{"id":"7beea43f-992a-4147-98ae-71fe19948be3","arxiv_id":"2501.08233","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A Rydberg tweezer array realizes a tunable bosonic t-J-V model and demonstrates dynamical phase separation, bound hole pairs, and sign-dependent pair mobility from next-nearest-neighbor tunneling.","lead":"Using a grid of laser-held rubidium atoms, this experiment builds a small quantum magnet and removes some atoms to create mobile holes, mimicking the doped antiferromagnets relevant to high-temperature superconductivity. It then shows that holes can clump together and that long-range hopping makes hole pairs light or heavy depending on the sign of the hopping.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Pair-mass mechanism is not fully isolated: teff relies on an underived chi(T), and the NN-vs-full control removes all long-range couplings, not just t'.","rationale":"The paper's experimental core—realizing a three-level Rydberg t-J-V model, observing dynamical phase separation at small |t/J|, and forming repulsively bound hole pairs—is well supported by exact diagonalization of the full Rydberg Hamiltonian with an explicit error model, by the isolated-pair calibrations in Fig. 1d-e, and by the 2D FM/AM comparisons in Fig. 5. I therefore see no reason to move the reader's verdict toward rejection. The weakest point is indeed the sign-dependent pair-mass mechanism, as the reader identified. My formulation adds specificity: the existing NN-vs-full comparison is not a clean intervention on t', because truncating to nearest-neighbor simultaneously deletes the long-range parts of J_perp, Jz, and V. In practice those tails are O(1/64) relative to the NN values, so this is a gap in presentation and control rather than a demonstrated error. The underived chi(T) is more serious, since the central formula cannot be used predictively without it and the paper's stated late-time value is not a derivation. A direct simulation toggling only t' would settle both points without changing the overall assessment; the CONDITIONAL verdict stands.","tokens_in":24458,"tokens_out":7038,"duration_ms":73563,"concrete_test":"Perform exact time evolution of the L=12 z-AFM hole-pair protocol at theta=49.7 and 59.7 under three Hamiltonians: (i) full Rydberg; (ii) full Rydberg with only the NNN tunneling amplitudes t_{i,i+2} set to zero, retaining all longer-range J and V couplings; (iii) full Rydberg with the sign of t_{i,i+2} reversed relative to NN t. If the t>0/t<0 pair-displacement asymmetry persists in (ii), the light/heavy effect is not caused by the t' interference; if it disappears only when the t' sign is changed, the attribution is confirmed. In parallel, compute chi(T) independently from the time-dependent spin-overlap in the second-order matrix element and check whether the same chi(T) at both angles reproduces the measured displacement ratio without fitting.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that light/heavy hole pairs depending on the sign of t are due to interference between NNN tunneling t' and second-order pair tunneling rests on teff = chi(T) t^2/(V - Jz/4) - t/8 (Methods, 'Influence of magnetic background on hole pair'). Two assumptions are load-bearing. First, chi(T), the spin-fluctuation overlap factor, is not derived; it is stated to be time- and background-dependent and only characterized at late times ('we find'), leaving an unspecified function that can absorb discrepancies in the cancellation producing the heavy pair. Second, the two comparison angles theta=49.7 and 59.7 differ not only in the sign of t but also in V/J_perp (1.2 vs 1.0), Jz/J_perp, and absolute J_perp (Table II). The numerical control in Extended Data Fig. 7 compares the full Rydberg Hamiltonian with a model truncated to nearest-neighbor couplings; that removes not only NNN tunneling but all 1/r^6 tails of J and V. If those tails or an angle-dependent chi(T) contribute, the attribution of the measured asymmetry specifically to the -t/8 interference term is not established by the data as presented.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a Rydberg-tweezer realization of a hard-core bosonic t-J-V model, with spin-up, spin-down, and hole encoded in three Rydberg states of 87Rb. The authors benchmark the bare interaction parameters on two-atom pairs, then study quench dynamics of doped 1D chains and a 5x5 2D array. They observe hole-spin domain separation for small |t/J|, repulsively bound hole pairs, a sign-dependent pair mobility asymmetry that they attribute to interference between NNN dipolar tunneling and second-order pair tunneling, and single-hole dynamics in 2D ferromagnetic and antiferromagnetic backgrounds. The central new claim is that the light/heavy pair asymmetry is caused by the 1/r^3 tail of the