REVIEW 3 major objections 5 minor 37 references
Separate Control of Transient Leakage Exposure and Endpoint Leakage in Fast Transmon Gates
T0 review · 3 major / 5 minor · reviewed 2026-08-02 · deepseek-v4-flash
Pith's one-line read Fast transmon gates leak in two separable ways — endpoint residue and transient exposure — and this paper shows a pulse can control each independently.
desk verdict The endpoint-vs-transient leakage distinction is sound and worth knowing; the 20% path-shaping gain is credible but white-noise-bound, so treat it as a mechanism proof rather than a hardware claim. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central objects are the running leakage amplitude, defined as the time integral of the drive envelope rotated at the |1>-|2> detuning, and the two functionals derived from it: a point evaluation of the control spectrum at the anharmonicity for endpoint leakage, and a band integral with a nonnegative triangular spectral filter for transient exposure. The composite ansatz, called the path-endpoint separation pulse (PESP), combines a cosine-basis path-shaping pulse with a generalized DRAG quadrature (DRAG being derivative-based pulse shaping that adds a quadrature component to cancel leakage at the transition) and two auxiliary endpoint-cancellation tones near the |2> and |3> transitions. T
What would settle it
Measure dephasing-induced excess leakage for a cosine-DRAG pulse and a path-shaped pulse with matched endpoint leakage on a transmon with T_phi near 10 microseconds; if the path-shaped pulse does not show roughly 20% lower excess leakage, the claimed separation fails. Alternatively, rerun the simulation with a measured 1/f-like noise spectrum S(omega): if the path advantage disappears, the white-noise assumption rather than the path–endpoint distinction is doing the work.
Extended reading notes
Core claim
The load-bearing claim is that the two leakage metrics are different functionals of the same drive spectrum. In first-order perturbation theory, leakage amplitude is a running Fourier integral of the drive envelope as seen from the leakage transition; endpoint leakage is the squared amplitude at the end of the gate, a single sample of the spectrum at the |1>-|2> detuning. Transient exposure is the time average of the running population, which expands into an integral over a band of width about 1/T centered at the same detuning, weighted by a nonnegative triangular spectral filter. A spectral null at the transition therefore fixes the endpoint without controlling the band, so endpoint-focused
Load-bearing premise
The quantitative claims rest on a white-noise pure-dephasing model and an idealized closed-system transmon Hamiltonian; under real 1/f noise, drive nonlinearity, or transfer-function distortion, the 20% reduction and 3e-8 endpoint floor could change materially.
Editorial extensions
If this is right
- Endpoint-focused pulse designs — DRAG, spectral-notch pulses, active leakage cancellation, and endpoint-only optimal control — can leave transient exposure essentially unchanged, so small final leakage does not imply small dephasing-induced leakage.
- Optimizing only the final leakage can produce pulses with large transient excursions; under pure dephasing those excursions become residual error (the paper's endpoint-only optimum behaves like cosine DRAG in dephasing simulations).
- Path shaping reduces dephasing exposure by about 21% and dephasing-induced excess leakage by about 20% relative to cosine DRAG, a reduction that persists across the tested 6–15 ns, 0.15–0.25 GHz grid where the optimizer converges.
- A two-tone endpoint correction lowers coherent endpoint leakage from about 7e-7 to 3e-8 without increasing transient exposure, and a five-level check shows negligible leakage into |4>.
- Leakage metrics should be reported as pairs (endpoint leakage, transient exposure) rather than a single number when comparing fast gate schemes.
Reading between the lines
- The paper's white-noise dephasing model is the flat-spectrum special case of a more general noise-response relation; once a device's measured noise spectrum is known, spectrum-tailored shaping (which the paper leaves to future work) could further reduce dephasing-induced leakage.
- The same path–endpoint separation should carry over to other weakly anharmonic qubit platforms and to two-qubit gates, where transient auxiliary-state population during entangling operations may produce even larger dephasing-induced leakage.
- Hardware benchmarks that quote only final or endpoint leakage may misattribute robustness; a fair comparison should match endpoint leakage first and then compare transient exposure under representative dephasing.
