{"id":"3467bbdb-c310-402a-af4e-bc120363397d","arxiv_id":"2412.16851","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Time-suspension pulse sequences outperform spectroscopic ones for dipolar decoupling in dense spin solids with local magnetic-field disorder, as shown by NMR experiments and numerical simulations.","lead":"Dense spin ensembles in solids are promising for quantum sensing and simulation, but the unavoidable magnetic dipolar interactions between spins destroy quantum information quickly. This paper benchmarks seven radiofrequency pulse sequences that suppress those interactions and shows that time-suspension sequences maintain coherence far better than spectroscopic sequences when the magnetic field is not perfectly uniform.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that local disorder is the distinguishing factor rests on a two-magnet/two-sample comparison; unmeasured phase transients or RF inhomogeneity on the 7 T setup could produce the same BR24-vs-CORY48 gap, and the paper explicitly states the frame-change correction was not applied.","rationale":"The reader and I identify the same load-bearing assumption: the cross-magnet/cross-sample comparison does not control for phase transients and B1 inhomogeneity. The paper's own simulations (Section IV B, Figure 8) show these errors affect sequences unequally, and the text states the frame-change technique was not applied, so phase transients are uncompensated. The liquid crystal control also changes sample type, making it an imperfect control. This is not a fatal flaw; it is testable, and the paper is otherwise careful and internally consistent, so CONDITIONAL remains the appropriate verdict. I considered the fidelity-metric issue raised in the reader's rationale but found it less load-bearing: local disorder causes real ensemble dephasing, which both the trace fidelity and the experimental T2,eff capture, and the experimental fits already remove the coherent global-offset oscillation. No ad hominem; the gap is one of missing control-error characterization, not of internal contradiction or fraud.","tokens_in":23280,"tokens_out":5532,"duration_ms":52003,"concrete_test":"Measure alpha_l and alpha_tr (phase-transient amplitudes) and B1 inhomogeneity on the 7 T adamantane setup, e.g., using the frame-change calibration of Ref. [54] or nutation/spectroscopy, then run the Section IV B simulations with these measured error levels for BR24 and CORY48 at tau = 4 microseconds. If the simulated BR24-to-CORY48 fidelity ratio reproduces the observed order-of-magnitude T2,eff gap, the disorder attribution is not established; if it does not, the attribution is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Central claim: local disorder is a distinguishing factor between spectroscopic and time-suspension sequences (abstract; Section IV A). The experimental support is the adamantane/7 T data vs the liquid-crystal/9.4 T control (Figure 7). This comparison changes both the sample and the magnet. The paper never reports phase-transient amplitudes (alpha_l, alpha_tr), RF B1 inhomogeneity, or pulse-calibration residuals for either setup, even though Section IV B and Figure 8 show these errors degrade sequences unequally, and Section IV B explicitly states the frame-change technique of Ref. [54] was not applied. If the 7 T probe/electronics had larger phase transients or B1 gradients, those errors, not static disorder, could account for the order-of-magnitude gap. The liquid-crystal control also differs in spin system: it is a nematic with only intramolecular couplings (Section IV A), not a 3D dipolar solid, so it does not isolate the role of static-field inhomogeneity.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper benchmarks seven dipolar decoupling pulse sequences—WHH, MREV8, MREV16, BR24, CORY48, YXX24, and YXX48—using 8-spin numerical simulations and 1H NMR experiments on adamantane in an unshimmed 7 T magnet. It reports that time-suspension sequences (CORY48 and the reinforcement-learning-discovered YXX sequences) preserve coherence for tens of milliseconds, whereas spectroscopic sequences decay in about a millisecond under the same conditions, and it attributes this gap to local static-field disorder because spectroscopic sequences preserve S_z-type terms. The attribution is supported by numerical disorder simulations and by a control experiment on a liquid crystal in a shimmed 9.4 T magnet, where BR24 and CORY48 perform similarly. The paper also extends the comparison to resonance-offset sensing and to protection of multiple-quantum coherences, where CORY48 increases the decay time by up to two orders of magnitude.","tokens_in":23492,"tokens_out":4304,"duration_ms":42013,"significance":"If