tunneling.","tokens_in":24707,"tokens_out":14271,"duration_ms":149121,"significance":"If the central claims hold, this is an important step: it is, to my knowledge, the first direct high-filling realization of a bosonic t-J-type model with site-resolved hole and spin readout and with NNN tunneling implemented natively by the dipolar interaction. The mapping is carefully benchmarked against two-atom exchange measurements, and the many-body predictions come from full Rydberg simulations rather than from fitting the many-body data, so the comparisons are not circular. The 2D single-hole interference pattern in the ferromagnetic case is a clear fingerprint of long-range tunneling. The main weakness is that the microscopic mechanism for the pair-mass asymmetry is not yet isolated from other angle-dependent couplings, as detailed below.","major_comments":[{"comment":"The formula teff = chi(T) t^2/(V - Jz/4) - t/8 is load-bearing for the light/heavy pair claim, but chi(T) is never derived. The Methods states only that chi is time- and spin-background-dependent and that 'at late times, we find' the perturbative amplitude is positive. No closed-form expression, normalization, or independent determination is given. Since the pair displacement is measured at T = 0.8 and 1.6 x 2pi/|2t|, which need not be the late-time regime in which chi was characterized, an unspecified chi could absorb part of the observed asymmetry. Please provide a derivation of chi for the relevant spin backgrounds and times, or determine it from a spin-only numerical simulation, and show that the resulting teff reproduces the full-Rydberg pair-displacement curves at both angles.","section":"Methods, 'Influence of magnetic background on hole pair'; main text near Fig. 3"},{"comment":"The two angles used to demonstrate the sign effect, theta = 49.7 deg and 59.7 deg, differ not only in the sign of t but also in V/J_perp (1.2 vs 1.0), Jz/J_perp, and the absolute coupling scale. The numerical control in Extended Data Fig. 7 truncates all couplings beyond nearest neighbors, thereby removing the 1/r^6 tails of J and V as well as the NNN tunneling t'. It therefore establishes that some long-range coupling is needed to reproduce the asymmetry, but it does not isolate the specific interference term -t/8. I request an additional control in which only the NNN tunneling is turned off while the long-range J and V tails are kept (or vice versa), or an angle pair in which the sign of t changes while V/J_perp and Jz/J_perp are held fixed, so that the attribution to t' is unambiguous.","section":"Table II and Extended Data Fig. 7"}],"minor_comments":[{"comment":"The phrase 'qualitative (quantitative) agreement with numerical simulations without (with) errors' is ambiguous; please specify which data are compared to which simulation and report a quantitative figure of merit for the agreement.","section":"Paragraph after Fig. 3e"},{"comment":"The pair displacement is defined using the operational cutoff of bond length l <= 2; please show how the central results depend on this cutoff (e.g., repeat the analysis for l <= 1 and l <= 3) so that the mobility comparison is not sensitive to the chosen definition.","section":"Fig. 3d,e and Fig. 4"},{"comment":"The arXiv metadata title includes 'with dipolar tunnelings' while the manuscript header omits this phrase; the two should be made consistent in the final version.","section":"Title"},{"comment":"The fitted angular coefficients F1, F2, F3 used to represent the van der Waals C6 interactions are not tabulated; providing these values (or a persistent source for the fits) would make the numerical simulations fully reproducible.","section":"Methods, 'Hamiltonian mapping'"}],"recommendation":"major_revision","confidential_remarks":"I believe the experimental work is sound and the qualitative claims are likely correct. The revision I request is focused: derive or independently fix chi(T), and add a numerical control that isolates NNN tunneling from the other long-range tails. If those are provided, the paper should be acceptable. I also note that the data availability policy ('available from the corresponding author on request') is weaker than current norms; I would encourage the editor to ask for public deposition of the raw site-resolved data and simulation snapshots."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The experiment is real and the main qualitative claims hold up. This is not a new model -- the three-level Rydberg encoding is from the same group's earlier proposal -- but the realization at high filling, with single-site control, and the observations of phase separation, bound hole pairs, and the dipolar-tail effects in 2D are new and well executed. The simulations are a strong point: they use the full Rydberg Hamiltonian with interaction strengths benchmarked on two-atom exchange data, not fitted to the many-body results. That gives real weight to the phase-separation and bound-pair observations.