- The endpoint floor's sensitivity to anharmonicity mismatch, particularly when the anharmonicity is smaller than assumed, implies that an experimental implementation would need periodic recalibration of the auxiliary tones.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper argues that endpoint leakage (population outside the computational subspace at the end of a gate) and transient leakage exposure (time-integrated or time-averaged leakage population during the gate) are distinct control objectives for fast transmon gates. Using a first-order perturbative treatment, the authors show that endpoint leakage is governed by the control spectrum at the anharmonicity, while transient exposure depends on spectral weight over a finite band around it. They introduce a staged pulse, PESP, consisting of a path-shaping pulse (PSP) that minimizes a time-averaged leakage proxy and a two-tone endpoint-cancellation pulse (ECP). For a 10 ns RX(pi/2) gate with eta/2pi = 0.2 GHz, they report a 21.3% reduction in dephasing exposure and about a 20% reduction in Lindblad excess leakage relative to cosine DRAG, and a lowering of the coherent four-level endpoint leakage from about 7e-7 to 3e-8. The endpoint/path distinction is supported by matched-budget baselines and by an independent Lindblad validation in which excess leakage is not part of the optimization cost.
Significance. If the central distinction holds, it is a useful design principle: endpoint-focused pulse shaping (DRAG, FAST DRAG, ALC) does not automatically reduce dephasing-induced leakage, and separate treatment of transient exposure is warranted. The paper's first-order spectral decomposition (Eqs. 9, 12, 14) is transparent and clean, and the claim is tested with a genuinely independent observable in the four-level Lindblad simulations (Sec. VII), which strengthens the result. The numerical work is careful in several respects: five-seed medians, time-step convergence, a five-level truncation check, and matched-budget baselines. The main caveat is that the quantitative advantage is established only under white dephasing noise; the practical significance under realistic 1/f noise remains unproven, a limitation that the paper acknowledges but does not fully address in its headline claims.
major comments (3)
- [Sec. VII.B, Eq. (24); Abstract; Sec. IX] The headline reductions of 21.3% in dephasing exposure and ~20% in Lindblad excess leakage are computed with S(omega)=S0 (white noise). For realistic 1/f-dominated dephasing, spectral weight is concentrated at low frequencies where |A~(omega)|^2 is suppressed, so the benefit may shrink or vanish. The paper explicitly defers spectrum-tailored optimization to future work, yet the abstract and conclusion quote the reductions without the white-noise qualifier. This is load-bearing for the quantitative claim. Please either add a colored-noise test (e.g., S(omega)=A/|omega| + S0) demonstrating the behavior, or qualify all quantitative reductions in the abstract and conclusion as white-noise reference values.
- [Sec. VI.A, Fig. 2, Table IV] The operating point w_path=30 is selected post hoc as the knee of the four-level leakage-floor sweep, and the endpoint improvement (7e-7 to 3e-8) and the quoted path reduction are evaluated at that point. Although the path reduction lies on a saturated plateau for w_path>=30, the two-tone endpoint floor is not flat across the plateau (Table VII shows w_path=20 gives 1.66e-8 while w_path=30 gives 2.96e-8). The selection rule therefore affects the headline endpoint claim. Please report the endpoint and path values across the plateau in the main text and state a prespecified criterion for choosing w_path, or demonstrate that the qualitative conclusions are unchanged for all plateau values.
- [Sec. V.D, Appendix A, Table V] The main results use a 'path-basin median' obtained by restricting the five seeds to those with P_path^3L / P_path,cosine^3L < 0.95. For the main operating point the occupancy is 5/5, but in the regime grid (Table X) the occupancy drops to 2/5 or 3/5 in many cells, and the reported reductions are medians over only the seeds that found the basin. Since the claim is about the method rather than the optimizer, please also report an all-seeds median or the worst-case seed for the main operating point, and clarify how the basin restriction affects the 21.3% value.
minor comments (5)
- [Table I and Sec. IV.B] The first ECP tone parameter list uses 'phi2' while the second tone also uses 'phi2'. Rename the first tone phase to 'phi' to avoid ambiguity.
- [Fig. 5 caption] The caption lists four panels (a)-(d), but the figure appears to contain six panels (a)-(f) including separate infidelity and leakage panels for amplitude and detuning errors, plus ZOH and low-pass panels. Update the caption to match the panel layout.
