{"id":"c256702d-e0a8-4f16-a8ec-c9a96595e0ec","arxiv_id":"2507.22768","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A numerical protocol shows that Yb(trensal) molecular spin qubit-qudit systems can violate generalized Bell inequalities under measured decoherence, with a proposed switchable molecular trimer for qudit-qudit CGLMP tests.","lead":"The authors simulate optimized pulse sequences showing that the Yb(trensal) molecule, a spin qubit coupled to a nuclear spin qudit, should violate generalized Bell inequalities even with measured decoherence. A smart generalist might care because this offers a concrete solid-state route to certify high-dimensional entanglement in molecular qudits.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The CGLMP trimer simulation may be too optimistic: it does not simulate the decoherence of the ancilla/switch during the full 15-pulse sequence, and the predicted violation at T2=5 us is a post-hoc amplitude-selected point that may not survive a complete Lindblad simulation with realistic switch…","rationale":"The reader's verdict is CONDITIONAL and its weakest_assumption correctly flags the exponential vs gaussian decoherence mismatch and residual couplings. I add a more specific and, in my view, more load-bearing concern: the CGLMP trimer numbers are not shown to come from a complete simulation of the entire protocol including switch dephasing and residual couplings. The paper's methodology in Appendix G does give fidelity values for the four measurement states (Table X) and I values (Tables XII-XIII), so the concern is not that the numbers are fabricated but that the central parameter robustness claim for the trimer rests on amplitude-selected points and on a switch whose dephasing is only stated to be minor because it is 'brought to a superposition only during the implementation of two-qudit gates'. The CHSH part of the manuscript is much stronger: Table II shows broad violation over T2e in [1,20] us with GRAPE pulses, and the CHSH violation at the measured T2e=2.4 us is robustly 2.34. Because the abstract's strongest practical claim is the Yb(trensal) CHSH violation, which is well supported within the stated Lindblad model, I do not recommend changing the reader's CONDITIONAL verdict; the concern is a reason to keep the trimer portion conditional, not to reject the paper. The most useful concrete check is a full self-consistent re-simulation of the trimer protocol, rather than an experimental or statistical check that the paper does not claim to have performed.","tokens_in":27813,"tokens_out":1930,"duration_ms":20828,"concrete_test":"Re-run the full CGLMP protocol for the Cr-Yb-Cr trimer as a single Lindblad simulation containing the ancilla with T2=1 us, the two qudits with T2 in {5,10,30} us, and the residual qudit-qudit couplings estimated from Eq. (10) with J12=J23=0 when the ancilla is in its ground state. Keep all 15 gates in the order of Appendix G and compute I from the four final density matrices without post-hoc per-T2 amplitude selection. If the I=2.61 value at T2=30 us falls below 2.61 or the T2=5 us point falls below 2, the claimed robustness of the trimer CGLMP violation is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing concern is in the two-qudit extension (Section 'Extension to a two-qudits dimer' and Appendix G, Tables VII, X, XII). The claimed CGLMP values, including the headline I=2.61 at T2=30 us and the marginal violation I=2.019 at T2=5 us, appear to be obtained from simulations in which the ancilla mediates the two-qudit controlled-Z gates but the readout fidelity tables (Table X) already show the largest errors in the measurement-state preparation, not in the entangling step. Even granting the pure-dephasing model, the paper does not demonstrate that the dephasing noise on the switch and all residual qudit-qudit couplings (Appendix G states the switch-off holds only 'strictly to first order, apart from possible residual couplings') are included self-consistently in the reported numbers. In particular, the reported I values come from amplitude choices that are optimized per T2 (Table XIII) and the T2=5 us entry at B1=20 G (2.058) is the only point above 2 in Table XII and is lower than the 9 G no-decoherence 2.74; this appears to be an artifact of a single amplitude scan rather than a robust parameter conclusion. The central abstract claim that inequalities are 'safely violated in a wide range of parameters' is therefore much more firmly established for the qubit-qudit CHSH case (Table II, violation >2.3) than for the CGLMP trimer prediction.