{"id":"249c6090-4e26-4f34-9ead-383959c1ea72","arxiv_id":"2412.17789","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Adding one or two linear constraints on the Fourier coefficients of an amplitude-modulated Mølmer-Sørensen pulse improves the leading-order gate-timing error from O(Δt²) to O(Δt⁶) or O(Δt¹⁰) with vanishing power overhead.","lead":"This paper shows how to shape the laser pulse of a trapped-ion two-qubit gate so that a small error in the gate timing barely lowers the gate fidelity. The method uses a Fourier-series pulse envelope and needs only one or two additional linear constraints, with almost no extra laser power when enough Fourier terms are included.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'power-optimized' claim depends on an unproven restriction to b_n=0; a b_n≠0 solution with higher phase per power would make the reported power overheads underestimates.","rationale":"The scaling result itself is well supported: the linear constraints, the reduction in Appendix C, and the Rayleigh-quotient optimization are analytically sound, and the numerical results reproduce the claimed exponents. The one place where the paper overclaims is the global power optimality of the b_n=0 ansatz. Appendix D supplies a useful bound (b_n contribution to A/P ≤ 1/3) but does not prove that the a-only optimum exceeds this bound for all N shown. The proposed eigenvalue test settles this directly: if the full-space optimum coincides with the a-only optimum, the concern is resolved; if not, the power-overhead numbers need revision. Other flagged issues (Eq. (14) typo, single-mode idealization) are real but do not threaten the central asymptotic claim. Since the reader's CONDITIONAL verdict already captures this checkable uncertainty, the verdict remains unchanged.","tokens_in":17338,"tokens_out":32958,"duration_ms":298346,"concrete_test":"Re-solve the generalized eigenvalue problem (35) on the full coefficient space (a0, a_n≥2, b_n≥2), imposing the even-i b-constraints (Σ b_n n^(i-1)=0 for i=2,4) and the a-constraints used for 1 LC and 2 LC, for N=5,10,20,100. Compare the maximum |A|/P eigenvalue with the a-only value from §IV and Appendix E. If any full-space eigenvalue exceeds the a-only value, recompute Fig. 8 including the optimal b_n; if all are equal (expected, because a-only ratios exceed 1/3), the b_n=0 ansatz is certified and the conditional on the power-optimal claim can be removed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central O(Δt^6)/O(Δt^10) scaling derivation is internally consistent; the fidelity expansion, linear constraints (28)-(29), and the Rayleigh-quotient optimization are all sound within the stated single-mode model. The load-bearing weakness is the assertion in §III.D and Appendix D that the sine coefficients b_n can be set to zero without loss of optimality. The argument is heuristic: nonzero b_n add positive contributions to both P and A, with per-power phase efficiency at most 1/3 since 1/(n^2-1)≤1/3 for n≥2, while the a0 term has efficiency 1. But the paper never proves that the optimized a-only ratio exceeds 1/3 for every N reported in Fig. 8; if for some N it did not, a b_n≠0 solution satisfying the even-derivative constraints (e.g., Σ b_n n=0 for O(Δt^6)) could yield a larger |A|/P, lowering the true minimum power. This would not change the scaling exponents but would invalidate the 'minimum-power pulse' and 'power-optimized' wording, and the overhead percentages would be upper bounds rather than minima.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper analyzes amplitude-modulated Mølmer-Sørensen gates for two trapped ions, expanding the laser amplitude envelope in a Fourier series. It derives closed-form expressions for the phase-space trajectory, converts the requirement of vanishing low-order derivatives of F and G at the gate time into linear constraints on the Fourier coefficients, and selects pulses by maximizing the geometric phase per unit average power subject to those constraints via a generalized eigenvalue problem. Numerical results show infidelity scaling O(Δt^6) with one constraint and O(Δt^10) with two constraints, with laser-power overhead that