- [Data Availability] The data availability statement says data are available from the authors on request. For a numerical study of this kind, depositing the pulse-generation and optimization code would improve reproducibility; consider adding a repository link.
- [Sec. III, Eq. (14)] The exact spectral representation is stated as used only for interpretation, with all reported values computed from the time domain. This is good practice, but the sentence 'the spectral representation is used for interpretation only' is easy to miss; move it to the main text near Eq. (14) for clarity.
- [Sec. VIII.A] The limitation list is comprehensive, but the sentence about the endpoint floor not being comparable to hardware leakage of Ref. [15] could be echoed in the conclusion, where the 3e-8 number is restated without that caveat.
Circularity Check
No significant circularity: the endpoint/path distinction is derived from the pulse-spectrum functionals and checked by a Lindblad observable outside the cost.
full rationale
I find no load-bearing circular step. The central distinction is derived, not assumed: the endpoint leakage is a point evaluation of the control spectrum (Eq. 9, |Λ~(η)|^2), while the transient exposure is a separate kernel-weighted band integral (Eqs. 10-14). Equation (14) explicitly shows that nulling the spectrum at η does not control the Fejér-weighted band integral that determines the path functional. The quantitative ~21% reduction in dephasing exposure is partly a designed outcome, because the Stage-1 cost (Eq. 22) includes the path proxy P̄_A, and P̄_A is numerically close to P̄_deph^A. However, the paper does not present this as a predictive test; it validates the proxy through the four-level Lindblad excess leakage P_excess, which is explicitly not included in any optimization objective (Sec. II.D and Sec. VII). The relation P_excess ≈ γφ T P̄_deph^A (Eq. 7) is derived by first-order perturbation theory, not imposed by definition. The matched-budget endpoint-only and FAST DRAG-L baselines provide controls that do not reproduce the path reduction, supporting the claimed separation of endpoint and path objectives. The white-noise assumption S(ω)=S0 and idealized Hamiltonian are model limitations, openly stated, and spectrum-tailored optimization is deferred to future work; these are robustness concerns, not circular inputs. There are no self-citations by the authors, no imported uniqueness theorem, and no ansatz smuggled in by citation: the Strauch functional is external, acknowledged prior work. The derivation chain is therefore self-contained within the stated model, and the headline claims do not reduce to their inputs by construction.
Assumptions & free parameters
free parameters (6)
- a2, a3 (cosine envelope coefficients) =
a2 ~ 0.28-0.30 at operating point; a3 not reported
- beta1, beta2 (DRAG quadrature coefficients) =
not reported separately
- delta_m (AC-Stark carrier detuning) =
not reported
- w_path (path-penalty weight) =
30
- ECP tone parameters (r_aux, delta_ratio, phi, b2, b3, r_aux2, delta_ratio2, phi2) =
r_aux2 constrained |r_aux2| <= 0.06; others not reported
- Cost weights (w_I, w_L, w_hf, rho_main=0.2, rho_aux=7, w_M=0.1*w_path) =
w_I, w_L, w_hf not specified numerically
assumptions (6)
- domain assumption Transmon is modeled by the rotating-frame Kerr-oscillator Hamiltonian H(t) = -(eta/2)n(n-1) + 1/2(Omega(t)a + Omega*(t)a^dagger) with eta/2pi ~ 200 MHz.
- domain assumption Leakage is dominated by the |1>->|2> transition and described to first order by c2(t) ~ -i*integral Lambda(t')e^{i eta t'} dt' (Eq. 8).
- domain assumption Pure dephasing is white-noise Lindblad with collapse operator n and gamma_phi = 2/T_phi (Eq. 6), giving S(omega)=S0.
- domain assumption Four-level truncation is adequate for the operating-point claims.
- domain assumption The projected computational-subspace fidelity with post-gate virtual-Z optimization (Eq. 2) is the correct figure of merit.
- domain assumption The spectral cutoff C_hf at 0.8 GHz represents the assumed DAC and mixer bandwidth.