\n\nThe soft spot is exactly where the reader put it: the pair-mass asymmetry. The effective tunneling teff = chi t^2/(V - Jz/4) - t/8 relies on a spin-fluctuation factor chi(T) that is stated to be time- and background-dependent, but never derived. The Methods say only \"we find\" that it is positive at late times. That is an honest admission, but it leaves an unspecified function that could absorb discrepancies. The two comparison angles, 49.7 and 59.7 degrees, differ not only in the sign of t but also in V/Jperp and Jz/Jperp. And the NN-versus-full numerical control removes all long-range couplings, not just t'. So the claim that the light/heavy pair asymmetry is specifically caused by interference with NNN tunneling is plausible but not isolated to the level the abstract states. The observation itself is credible -- the asymmetry is there and the full model reproduces it -- but the mechanism attribution is not bulletproof.\n\nTwo smaller things. First, the paper says the simulations describe the pair dynamics \"qualitatively but not quantitatively\" in Fig. 3d, and mention correlated initial-state errors as a possible reason. That is honest, but it means the quantitative aspect of the pair-binding comparison is weaker than the text sometimes implies. Second, the methods state the experiments use B = 46 G while the pairinteraction calculations in Extended Data Fig. 5 use B = 50 G. That discrepancy is never explained; it may be harmless, but it should be addressed. Data being \"available on request\" is also a mild negative; this is the kind of experiment where deposited data would help.\n\nWho is this for? Anyone working on doped quantum magnets or analog quantum simulation with Rydberg arrays. It deserves a serious referee. My recommendation: send it out, but the authors should be asked to either derive or measure chi(T) more carefully, or to soften the specific interference claim. The central results will stand either way.","headline":"Genuinely new experimental capability with credible qualitative results; the sign-dependent pair-mass attribution is the one place where the evidence is softer than the abstract suggests.","tokens_in":762,"tokens_out":835,"would_cite":true,"duration_ms":27656,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper reports the realization of a doped quantum antiferromagnet in a Rydberg tweezer array, using three Rydberg states per atom to encode spins and hard-core holes, and the observation of hole-spin phase separation, repulsively…","keywords":["Rydberg tweezer array","doped quantum antiferromagnet","bosonic t-J-V model","hard-core bosonic holes","dipolar tunneling","repulsively bound hole pairs","next-nearest-neighbor tunneling","single-site control"],"falsifier":"Measure the center-of-mass displacement of a bound hole pair at two angles with opposite signs of $t$ but with $V$, $J_z$, and $J_\\perp$ held fixed (by compensating the angle change with a change in lattice spacing), and compare with a simulation in which the next-nearest-neighbor tunneling $t'$ is set to zero; if the displacement asymmetry persists, the claimed NNN interference mechanism is not the cause.","tokens_in":24225,"feed_emoji":"⚛️","tokens_out":9734,"duration_ms":86985,"temperature":0.7,"pith_summary":"This paper reports the realization of a doped quantum antiferromagnet in a Rydberg tweezer array, using three Rydberg states per atom to encode spin-up, spin-down, and a hard-core hole. The central achievement is a tunable bosonic $t$-$J$-$V$ Hamiltonian whose hole tunneling $t$, spin couplings $J_\\perp,J_z$, and hole-hole repulsion $V$ can be varied by rotating the array relative to the quantization axis. With that control, the authors observe dynamical phase separation between hole-rich and spin-rich regions at $|t/J|\\ll1$, the formation of repulsively bound hole pairs, and a sign-dependent pair mobility caused by interference between next-nearest-neighbor tunneling and second-order pair tunneling. They also follow a single hole in a 2D square lattice with ferromagnetic and antiferromagnetic spin backgrounds. If correct, the platform opens the high-density regime of the $t$-$J$ model, the regime relevant to doped Mott insulators and high-temperature superconductivity, to direct quantum simulation with single-site resolution.","feed_headline":"Rydberg tweezers realize tunable doped quantum antiferromagnet","feed_subtitle":"Three Rydberg levels encode spins and holes, reaching t-J model regimes optical lattices cannot reach.","key_machinery":"The carrying object is the hard-core bosonic $t$-$J$-$V$ Hamiltonian, Eq. (1), with tunneling $\\hat H_t$ between a hole and a spin at distance $r$ with amplitude $t_\\sigma/r^3$, spin exchange $J_\\perp/(2r^6)(\\hat S_i^+\\hat S_j^-+\\mathrm{h.c.