the central attribution holds, the paper gives a practically useful rule for quantum simulation and sensing in dense spin ensembles: under inhomogeneous static fields, time-suspension sequences are preferable to spectroscopic sequences, and machine-learned sequences such as YXX24/YXX48 are competitive with the CORY48 gold standard. The strength of the paper is its systematic experimental benchmark with raw decay curves, full pulse-sequence tables, F-matrix representations, and a direct comparison with Average Hamiltonian Theory. The numerical simulations are straightforward and reproducible, and the liquid-crystal control experiment is a genuinely falsifiable check of the disorder hypothesis. However, the headline causal claim is currently under-supported because the experimental comparison changes both the sample and the magnet without quantifying other control errors that the paper's own simulations show can affect sequences unequally.","major_comments":[{"comment":"The central claim that local disorder is the distinguishing factor between spectroscopic and time-suspension sequences is not fully established by the experimental comparison. The adamantane/7 T data are compared with a liquid-crystal/9.4 T control in Fig. 7, but this changes not only the static-field homogeneity but also the spin system (a 3D dipolar solid vs. a nematic with only intramolecular couplings, as stated in §IV A) and the observable (Cavg for adamantane vs. Z autocorrelation for the liquid crystal). The paper does not report phase-transient amplitudes, RF B1 inhomogeneity, or pulse-calibration residuals for either setup, and §IV B explicitly states that the frame-change correction of Ref. [54] was not applied. Since the simulations in Fig. 8 show that rotation errors and phase transients degrade different sequences unequally, setup-specific control errors could in principle account for the order-of-magnitude gap in Fig. 1(b). Please either add control-error characterization for both magnets or reword the causal attribution to 'consistent with local disorder' rather than 'distinguishing factor.'","section":"§IV A, Fig. 7, abstract"},{"comment":"The quantitative claim of 'order of magnitude better coherence time' rests on fits to stretched exponentials with no reported uncertainties or parameter values. Equation (6) has five free parameters (C0, f, g, T2,eff, C1), and the fits constrain g to a range rather than reporting its fitted value for each condition. The qualitative gap between sequence classes is clearly visible in the raw decay curves, but the numerical T2,eff comparisons in Fig. 4(e)–(f) and Fig. 5(a)–(b) should report fit uncertainties and goodness-of-fit information so that the reader can assess whether the apparent differences are statistically significant.","section":"§III B, Eqs. (5)–(6), Fig. 4(e)–(f)"}],"minor_comments":[{"comment":"The fidelity metric F = Tr(U†th Uexp^{1/M}) is not normalized; as written, the trace for a perfect unitary is 2^N, so 1−F would be negative for N>0. The plots show values between 0 and 1, so a normalized fidelity is clearly intended; please state the normalization explicitly.","section":"§III A"},{"comment":"The text says the liquid-crystal field inhomogeneity is 'of the order of 30 Hz' and that the individual NMR peaks are about 30 Hz wide, but the Fig. 6 caption states the green shaded region represents 10 Hz. Please reconcile these numbers.","section":"§IV A, Fig. 6 caption"},{"comment":"The spectroscopic fit is written as Cavg,spectro = C0 cos(2πft) + C1, which omits the stretched-exponential decay factor e^{-t^g/T2,eff} that appears in Eq. (6); this is presumably a typographical omission.","section":"Appendix B 3"},{"comment":"The caption contains a duplicated panel label '(c) CORY48(c) CORY48(c) CORY48'; please clean up the caption and ensure each panel is labeled once.","section":"Fig. 3 caption"},{"comment":"The text says maximum dipolar coupling in adamantane is approximately 420 Hz, while the simulations in §III A use 3σ = 5000 Hz as the maximum coupling. This mismatch is acknowledged as a mid-range choice, but a brief justification of why a 5000 Hz distribution is representative for the experimental comparison would help the reader connect the numerics to the adamantane data.","section":"§III B"}],"recommendation":"major_revision","confidential_remarks":"The paper is experimentally solid and well within the scope of quant-ph, but the headline attribution of the sequence-performance gap to local disorder needs either additional control-error characterization on both magnets or a more modest wording. I do not see grounds for rejection; the required changes are local and feasible."