decreases as the number of Fourier components increases.","tokens_in":17561,"tokens_out":19013,"duration_ms":165616,"significance":"The central scaling result is an analytical, parameter-free prediction within the stated single-mode model: the constraints are linear, the optimization is a Rayleigh quotient, and the asymptotic improvements in Δt are verified numerically rather than inferred from fits. This is a useful and clean demonstration that amplitude-modulated MS gates can suppress gate-timing errors with vanishing power overhead as N grows. The main limitation is that the 'power-optimized' claim is proved only for the cosine-only subspace; the global optimality over all Fourier coefficients is asserted rather than proven. With that point repaired or appropriately qualified, the paper would be a solid contribution to pulse-shaping methods for trapped-ion gates.","major_comments":[{"comment":"The claim that the optimized pulse is 'minimum-power' (abstract and Section III.D) is not established for the full Fourier space. Appendix D argues that b_n should be zero because nonzero b_n add positive contributions to both P and A, and because a b-only solution has |A|/P ≤ 1/3. This is heuristic: it does not rule out a mixed a_n,b_n solution with larger |A|/P, and the '≤1/3' bound is not compared with the actual optimized a-only ratio for each N reported in Fig. 8. The missing step is a convexity argument: for the optimized a-only solution with negative A_a, any nonzero b vector adds A_b>0 and P_b>0, which strictly lowers |A_a|/P_a; if such a proof is supplied (or an explicit check that the chosen eigenvector has A_a<0 for the relevant N), the restriction is justified. Without it, the wording should be softened to 'optimal within the cosine-only subspace,' and the overhead percentages in Fig. 8 are upper bounds rather than minima.","section":"Section III.D and Appendix D"}],"minor_comments":[{"comment":"Equation (14) is inconsistent with Appendix A: using periodicity of F and g, A(T+Δt)=A(T)+A(Δt)=π/2+A(Δt), so ΔA=A(Δt), not A(Δt)-π/2. Appendix A in fact uses the correct relation. Please correct Eq. (14) and the sentence that refers to it.","section":"Eq. (14) and Appendix A"},{"comment":"The coefficient 1/128 on the dropped term in Eq. (A1) does not follow from the expansion of Eq. (11) as displayed, and the accompanying statement 'A=O(Δt²)' is not correct for the MS pulse (one finds ΔA=O(Δt³) for the constant-amplitude case). Please recompute the dropped term and its order; the leading-order term in Eq. (15) is unaffected.","section":"Appendix A, Eq. (A1)"},{"comment":"The phrase 'one linear constraint' should be qualified. If the b_n are retained, the reduced system for the first nontrivial order contains both a0/2+Σ a_n=0 and Σ b_n n=0; the latter is automatic only in the cosine-only subspace. Please make the counting explicit.","section":"Section III.C and abstract"},{"comment":"The caption should state the normalization of the plotted power (apparently P=1 for the unmodulated MS pulse) and define 'additional power' as (P-1)×100%.","section":"Fig. 8"}],"recommendation":"major_revision","confidential_remarks":"To the editor: the manuscript is a clean analytical/numerical study, but the global power-optimality claim needs either a proof or a qualification before publication. The authors should also be asked to clarify the relationship to Ref. [13], which already contains a provably power-optimal construction for multimode gates; the present paper's contribution is the simple two-ion analytical scaling and the cosine-only optimization. If the authors supply the missing argument and correct the equation errors, I would support publication. I did not find any indication of circularity or fitted-parameter concerns."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the O(Δt^6) and O(Δt^10) scaling for timing errors is derived cleanly and I believe it holds within the stated single-mode model. The power-optimized label is not fully earned: the paper restricts to the cosine (a_n) subspace without a proof that sine coefficients can't do better, though my own spot-check suggests they likely can't in the parameter range shown.