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes and numerically simulates a protocol for testing generalized Bell inequalities in molecular spin systems. For a Yb(trensal) electro-nuclear system, treated as a qubit-qudit pair, the authors design a pulse sequence, use GRAPE-optimized unitaries to realize the four rotated measurement settings, and simulate the full protocol with a Lindblad pure-dephasing model using experimentally measured T2e = 2.4 us and T2n = 560 us. They report CHSH expectation values up to about 2.34, above the local bound of 2. For a hypothetical Cr-Yb-Cr trimer with two S = 3/2 qudits and a central spin-1/2 ancilla switch, they propose a 15-pulse sequence to prepare a maximally entangled two-qudit state and simulate CGLMP values up to I = 2.61 at T2 = 30 us and a marginal violation at T2 = 5 us. The central abstract claim is that the inequalities are 'safely violated in a wide range of parameters'.","tokens_in":28130,"tokens_out":5352,"duration_ms":63284,"significance":"If the results hold, this would be a valuable step toward using molecular nanomagnets for high-dimensional Bell tests, with a realistic experimental platform and concrete pulse sequences. The paper's strengths include the use of experimentally measured coherence times for Yb(trensal), detailed pulse tables in the appendices, an explicit construction of the Bell operator and optimal observables, and a clear separation between the established qubit-qudit part and the more speculative trimer proposal. The qubit-qudit CHSH simulation is internally consistent and appears technically sound. The trimer part, however, is a proposal based on a hypothetical Hamiltonian and parameter choices, and its claimed robustness depends on modeling assumptions that are not fully demonstrated.","major_comments":[{"comment":"The Lindblad simulations of the 15-pulse trimer sequence appear to include pure dephasing only on the two qudits, not on the ancilla switch. The text states that the ancilla T2 = 1 us has 'only a limited role' because it is in a superposition only during two-qudit gates, but no evidence is given that ancilla dephasing was actually included in the master equation. Since each controlled-Z gate in Eqs. (G7), (G8), and (G10) excites and de-excites the ancilla through conditional pi rotations, omitting its dephasing could inflate the fidelities in Table VII and the CGLMP values in Table XII. Please state explicitly which jump operators are used in the trimer simulation; if ancilla dephasing is omitted, rerun the simulations with it included and report the revised I values.","section":"Main text, 'Extension to a two-qudits dimer'; Appendix G"},{"comment":"The switch-off assumption is load-bearing: the text says the ancilla-mediated switch 'holds strictly to first order, apart from possible residual couplings', yet no quantitative estimate of the residual qudit-qudit coupling is given and no such term appears in the master equation. This assumption matters especially for the single-qudit rotations and the measurement unitaries, where Table X already shows state fidelities as low as 0.58 at T2 = 5 us. Please provide a numerical estimate of the effective residual coupling when the ancilla is in its ground state, or include it in the simulation and show that the reported CGLMP values are unchanged.","section":"Appendix G, 'Realization of two-qudit gates via ancilla-controlled interactions'"},{"comment":"The CGLMP violation at T2 = 5 us is marginal: with uniform amplitudes the only value above 2 is I = 2.058 at B1 = 20 G, and with per-state amplitude optimization the best value is I = 2.165. Given the unmodeled ancilla dephasing and residual couplings discussed above, the abstract's claim that the inequalities are 'safely violated in a wide range of parameters' is not established for the trimer part. Please either restrict the robustness claim to the qubit-qudit CHSH case or provide a conclusive simulation that includes all relevant decoherence channels and demonstrates the violation over a well-defined parameter region.","section":"Table XII, Table XIII, and Fig. 2 (lower panel)"}],"minor_comments":[{"comment":"There are typos in the text: 'Kocken-Speker' should be 'Kochen-Specker' and 'lager Hilbert space' should be 'larger Hilbert space'.","section":"Introduction"},{"comment":"The sentence 'the relevant jump operators in Eq. (6) are to S_z and I_z' should read 'are proportional to S_z and I_z' or should specify the actual operators.","section":"Eq. (6)"},{"comment":"The first column, labeled 'Amp State', is not defined in the text; it appears to list pulse amplitudes, but the values (0.004, 0.003, 0.002, 0.001, 0.0009) are not clearly connected to the amplitudes used elsewhere. Please add a definition or remove the column.","section":"Table X"},{"comment":"The measured nuclear coherence decay is noted to 'tend to a gaussian