations of Computer Science (FOCS)
  9. 10.4086/toc.2009.v005a011 · DOI

Cited by

Sorted by citations

Related

Authors
Magesan, E., Gambetta, J. M., & Emerson, J.
Affiliations
  • University of Waterloo (CA)
  • IBM Research - Thomas J. Watson Research Center (US)
ORCID
Venue
Physical Review A
Publisher
American Physical Society (APS)
Published
2012-04-11
Volume
85
Issue
4
ISSN
  • 0556-2791 (ISSN-L)
  • 1050-2947
  • 1094-1622
DOI
10.1103/physreva.85.042311
OpenAlex
W2017477211
Cited by
464
References
50
Citations by year
2026: 33 · 2025: 67 · 2024: 48 · 2023: 50 · 2022: 47 · 2021: 54 · 2020: 42 · 2019: 35 · 2018: 15 · 2017: 19 · 2016: 15 · 2015: 16 · 2014: 14 · 2013: 5 · 2012: 4
FWCI
12.19
Type
journal article
Language
en
Open access
Licence
link.aps.org/licenses/aps-default-license (vor)
Locations
  • Physical Review A (publishedVersion)
  • arXiv (Cornell University) (publishedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Quantum Computing Algorithms and Architecture
  • Physical Sciences › Computer Science › Artificial Intelligence › Quantum Information and Cryptography
  • Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Quantum many-body systems
Keywords
randomized benchmarking, quantum gate characterization, gate noise, scalability, state preparation and measurement errors, quantum error rates

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-10.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-10.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review