\n\nWhat's genuinely new: the reduction of timing-error robustness to a compact set of linear constraints on the Fourier coefficients of an AM-only pulse, and the numerical demonstration that the extra power cost tends to zero as the number of coefficients grows. The derivation in Appendix C is straightforward and correct, and the generalized-eigenvalue optimization is standard but neatly executed. The plots in Fig. 7 clearly show the improved scaling.\n\nSoft spots, in order of importance. First, the b_n=0 argument in Appendix D is heuristic. They note that a sine-only solution has geometric phase per power at most 1/3, and they suggest the optimized cosine solution beats that, but they never verify it for every N. For 2 LC with N=4, for example, the cosine-only ratio is about 0.36, above 1/3, so it's plausibly fine; but for N=3 with 2 LC the cosine ratio is well below 1/3, though there are no nontrivial sine solutions there. The missing proof matters because the title says 'power-optimized' and the reported overheads are minima only if the restriction is optimal. If sine coefficients can help, those numbers are upper bounds. This should be either proven or the language softened to 'power-optimized within the cosine subspace.'\n\nSecond, Eq. (14) misdefines ΔA: from the periodicity, A(T+Δt) = π/2 + A(Δt), so ΔA should be A(Δt), not A(Δt)-π/2. The text and Appendix A effectively use the correct definition, so it's a typo, but it's confusing. Appendix A also has a wrong coefficient (and sign) on the squared term in the fidelity expansion; it's higher order and drops out, so the main argument survives.\n\nThird, the model is deliberately simple—single mode, no carrier—and the paper honestly says timing errors aren't the leading error source in real gates. So the practical significance is moderate; the value is in the clear analytical framework.\n\nWho this is for: anyone working on amplitude-modulated ion gates, especially the power-robustness trade-off. It deserves review; the core result is sound and useful. I'd recommend sending it to referees, with a request to tighten the power-optimality argument and fix the small errors.","headline":"Solid analytical derivation of timing-error robustness for Fourier-AM MS gates, with a power-optimality claim that needs tightening but is probably correct.","tokens_in":18077,"tokens_out":13352,"would_cite":true,"duration_ms":106718,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81P68"],"pacs":["03.67.Lx"],"model":"deepseek-v4-flash","headline":"This paper establishes that amplitude-modulated Mølmer–Sørensen gates can reduce leading-order gate-timing error from quadratic to sixth- or tenth-order scaling with only linear constraints on the Fourier coefficients and near-constant…","keywords":["trapped-ion quantum computing","Mølmer–Sørensen gate","amplitude-modulated pulse","Fourier pulse shaping","gate-timing error","power optimization","gate fidelity","linear constraints"],"falsifier":"Run the same constrained power optimization with $b_n$ allowed to be nonzero for a fixed $N$ (say $N=10$), subject to the same one- or two-constraint system, and compare the maximum attainable geometric phase per unit power with the $b_n=0$ result; any strictly larger ratio would disprove the claim that cosine-only pulses are power-optimal. Experimentally, one could measure infidelity versus gate-time offset for an $N=5$ or $N=10$ pulse and check whether the predicted $\\Delta t^6$ or $\\Delta t^{10}$ plateau appears over the claimed stability window.","tokens_in":17122,"feed_emoji":"⚛️","tokens_out":6850,"duration_ms":58754,"temperature":0.7,"pith_summary":"The paper studies amplitude-modulated Mølmer–Sørensen gates on two trapped ions, writing the laser amplitude as a truncated Fourier series. It shows that requiring the first few derivatives of the phase-space trajectory to vanish at the gate time reduces to simple linear moment conditions on the Fourier coefficients. With one such constraint the leading infidelity from a gate-timing error $\\Delta t$ drops from $O(\\Delta t^2)$ to $O(\\Delta t^6)$, and with two constraints to $O(\\Delta t^{10})$. The pulse satisfying those constraints can be chosen to minimize