behavior', while the simulations use exponential decay. Because T2n = 560 us is much longer than the pulse sequence durations (about 1 us), the numerical difference is likely negligible, but a one-sentence quantitative statement would preempt concern.","section":"Appendix C"},{"comment":"The protocol is called 'semi-device independent', but the precise assumptions (trusted dimension, trusted measurement settings, etc.) are only mentioned in the introduction; please state them explicitly in the protocol description.","section":"Main text, CHSH section"}],"recommendation":"major_revision","confidential_remarks":"The qubit-qudit CHSH part is solid and likely publishable after minor clarifications. The main risk is the trimer CGLMP section, which currently supports a weaker claim than the abstract makes. If the authors can supply a complete Lindblad simulation including the ancilla and residual couplings, or explicitly restrict the robustness claim, the paper could be accepted after revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThis paper is a serious feasibility study, and the CHSH half is worth taking seriously; the CGLMP trimer half is more fragile than the abstract suggests. What is new: an optimized GRAPE pulse sequence for the Yb(trensal) electro-nuclear qubit-qudit that, in Lindblad simulation with measured T2e=2.4 µs and T2n=560 µs, gives Bell values around 2.34, and a proposed Cr-Yb-Cr trimer for qudit-qudit CGLMP tests. The underlying Bell bounds come from Ref. [48] and the Hamiltonian is known, but the system-specific protocol, the pulse tables, and the realistic decoherence inputs are new.\n\nThe paper does its homework: it reports measured coherence times, provides detailed pulse parameters and fidelities, and the appendices flag important caveats. In particular, Appendix C honestly notes that the measured nuclear decay tends to a Gaussian shape typical of a nuclear-spin bath, not the exponential model used in the simulations. That is a real limitation, not a hidden one.\n\nThe main soft spots: the abstract says the inequalities are 'safely violated in a wide range of parameters, proving the robustness of entanglement.' For the CHSH case that is close to true, though 'proving' outruns a simulation with a simplified dephasing model. For the CGLMP trimer, the situation is weaker. At T2=5 µs the violation is marginal (I≈2.02–2.06) and depends on which amplitude you pick; the no-decoherence bound is 2.74, and the 5 µs points are a small island above threshold. The paper also does not show that the ancilla/switch dephasing (T2=1 µs) is included consistently in all reported numbers; the text says it has 'limited role' but I did not find a self-contained demonstration, and the state-preparation fidelities in Table X are already quite low at T2=5 µs. The stress-test concern about post-hoc amplitude selection is fair.\n\nThe central CHSH result holds up. The trimer is a speculative proposal that needs either experimental input or a more complete noise model before the abstract's robustness claim can be taken at face value.\n\nRecommendation: send to peer review, but ask the authors to temper the abstract, quantify statistical/calibration uncertainties, include the ancilla dephasing in the trimer simulation explicitly, and ideally release code and data.","headline":"A credible CHSH feasibility study on Yb(trensal) with an overreaching abstract and a much more fragile CGLMP trimer proposal.","tokens_in":28793,"tokens_out":2962,"would_cite":false,"duration_ms":31456,"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":"The paper argues that Yb(trensal) molecular nanomagnets can violate generalized Bell inequalities in realistic simulations.","keywords":["Bell inequalities","qudits","molecular nanomagnets","Yb(trensal)","CHSH inequality","CGLMP inequality","GRAPE optimal control","spin qudit entanglement"],"falsifier":"Run the published GRAPE pulse sequence on a $0.05\\%$ $^{173}$Yb(trensal):Lu(trensal) single crystal at $B_z=0.3$ T and measure the four joint diagonal expectation values; if the resulting CHSH value is at or below $2$, the central claim fails. Alternatively, fit the measured nuclear Hahn-echo decays to a Gaussian rather than an exponential and re-run the master-equation simulation; if Gaussian dephasing with the same $T_{2n}$ lowers the simulated Bell value below $2$, the robustness claim fails.","tokens_in":27552,"feed_emoji":"🧲","tokens_out":6033,"duration_ms":67435,"temperature":0.7,"pith_summary":"The paper argues that a single molecular nanomagnet, Yb(trensal), can be the stage for a Bell test that goes beyond two-level systems. Its electronic spin acts as a qubit, and four states of its nuclear spin form a four-level qudit, together making a