Cite this review
Pith. "Pith review of Separate Control of Transient Leakage Exposure and Endpoint Leakage in Fast Transmon Gates." pith.science (2026). https://pith.science/paper/YROFDVIJ
@misc{pith2026260705779,
author = {Pith},
title = {Pith review of: Separate Control of Transient Leakage Exposure and Endpoint Leakage in Fast Transmon Gates},
year = {2026},
howpublished = {\url{https://pith.science/paper/YROFDVIJ}},
note = {Machine review of arXiv:2607.05779}
}
abstract
Leakage to noncomputational states limits the speed of single-qubit gates in weakly anharmonic transmons. Conventional pulse-shaping methods, including derivative removal by adiabatic gate (DRAG), primarily suppress the leakage remaining at the end of a gate. Here we demonstrate that endpoint leakage and the transient leakage population that accumulates during the gate represent distinct control objectives. Endpoint leakage is associated with the drive spectrum at the anharmonicity, whereas transient exposure depends on spectral weight over a finite frequency band and governs the additional leakage induced by dephasing. A spectral null at the leakage transition therefore suppresses the endpoint amplitude without necessarily reducing transient exposure. Based on this distinction, we introduce a path--endpoint separation pulse that combines transient-path shaping with a two-tone endpoint correction. For a $10$ ns $R_X(\pi/2)$ gate with an anharmonicity magnitude of $0.2$ GHz, numerical simulations show a $21\%$ reduction in transient exposure relative to cosine DRAG and a corresponding $20\%$ reduction in dephasing-induced excess leakage. The two correction tones further suppress residual leakage through the $|2\rangle$ and $|3\rangle$ channels, lowering the coherent endpoint leakage from approximately $7\times10^{-7}$ to $3\times10^{-8}$ without increasing transient exposure. These results establish transient exposure and endpoint leakage as complementary targets for the design of fast transmon gates.
Figures
Figures from the paper (2 more)
Reference graph
Works this paper leans on
-
[1]
First tone The endpoint-cancellation pulse (ECP) is built from two tones, one near each leakage transition. With the PSP fixed, the first tone, near the|1⟩–|2⟩transition, is Ωaux(t) =A mraux ei(δauxt+ϕ) × i ˙Fθ δaux +W(t) b2 ¨Fθ η2 +ib 3 F (3) θ η3 .(20) 5 0 2 4 6 8 10 Time (ns) −0.2 −0.1 0.0 0.1 0.2 0.3 Ω(t) (rad/ns) (a) Path pulse — time domain cosine D...
-
[2]
path basin
Second tone In the rotating frame of Eq. (1) the level energies are En =−η n(n−1)/2, so the|1⟩–|2⟩gap isE 2 −E 1 =−η (the transition the first tone addresses) and the|2⟩–|3⟩ gap isE 3−E2 =−2η. The|3⟩component of the four-level endpoint floor (Sec. VI D) arises from a cascade through the transiently populated|2⟩state, with first-order end- point amplitudec...
2000
-
[3]
The decrease in|2⟩back-action and the increase in the|3⟩cascade explain the four-level knee used in the main text
F our-level endpoint floor and per-level decomposition The sweep overw path reveals a non-monotonic four- level endpoint floor. The decrease in|2⟩back-action and the increase in the|3⟩cascade explain the four-level knee used in the main text. Across this knee, the bare floor is nearly independent of the endpoint module: the main pulse, the single-tone ECP...
-
[4]
Figure 4 reports the per-level decomposition of the five-level endpoint floor, seed by seed
Five-level truncation verification To check that the second ECP tone does not merely push the leakage cascade from|3⟩into|4⟩, we evaluated the PESP-C design at the operating pointw path = 30 with a five-level model (n steps = 2400). Figure 4 reports the per-level decomposition of the five-level endpoint floor, seed by seed. The|4⟩population is∼6.5×10 −12 ...
-
[5]
Table VIII reports the five-seed median atn steps = 1200,2400,4800
Numerical convergence, peak-leakage sensitivity , and direct four-level sanity check The two-tone operating-point result is stable against time-step refinement. Table VIII reports the five-seed median atn steps = 1200,2400,4800. The endpoint floor changes by less than 0.1% from 2400 to 4800 steps. As a separate sanity check, Table IX reports one direct fo...