average laser power by solving a generalized eigenvalue problem, and the extra power cost shrinks toward zero as more Fourier terms are included. The paper's point is that robust, power-efficient entangling gates need only a modest Fourier basis.","feed_headline":"Pulse shaping cuts trapped-ion gate errors from Δt² to Δt¹⁰","feed_subtitle":"Fourier-amplitude MS gate adds one or two linear constraints; extra laser power falls below 1.2 percent.","key_machinery":"The machinery is the Fourier amplitude envelope $\\Omega(t)=a_0/2+\\sum_{n=1}^N [a_n\\cos(n\\xi_0 t)+b_n\\sin(n\\xi_0 t)]$ together with exact integral expressions for the phase-space couplings $F(t)$ and $G(t)$. Because every derivative $F^{(i)}(T)$ and $G^{(i)}(T)$ becomes a sum over $n^i$ times the Fourier coefficients, the robustness conditions are linear moment constraints on those coefficients. Average laser power and geometric phase are both quadratic forms in the coefficients, so after projecting out the linear constraints the minimum-power pulse is the eigenvector of $P^{-1/2}AP^{-1/2}$ with the largest absolute eigenvalue, with $A(T)$ fixed to $\\pi/2$ by rescaling.","core_discovery":"The central claim is that a Fourier-expanded amplitude envelope for the MS gate converts robustness to gate-timing errors into a short list of homogeneous linear equations in the Fourier coefficients, and that the minimal-power solution to those equations is found by diagonalizing one matrix. The paper derives closed-form amplitude-modulated trajectories $F(t)$ and $G(t)$, and shows that $F^{(i)}(T)=0$ and $G^{(i)}(T)=0$ collapse to the moment conditions $\\sum_n a_n n^{i-1}=0$ for odd $i$ and $\\sum_n b_n n^{i-1}=0$ for even $i$, with $a_1=b_1=0$ enforced by closure. Enforcing one constraint makes $1-F_{\\rm MS}$ scale as $\\Delta t^6$; enforcing two makes it scale as $\\Delta t^{10}$. The power-optimized pulses use only cosine coefficients ($b_n=0$), reducing the constrained optimization to the Rayleigh quotient of a pair of quadratic forms, and numerical results for $N$ up to 100 show the power overhead falling to $0.51\\%$ for one constraint and $1.2\\%$ for two.","pith_inferences":["If the $b_n=0$ ansatz is wrong, the reported power overheads are optimistic; a numerical search over the full coefficient space for fixed $N$ would settle this without new physics.","The two-ion single-mode model omits carrier transitions and higher Lamb-Dicke terms that the paper itself estimates contribute coherent errors near $10^{-4}$, so the asymptotic scaling may not dominate in a full multi-mode device.","An experimental test only needs an arbitrary waveform generator and a two-ion chain; measuring the width of the flat infidelity region for increasing $N$ would directly check whether the stability region narrows as predicted.","Because the method only reshapes the amplitude envelope, it can likely be stacked with frequency or phase modulation and with standard error-mitigation techniques, though the paper does not demonstrate such combinations."],"forward_implications":["One added linear constraint changes the leading infidelity from timing error from $O(\\Delta t^2)$ to $O(\\Delta t^6)$; two constraints change it to $O(\\Delta t^{10})$.","The linear constraints are homogeneous and can in principle be extended to arbitrarily high order, since each additional constraint removes four powers of $\\Delta t$ from the leading error.","The additional average laser power is small and shrinks with $N$: $0.51\\%$ overhead at $N=100$ for one constraint and $1.2\\%$ for two.","The amplitude-modulated pulses develop a soft start, which leaves smaller residual phase-space displacement and a wider region of near-unit fidelity around the nominal gate time.","The same strategy of expanding fidelity in a control parameter and imposing linear coefficient constraints can be applied to other error sources such as laser frequency fluctuations."],"supporting_citations":[{"why":"Provides the MS Hamiltonian, exact propagator, and ideal gate-time conditions F(T)=G(T)=0, A(T)=π/2 that define the baseline trajectory.","marker":"[7]"},{"why":"Demonstrates expanding the MS fidelity in a small parameter to derive