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a genuinely useful experimental benchmark: seven dipolar decoupling sequences (WHH, MREV8/16, BR24, CORY48, YXX24/48) run side by side on adamantane in an inhomogeneous 7 T magnet, with the clear result that time-suspension sequences hold coherence for tens of milliseconds while spectroscopic sequences die in about a millisecond. Second, the paper's interpretive claim — that local disorder is what separates the two classes — is plausible but not airtight, because the supporting control experiment changes both the sample and the magnet (liquid crystal in a shimmed 9.4 T vs adamantane in unshimmed 7 T). The reader's concern about unmeasured phase transients and RF inhomogeneity is legitimate, and the paper itself admits the frame-change correction of Ref [54] was not applied.\n\nWhat's new: the head-to-head is new; previous reports tend to test one or two sequences. The liquid-crystal control, while imperfect, is a smart idea. The numerical error-scaling analysis (disorder, rotation errors, phase transients) is coherent and consistent with the AHT table. The MQC protection data with CORY48 (two orders of magnitude extension of decay time) is a nice demonstration.\n\nSoft spots, in order. (1) The disorder attribution rests on the two-magnet/two-sample comparison; the paper never reports phase-transient amplitudes or B1 inhomogeneity for either setup, though its own simulations show these degrade sequences unequally. This is a significant gap, but fixable: measure or bound those errors, or run adamantane on a shimmed magnet. (2) The reported T2,eff values come from five-parameter stretched-exponential fits with no error bars, and no code/data are provided. For a benchmarking paper, data availability matters. (3) The numerical fidelity metric sets Uth to identity, which penalizes spectroscopic sequences for doing exactly what they are designed to do — preserve offset-derived terms. The experimental Cavg metric is fairer; the simulations should be read as illustrating error scalings, not as the primary evidence. (4) The paper itself flags that absolute coherence times are below literature highs; that is honest but means the benchmark is relative, not state-of-the-art.\n\nBottom line: this paper deserves a serious referee. A reviewer should push for error bars, data/code, and a cleaner isolation of disorder. The central experimental observation will survive, but the interpretation needs tightening. I'd bring it to a reading group for the NMR/quantum-sensing crowd, and I'd cite the head-to-head numbers.","headline":"A useful head-to-head experimental benchmark of seven dipolar decoupling sequences, with a plausible but not airtight case that local disorder explains why time-suspension beats spectroscopic sequences.","tokens_in":24034,"tokens_out":1959,"would_cite":true,"duration_ms":17793,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"In dense spin solids with inhomogeneous fields, time-suspension pulse sequences hold coherence for tens of milliseconds while spectroscopic sequences decay in about a millisecond, and the gap is traced to local disorder.","keywords":["dipolar decoupling","Average Hamiltonian Theory","time-suspension sequences","spectroscopic sequences","local disorder","solid-state NMR","multiple-pulse control","quantum coherence"],"falsifier":"Repeat the adamantane decoupling experiment on a well-shimmed magnet with matched pulse calibration: if BR24 remains an order of magnitude worse than CORY48, the local-disorder explanation fails, while if the gap disappears when the 244 Hz inhomogeneity is removed, the claim is supported. A complementary check is to measure phase-transient amplitudes and $B_1$ inhomogeneity on both magnets and confirm in simulation that they cannot account for the observed gap.","tokens_in":23052,"feed_emoji":"🧲","tokens_out":10060,"duration_ms":77785,"temperature":0.7,"pith_summary":"The paper benchmarks seven multiple-pulse sequences that aim to suppress magnetic dipolar couplings in dense spin solids, combining numerical simulations on small spin systems with solid-state NMR experiments on adamantane. Its central claim is that in an inhomogeneous static field, time-suspension sequences (which refocus all internal-Hamiltonian terms) preserve spin coherence for tens of milliseconds, while spectroscopic sequences (which keep terms proportional to $S_z$) decay in about one millisecond. The authors trace this gap to local disorder: spectroscopic sequences preserve the very $S_z$ terms that carry the disorder, whereas time-suspension sequences average those terms away. A liquid-crystal experiment in a well-shimmed magnet, where disorder is small, shows the two classes performing comparably, which supports the attribution. If