qubit–qudit pair. Using measured Hamiltonian parameters and measured coherence times (electronic $T_{2e}\\approx 2.4\\,\\mu$s, nuclear $T_{2n}\\approx 560\\,\\mu$s), the authors simulate an optimized pulse sequence and find a generalized CHSH Bell value near $2.34$, above the classical bound of $2$. They also propose a Cr–Yb–Cr trimer in which two spin-$3/2$ qudits are entangled through a switchable Yb spin, giving CGLMP values up to about $2.61$ when each qudit has $T_2=30\\,\\mu$s. If correct, molecular magnets can certify high-dimensional entanglement in a semi-device-independent way using only ensemble expectation values.","feed_headline":"A single molecule passes a high-dimensional Bell test at 2.34","feed_subtitle":"Measured decoherence still lets Yb(trensal) beat CHSH and CGLMP bounds, opening molecular magnets to qudit tests.","key_machinery":"The argument runs through the generalized CHSH Bell operator $\\hat{O}_{\\rm Bell}=A\\otimes(B+B')+A'\\otimes(B-B')$ for a qubit–qudit pair, with the qudit observables $B,B'$ acting on a four-level subspace. The observable is re-expressed as four joint diagonal measurements after local unitary rotations, so the protocol needs only ensemble expectation values rather than projective measurements. The pulse sequence is built from resonant square pulses for state preparation and from GRAPE-optimized control fields for the measurement unitaries, simulated under the Lindblad master equation with pure-dephasing jump operators $S_z$ and $I_z$ at rates $1/T_{2e}$ and $1/T_{2n}$. For the trimer, the key mechanism is an ancilla-mediated controlled-phase gate: the central Yb spin acts as a switchable mediator, and a $\\pi$ phase is applied to a selected two-qudit component through a conditional excitation and de-excitation of the ancilla, after which the ancilla returns to its ground state.","core_discovery":"The central claim is that Yb(trensal), an effective spin-$1/2$ electron coupled by hyperfine interaction to a nuclear spin $5/2$, provides a practical platform for violating generalized Bell inequalities on a qubit–qudit system. The authors construct the entangled state $|\\psi\\rangle=\\frac12(|\\uparrow,3/2\\rangle+|\\uparrow,-1/2\\rangle+|\\downarrow,-1/2\\rangle+|\\downarrow,-3/2\\rangle)$, whose ideal CHSH value is $2.64575$, and show by Lindblad master-equation simulations with experimentally measured dephasing that the inequality is violated across a wide range of parameters; at the measured $T_{2e}=2.4\\,\\mu$s they obtain $\\langle\\hat{O}_{\\rm Bell}\\rangle\\simeq 2.34$. For the two-qudit extension, they propose a switchable Cr–Yb–Cr trimer and show that the CGLMP functional exceeds $2$, reaching about $2.61$ for $T_2=30\\,\\mu$s, with optimized fidelities above $0.91$. The authors state this proves the robustness of entanglement in the investigated molecular spin systems and opens the way to an actual broadband NMR experiment.","pith_inferences":["Because the protocol uses only ensemble expectation values of diagonal spin operators, it could be run on bulk molecular crystals without single-shot readout, though closing detection and freedom-of-choice loopholes would still require additional experimental design.","The paper notes in Appendix C that the measured nuclear-spin coherence decay tends to a Gaussian form typical of a nuclear-spin bath; testing the same pulse sequence under Gaussian rather than exponential pure dephasing would show whether the predicted Bell values survive a more realistic noise model.","The same GRAPE-based approach could be applied to other lanthanide or transition-metal molecular qudits to certify high-dimensional entanglement, not just for CHSH and CGLMP but for non-dichotomic qubit–qudit inequalities the authors list as future work.","For the trimer, measuring the residual qudit–qudit coupling when the switch is off would directly bound the main systematic error in the CGLMP protocol."],"forward_implications":["A Bell inequality can be violated in a single molecular crystal at $B_z=0.3$ T using existing broadband NMR and EPR techniques, without projective measurements.","The violation persists in a wide range of pulse amplitudes and coherence times, so the protocol is robust enough for an experimental implementation in currently available Yb(trensal) samples.","The switchable Cr–Yb–Cr trimer provides a route to qudit–qudit CGLMP violations with the entangling interaction turned on only during gates, in principle allowing space-like separated measurements.","Molecular nanomagnets become viable platforms for high-dimensional entanglement certification, complementing trapped-ion, Rydberg, and transmon qudit systems.","The GRAPE-optimized pulses are essential: hand-optimized sequences fail