robustness constraints, the technique this paper adapts to gate-timing errors.","marker":"[22]"},{"why":"Gives the power-optimal Fourier-based gate construction whose lower bound P≥π/2 and stabilized-gate framework the present optimization is compared against.","marker":"[13]"},{"why":"Introduces simultaneous amplitude and frequency modulation with a Fourier sine series, the closest prior scheme that this work simplifies to amplitude-only modulation.","marker":"[12]"},{"why":"Supplies the method for extremizing a ratio of quadratic forms subject to linear constraints, which reduces the pulse optimization to an eigenvalue problem.","marker":"[28]"},{"why":"Defines the average-power convention and optimal-control framing used when quoting power overheads against the MS gate.","marker":"[10]"},{"why":"Motivates soft-start amplitude envelopes and the residual spin-motion entanglement error that robustness constraints address.","marker":"[11]"}],"fun_headline_variants":["Two constraints boost trapped-ion gate timing robustness to Δt¹⁰","Fourier amplitude modulation achieves Δt¹⁰ error suppression with <1.2% power","Minimal power AM gate: timing error order Δt⁶ to Δt¹⁰ with linear constraints","One constraint lifts trapped-ion gate timing to Δt⁶, two to Δt¹⁰","AM pulse with one extra moment condition: Δt²→Δt⁶; two: Δt²→Δt¹⁰"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that zeroing the sine Fourier coefficients $b_n$ sacrifices no optimality in the ratio of geometric phase to average laser power; if a nonzero-$b_n$ solution under the same constraints beats the $b_n=0$ one, the reported power costs would be underestimates and the power-optimal claim would be unsupported.","fun_headline_variants_meta":{"raw":{"variants":["Two constraints boost trapped-ion gate timing robustness to Δt¹⁰","Fourier amplitude modulation achieves Δt¹⁰ error suppression with <1.2% power","Minimal power AM gate: timing error order Δt⁶ to Δt¹⁰ with linear constraints","One constraint lifts trapped-ion gate timing to Δt⁶, two to Δt¹⁰","AM pulse with one extra moment condition: Δt²→Δt⁶; two: Δt²→Δt¹⁰"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000838,"raw_usage":{"total_tokens":3726,"prompt_tokens":1090,"completion_tokens":2636,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":706,"completion_tokens_details":{"reasoning_tokens":2508}},"tokens_in":706,"tokens_out":2636,"duration_ms":17884,"temperature":1.0,"reasoning_tokens":2508,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:10:54.729219+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same constrained power optimization with $b_n$ allowed to be nonzero for a fixed $N$ (say $N=10$), subject to the same one- or two-constraint system, and compare the maximum attainable geometric phase per unit power with the $b_n=0$ result; any strictly larger ratio would disprove the claim that cosine-only pulses are power-optimal. Experimentally, one could measure infidelity versus gate-time offset for an $N=5$ or $N=10$ pulse and check whether the predicted $\\Delta t^6$ or $\\Delta t^{10}$ plateau appears over the claimed stability window.","supporting_citations":[{"cited_title":"Bl¨ umel, N","cited_arxiv_id":null,"evidence_quote":"Gives the power-optimal Fourier-based gate construction whose lower bound P≥π/2 and stabilized-gate framework the present optimization is compared against."},{"cited_title":"Bl¨ umel, N","cited_arxiv_id":null,"evidence_quote":"Introduces simultaneous amplitude and frequency modulation with a Fourier sine series, the closest prior scheme that this work simplifies to amplitude-only modulation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the method for extremizing a ratio of quadratic forms subject to linear constraints, which reduces the pulse optimization to an eigenvalue problem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the average-power convention and optimal-control framing used when quoting power overheads against the MS gate."},{"cited_title":"Zarantonello, H","cited_arxiv_id":null,"evidence_quote":"Motivates soft-start amplitude envelopes and the residual spin-motion entanglement error that robustness constraints address."}],"review_version":1}