correct, the result gives quantum-information experiments a concrete rule: in dense ensembles with inhomogeneous fields, use time-suspension sequences such as CORY48.","feed_headline":"Time-suspension pulses stretch dense-spin coherence to ~56 ms","feed_subtitle":"Spectroscopic sequences die in ~1 ms; time-suspension refocusing keeps dense-spin coherence for tens of ms.","key_machinery":"The load-bearing mechanism is Average Hamiltonian Theory with the Magnus expansion, which represents the stroboscopic effect of a periodic pulse train as an effective Hamiltonian $\\bar H^{(0)}+\\bar H^{(1)}+\\dots$. The paper classifies each sequence by which terms in this expansion vanish: spectroscopic sequences (WHH, MREV8, MREV16, BR24) cancel low-order dipolar terms but leave a rescaled term $\\Delta_{\\text{SF}}\\sum_i a_i S_{z'}$, with $a_i=\\delta_i+h_i+\\Delta\\omega$, while time-suspension sequences (CORY48, YXX24, YXX48) leave $\\bar H^{(0)}=0$ and cancel the stated higher-order terms as well. The quantitative comparisons are carried by a trace fidelity $F=\\operatorname{Tr}(U^\\dagger_{\\text{th}} U_{\\text{exp}}^{1/M})$ for 8-spin simulations and by the geometric-mean autocorrelation $C_{\\text{avg}}=(C_{xx}C_{yy}C_{zz})^{1/3}$ in experiments, which approximates state fidelity. The local-disorder simulations then act as the discriminator: they introduce a Gaussian spread of offsets and show the signature $\\sigma_h^2$ growth of infidelity that identifies the preserved disorder term as the cause of spectroscopic-sequence failure.","core_discovery":"At its core, the paper argues that the practical quality of a decoupling sequence in a dense spin solid is governed less by how many orders of the dipolar Hamiltonian it cancels and more by whether it preserves the local disordered Zeeman term. In experiments on adamantane in an unshimmed 7 T magnet, CORY48 achieves effective coherence times up to 56.6 ms at an inter-pulse delay of 5.2 µs, while spectroscopic sequences WHH and BR24 remain below about 1 ms under the same conditions. Numerical simulations show that when local offsets are drawn from a Gaussian distribution with standard deviation near the experimental inhomogeneity, spectroscopic-sequence infidelity grows as $\\sigma_h^2$, while time-suspension sequences are affected far less at the same disorder strengths; this scaling matches the algebraic fact that spectroscopic sequences have a nonzero average Hamiltonian $\\bar H^{(0)} \\propto \\sum_i a_i S_{z'}$ with $a_i=\\delta_i+h_i+\\Delta\\omega$. The paper also demonstrates that CORY48 can protect multiple-quantum correlations, extending their decay times by up to two orders of magnitude relative to free dipolar evolution, and that the machine-learned YXX24 and YXX48 sequences perform comparably to CORY48.","pith_inferences":["Beyond the paper: a decisive controlled test would be to sweep the inhomogeneous broadening on a single sample (for example with gradients or susceptibility inserts) and verify that the CORY48-versus-BR24 gap widens monotonically with disorder; the paper only compares two different magnets with different samples.","Beyond the paper: the same spectroscopic-versus-time-suspension distinction should appear in electron-spin ensembles such as nitrogen-vacancy centers, where g-factor and strain variations create an analogous local disorder term, so time-suspension sequences should be tested there for quantum sensing and memory applications.","Beyond the paper: the slow convergence of the Magnus expansion reported for WHH suggests that for strongly coupled systems, Hamiltonian-engineering design may need to move beyond truncation-based analytical methods; the paper raises this as an open question rather than a demonstrated result.","Beyond the paper: because chemical shifts and local disorder have identical symmetry, the result implies an unavoidable trade-off between spectroscopy and coherence in dense solids; any sequence that learns the chemical-shift term will also preserve the disorder that kills coherence."],"forward_implications":["Quantum-information experiments in dense dipolar solids with inhomogeneous static fields should default to time-suspension sequences such as CORY48, since spectroscopic sequences cap coherence near one millisecond.","The machine-learned YXX24 and YXX48 sequences perform comparably to CORY48, indicating that sequences discovered by automated search can serve as practical alternatives to analytically designed ones.","Spectroscopic sequences remain the right tool for measuring resonance offsets or chemical shifts, but in the presence of local disorder they cannot simultaneously preserve quantum coherence.","Time-suspension decoupling can extend the lifetime of