-
[6]
Peak drive amplitude Path shaping does not increase the peak drive strength. In the three-level proxy at the operating point, the peak Rabi rate of the PSP is 0.259 rad ns −1 versus 0.313 rad ns−1 for the calibrated cosine-DRAG reference, an≈17% reduction that is identical across the five seeds, and adding the ECP tones changes the peak by less than 0.1%....
-
[7]
Amplitude and anharmonicity mismatch Table XI (top) and Fig. 5 report sensitivity to ampli- tude andηerrors, evaluated for the PESP-C design at the operating point (w path = 30; ideal 1−F= 3.0×10 −7, −20 −10 0 10 20 Amplitude error (\%) 10−6 10−5 10−4 10−3 10−2 1 − F (a) Amplitude miscal. → infidelity cosine DRAG FAST baseline PESP-C −20 −10 0 10 20 Ampli...
-
[8]
Sampling and filtering Table XI (bottom) reports the effect of zero-order- hold (ZOH) sampling and analog low-pass (LP) filter- ing for the PESP-C design at the operating point (ideal 1−F= 3.0×10 −7). Linear low-pass distortion is a lin- ear filter and is recoverable close to the ideal floor by Wiener predistortion (ϵ= 10 −3); raw, it costs 2.4×10 −4 at 0...
Show all 37 references
-
[9]
J. Koch, T. M. Yu, J. Gambetta, A. A. Houck, D. I. Schuster, J. Majer, A. Blais, M. H. Devoret, S. M. Girvin, and R. J. Schoelkopf, Charge-insensitive qubit design de- rived from the Cooper pair box, Phys. Rev. A76, 042319 (2007)
2007
-
[10]
Blais, A
A. Blais, A. L. Grimsmo, S. M. Girvin, and A. Wallraff, Circuit quantum electrodynamics, Rev. Mod. Phys.93, 025005 (2021)
2021
-
[11]
Krantz, M
P. Krantz, M. Kjaergaard, F. Yan, T. P. Orlando, S. Gus- tavsson, and W. D. Oliver, A quantum engineer’s guide to superconducting qubits, Appl. Phys. Rev.6, 021318 (2019)
2019
-
[12]
Kjaergaard, M
M. Kjaergaard, M. E. Schwartz, J. Braum¨ uller, P. Krantz, J. I.-J. Wang, S. Gustavsson, and W. D. Oliver, Superconducting qubits: Current state of play, Annu. Rev. Condens. Matter Phys.11, 369 (2020)
2020
-
[13]
Motzoi, J
F. Motzoi, J. M. Gambetta, P. Rebentrost, and F. K. Wil- helm, Simple pulses for elimination of leakage in weakly nonlinear qubits, Phys. Rev. Lett.103, 110501 (2009)
2009
-
[14]
Chenet al., Measuring and suppressing quantum state leakage in a superconducting qubit, Phys
Z. Chenet al., Measuring and suppressing quantum state leakage in a superconducting qubit, Phys. Rev. Lett.116, 020501 (2016)
2016
-
[15]
C. J. Wood and J. M. Gambetta, Quantification and characterization of leakage errors, Phys. Rev. A97, 032306 (2018)
2018
-
[16]
A. G. Fowler, M. Mariantoni, J. M. Martinis, and A. N. Cleland, Surface codes: Towards practical large-scale quantum computation, Phys. Rev. A86, 032324 (2012). 16
2012
-
[17]
McEwenet al., Removing leakage-induced correlated errors in superconducting quantum error correction, Nat
M. McEwenet al., Removing leakage-induced correlated errors in superconducting quantum error correction, Nat. Commun.12, 1761 (2021)
2021
-
[18]
K. C. Miaoet al., Overcoming leakage in quantum error correction, Nat. Phys.19, 1780 (2023)
2023
-
[19]
F. W. Strauch, Dephasing-induced leakage in multilevel superconducting quantum circuits, Phys. Rev. A112, 012601 (2025)
2025
-
[20]
J. M. Gambetta, F. Motzoi, S. T. Merkel, and F. K. Wil- helm, Analytic control methods for high-fidelity unitary operations in a weakly nonlinear oscillator, Phys. Rev. A 83, 012308 (2011)
2011
-
[21]
Lucero, J