to violate the CHSH inequality at the measured $T_{2e}$, while the optimized ones reach about $2.34$."],"supporting_citations":[{"why":"Supplies the generalized CHSH formalism for qubit–qudit systems, including the optimal observables and the maximum violation $2.64575$ used as benchmark.","marker":"[48]"},{"why":"Defines the CGLMP inequality for pairs of $d$-dimensional systems, which the trimer protocol is designed to violate.","marker":"[46]"},{"why":"Provides the Yb(trensal) molecular nanomagnet with its effective spin-$1/2$ electron and nuclear spin-$5/2$, the physical platform of the qubit–qudit test.","marker":"[52]"},{"why":"Provides the GRAPE optimal-control implementation used to design high-fidelity pulses for the measurement unitaries.","marker":"[45]"},{"why":"Gives the original CHSH inequality and the local-hidden-variable threshold of $2$ that the qubit–qudit protocol must beat.","marker":"[47]"},{"why":"Underlies the Lindblad master equation used to simulate pure dephasing at the measured rates $1/T_{2e}$ and $1/T_{2n}$.","marker":"[53]"},{"why":"Provides the explicit local unitary transformations used to convert the CGLMP measurement settings into diagonal $S_z$ measurements.","marker":"[66]"},{"why":"Establishes molecular nanomagnets as qudit platforms and describes the ancilla-mediated switchable coupling used for the trimer entangling gate.","marker":"[10]"}],"fun_headline_variants":["Single molecule qubit-qudit violates Bell at 2.34","Yb(trensal) beats Bell bound despite decoherence","Qubit-qudit Bell violation robust in molecular magnet","High-dimensional Bell test passed by single molecule"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The simulations assume that pure dephasing with exponential decay at the measured rates $1/T_{2e}$ and $1/T_{2n}$ is the only significant noise; Appendix C admits the nuclear-spin-bath decay tends to a Gaussian behavior, and the trimer's residual qudit–qudit couplings are negligible only to first order, so if either assumption fails the predicted violations may shrink.","fun_headline_variants_meta":{"raw":{"variants":["Single molecule qubit-qudit violates Bell at 2.34","Yb(trensal) beats Bell bound despite decoherence","Qubit-qudit Bell violation robust in molecular magnet","High-dimensional Bell test passed by single molecule"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000936,"raw_usage":{"total_tokens":3989,"prompt_tokens":919,"completion_tokens":3070,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":535,"completion_tokens_details":{"reasoning_tokens":3002}},"tokens_in":535,"tokens_out":3070,"duration_ms":27661,"temperature":1.0,"reasoning_tokens":3002,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T11:19:09.683061+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the published GRAPE pulse sequence on a $0.05\\%$ $^{173}$Yb(trensal):Lu(trensal) single crystal at $B_z=0.3$ T and measure the four joint diagonal expectation values; if the resulting CHSH value is at or below $2$, the central claim fails. Alternatively, fit the measured nuclear Hahn-echo decays to a Gaussian rather than an exponential and re-run the master-equation simulation; if Gaussian dephasing with the same $T_{2n}$ lowers the simulated Bell value below $2$, the robustness claim fails.","supporting_citations":[{"cited_title":"Chiesa, G","cited_arxiv_id":null,"evidence_quote":"Supplies the generalized CHSH formalism for qubit–qudit systems, including the optimal observables and the maximum violation $2.64575$ used as benchmark."},{"cited_title":"Atzori, L","cited_arxiv_id":null,"evidence_quote":"Defines the CGLMP inequality for pairs of $d$-dimensional systems, which the trimer protocol is designed to violate."},{"cited_title":"Chiesa, P","cited_arxiv_id":null,"evidence_quote":"Provides the Yb(trensal) molecular nanomagnet with its effective spin-$1/2$ electron and nuclear spin-$5/2$, the physical platform of the qubit–qudit test."},{"cited_title":"Atzori and R","cited_arxiv_id":null,"evidence_quote":"Provides the GRAPE optimal-control implementation used to design high-fidelity pulses for the measurement unitaries."},{"cited_title":"Moreno-Pineda, C","cited_arxiv_id":null,"evidence_quote":"Gives the original CHSH inequality and the local-hidden-variable threshold of $2$ that the qubit–qudit protocol must beat."},{"cited_title":"Chiesa, F","cited_arxiv_id":null,"evidence_quote":"Underlies the Lindblad master equation used to simulate pure dephasing at the measured rates $1/T_{2e}$ and $1/T_{2n}$."},{"cited_title":"Hansen, C","cited_arxiv_id":null,"evidence_quote":"Provides the explicit local unitary transformations used to convert the CGLMP measurement settings into diagonal $S_z$ measurements."}],"review_version":1}