multi-spin correlated states by up to two orders of magnitude, making many-body correlation measurements feasible on longer timescales.","Sequence ranking is device-dependent: the best sequence for a given experiment is the one matched to that platform's disorder, pulse-width, rotation-error, and phase-transient budget."],"supporting_citations":[{"why":"Supplies the Average Hamiltonian Theory framework used to classify and compare all sequences.","marker":"[22]"},{"why":"Introduces the WHH sequence, the shortest cyclic sequence that cancels zeroth-order dipolar coupling.","marker":"[18]"},{"why":"Introduces BR24, the highest-order spectroscopic sequence used as the main spectroscopic benchmark.","marker":"[52]"},{"why":"Defines CORY48 as the canonical time-suspension sequence and the gold standard for decoupling.","marker":"[53]"},{"why":"Provides the machine-learned YXX24/YXX48 sequences and the Cavg fidelity approximation used in experiments.","marker":"[40]"},{"why":"Supplies the frame-matrix representation used to analyze toggling-frame orientations and sequence design rules.","marker":"[36]"},{"why":"The reference for selective averaging and offset effects, used to explain second-averaging behavior of spectroscopic sequences.","marker":"[49]"},{"why":"Earlier demonstration of decoupling protection of highly correlated spin states; the paper's MQC results are compared with it.","marker":"[63]"}],"fun_headline_variants":["CORY48 keeps dense-spin coherence 50x longer than spectroscopic pulses","Preserving local disorder is the real key to spin decoupling","Spectroscopic sequences die at 1 ms; CORY48 survives to 56 ms","Time-suspension pulses quench dipolar decoherence in dense spins"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusion that local disorder, not some other experimental difference, drives the performance gap assumes that the unshimmed 7 T and shimmed 9.4 T setups differ only in static-field homogeneity; the paper does not report phase-transient amplitudes, radiofrequency inhomogeneity, or pulse-calibration residuals for either setup.","fun_headline_variants_meta":{"raw":{"variants":["CORY48 keeps dense-spin coherence 50x longer than spectroscopic pulses","Preserving local disorder is the real key to spin decoupling","Spectroscopic sequences die at 1 ms; CORY48 survives to 56 ms","Time-suspension pulses quench dipolar decoherence in dense spins"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000409,"raw_usage":{"total_tokens":2167,"prompt_tokens":1033,"completion_tokens":1134,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":649,"completion_tokens_details":{"reasoning_tokens":1052}},"tokens_in":649,"tokens_out":1134,"duration_ms":8010,"temperature":1.0,"reasoning_tokens":1052,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T10:15:36.902735+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Repeat the adamantane decoupling experiment on a well-shimmed magnet with matched pulse calibration: if BR24 remains an order of magnitude worse than CORY48, the local-disorder explanation fails, while if the gap disappears when the 244 Hz inhomogeneity is removed, the claim is supported. A complementary check is to measure phase-transient amplitudes and $B_1$ inhomogeneity on both magnets and confirm in simulation that they cannot account for the observed gap.","supporting_citations":[{"cited_title":"Mehring and J","cited_arxiv_id":null,"evidence_quote":"Supplies the Average Hamiltonian Theory framework used to classify and compare all sequences."},{"cited_title":"Cappellaro and M","cited_arxiv_id":null,"evidence_quote":"Introduces the WHH sequence, the shortest cyclic sequence that cancels zeroth-order dipolar coupling."},{"cited_title":"Magnus, On the exponential solution of differential equations for a linear operator, Communications on Pure and Applied Mathematics 7, 649 (1954)","cited_arxiv_id":null,"evidence_quote":"Introduces BR24, the highest-order spectroscopic sequence used as the main spectroscopic benchmark."},{"cited_title":"Haeberlen, Line Narrowing by Multiple Pulse Tech- niques III","cited_arxiv_id":null,"evidence_quote":"Defines CORY48 as the canonical time-suspension sequence and the gold standard for decoupling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the machine-learned YXX24/YXX48 sequences and the Cavg fidelity approximation used in experiments."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the frame-matrix representation used to analyze toggling-frame orientations and sequence design rules."},{"cited_title":"Ernst, A","cited_arxiv_id":null,"evidence_quote":"Earlier demonstration of decoupling protection of highly correlated spin states; the paper's MQC results are compared with it."}],"review_version":1}