E. Lucero, J. Kelly, R. C. Bialczak, M. Lenander, M. Mariantoni, M. Neeley, A. D. O’Connell, D. Sank, H. Wang, M. Weides, J. Wenner, T. Yamamoto, A. N. Cleland, and J. M. Martinis, Reduced phase error through optimized control of a superconducting qubit, Phys. Rev. A82, 042339 (2010)
2010
-
[22]
Hyypp¨ aet al., Reducing leakage of single-qubit gates for superconducting quantum processors using analytical control pulse envelopes, PRX Quantum5, 030353 (2024)
E. Hyypp¨ aet al., Reducing leakage of single-qubit gates for superconducting quantum processors using analytical control pulse envelopes, PRX Quantum5, 030353 (2024)
2024
-
[23]
Chiaro and Y
B. Chiaro and Y. Zhang, Active leakage cancellation in single qubit gates, Phys. Rev. Lett.135, 130601 (2025)
2025
-
[24]
Khaneja, T
N. Khaneja, T. Reiss, C. Kehlet, T. Schulte-Herbr¨ uggen, and S. J. Glaser, Optimal control of coupled spin dynam- ics: design of NMR pulse sequences by gradient ascent algorithms, J. Magn. Reson.172, 296 (2005)
2005
-
[25]
Werninghaus, D
M. Werninghaus, D. J. Egger, F. Roy, S. Machnes, F. K. Wilhelm, and S. Filipp, Leakage reduction in fast super- conducting qubit gates via optimal control, npj Quantum Inf.7, 14 (2021)
2021
-
[26]
Caneva, T
T. Caneva, T. Calarco, and S. Montangero, Chopped random-basis quantum optimization, Phys. Rev. A84, 022326 (2011)
2011
-
[27]
Glaseret al., Closed-loop optimization for high-fidelity controlled-Zgates in superconducting qubits, Phys
N. Glaseret al., Closed-loop optimization for high-fidelity controlled-Zgates in superconducting qubits, Phys. Rev. Applied24, 024048 (2025)
2025
-
[28]
H. N. Nguyen, F. Motzoi, M. Metcalf, K. B. Wha- ley, M. Bukov, and M. Schmitt, Reinforcement learning pulses for transmon qubit entangling gates, Mach. Learn.: Sci. Technol.5, 025066 (2024)
2024
-
[29]
D. C. McKay, C. J. Wood, S. Sheldon, J. M. Chow, and J. M. Gambetta, EfficientZgates for quantum comput- ing, Phys. Rev. A96, 022330 (2017)
2017
-
[30]
Wanget al., Suppressing spurious transitions using spectrally balanced pulse, Phys
R. Wanget al., Suppressing spurious transitions using spectrally balanced pulse, Phys. Rev. Lett.135, 160804 (2025)
2025
-
[31]
B. Li, T. Calarco, and F. Motzoi, Experimental error suppression in cross-resonance gates via multi-derivative pulse shaping, npj Quantum Inf.10, 66 (2024)
2024
-
[32]
Cywi´ nski, R
L. Cywi´ nski, R. M. Lutchyn, C. P. Nave, and S. Das Sarma, How to enhance dephasing time in superconduct- ing qubits, Phys. Rev. B77, 174509 (2008)
2008
-
[33]
T. J. Green, J. Sastrawan, H. Uys, and M. J. Biercuk, Arbitrary quantum control of qubits in the presence of universal noise, New J. Phys.15, 095004 (2013)
2013
-
[34]
G. A. Paz-Silva and L. Viola, General transfer-function approach to noise filtering in open-loop quantum control, Phys. Rev. Lett.113, 250501 (2014)
2014
-
[35]
Y. Oda, D. Lucarelli, K. Schultz, B. D. Clader, and G. Quiroz, Optimally band-limited noise filtering for single-qubit gates, Phys. Rev. Applied19, 014062 (2023)
2023
-
[36]
P. M. Poggi and A. Kiely, Suppressing leakage and main- taining robustness in transmon qubits: Signatures of a trade-off relation, Control Eng. Pract.172, 106855 (2026)
2026
-
[37]
J. J. McCord, M. Kuzmanovi´ c, and G. S. Paraoanu, Pareto-optimality of pulses for robust population trans- fer in a ladder-type qutrit, EPJ Quantum Technol.12, 121 (2025)
2025
Reviewed August 2, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.