})$ plus Ising $J_z \\hat S_i^z \\hat S_j^z/r^6$, and hole-hole repulsion $V \\hat n_i^h \\hat n_j^h/r^6$. The mechanism that makes the experiment work is the dual power-law structure of Rydberg interactions: the $1/r^3$ dipolar tail produces both nearest-neighbor tunneling $t$ and next-nearest-neighbor tunneling $t'=t/8$, while the $1/r^6$ van der Waals terms produce spin and hole interactions that can be tuned by the angle $\\theta$ between the array and the quantization axis. At the magic angle $\\theta_m\\approx54.7^\\circ$ the tunneling vanishes, placing the system at $|t|\\ll J$; on either side the sign of $t$ flips, which flips the sign of the NNN contribution and makes the effective pair tunneling $t_{\\mathrm{eff}}=\\chi t^2/(V-J_z/4)-t/8$ either constructive or destructive, thereby controlling the pair's effective mass $m_{\\mathrm{eff}}\\propto 1/(2t_{\\mathrm{eff}})$.","core_discovery":"The paper claims that by mapping the three Rydberg states $|\\downarrow\\rangle=|60S_{1/2},m_J=1/2\\rangle$, $|\\uparrow\\rangle=|61S_{1/2},m_J=1/2\\rangle$, and $|h\\rangle=|60P_{3/2},m_J=-1/2\\rangle$ of $^{87}$Rb atoms to spin-down, spin-up, and hole, the dipole-dipole exchange $\\propto 1/r^3$ and van der Waals interactions $\\propto 1/r^6$ faithfully implement a hard-core bosonic $t$-$J$-$V$ Hamiltonian with at most one particle per site. The authors show that tilting the chain angle across the magic angle $\\theta_m=54.7^\\circ$ tunes the hole tunneling $t$ through zero and reverses its sign, while the spin couplings $J_\\perp,J_z$ and hole-hole interaction $V$ remain sizeable, allowing them to enter the high-density $|t/J|\\ll1$ regime. On this platform they observe dynamical phase separation of holes and spins, repulsively bound hole pairs, and a sign-dependent pair mobility that they attribute to constructive or destructive interference between next-nearest-neighbor tunneling $t'=-t/8$ and second-order pair tunneling $\\propto t^2/(V-J_z/4)$. They further show single-hole coherent dynamics in a 2D square array with ferromagnetic and antiferromagnetic backgrounds, where the dipolar tail of the tunneling is visible in the interference pattern.","pith_inferences":["A closed-form derivation of the spin-fluctuation prefactor $\\chi(T)$ would turn $t_{\\mathrm{eff}}=\\chi t^2/(V-J_z/4)-t/8$ into a quantitative prediction; without it, the sign-dependent pair mass is identified but not fully explained.","Because $t'=-t/8$ is fixed by lattice geometry, the same interference mechanism could be used to engineer the mobility of larger hole clusters by flipping the sign of $t$, a control that nearest-neighbor models do not offer.","The 2D antiferromagnetic single-hole data suggest that spin memory suppresses path interference; an adiabatic extension from staggered states would test whether this suppression survives in the low-energy sector.","If the pair-mass asymmetry is confirmed in other spin backgrounds, the simulator becomes a testbed for kinetic-magnetism and pairing mechanisms in bosonic $t$-$J$ models at finite doping."],"forward_implications":["The platform reaches the high-particle-density $|t/J|\\ll1$ regime that optical-lattice superexchange simulators cannot access, making dynamical phase separation and hole clustering directly observable.","Since $\\theta$ tunes the sign of $t$, one setup can compare constructive and destructive interference between perturbative pair tunneling and the $1/r^3$ tunneling tail, giving control over the effective pair mass.","In the 2D square array with a ferromagnetic background, the hole occupation develops diagonal interference peaks that a nearest-neighbor-only simulation does not reproduce, directly showing the dipolar tail.","The same three-level encoding extends the toolbox beyond spin-1/2 to spin-1 chains and Haldane physics, as the paper states.","Single-site initialization permits controlled studies of one-hole and few-hole dynamics in ferromagnetic and antiferromagnetic backgrounds, relevant to magnetic polaron physics."],"supporting_citations":[{"why":"Supplies the proposed three-Rydberg-level encoding of spins and holes as an antiferromagnetic bosonic t-J model.","marker":"[12]"},{"why":"Provides the bosonic t-J-V Hamiltonian form into which the Rydberg pair interactions are mapped.","marker":"[5]"},{"why":"Benchmarks the dipolar and van der Waals interaction amplitudes used to validate the model and simulations.","marker":"[39]"},{"why":"Introduces repulsively bound pairs, the concept used to interpret the hole-pair bound states.","marker":"[41]"},{"why":"Predicts phase separation in hard-core bosonic t-J models, which the dynamical phase-separation observation addresses.","marker":"[29]"},{"why":"Predicts quantum-interference-induced pairing in the antiferromagnetic bosonic t-J model, supporting the pairing interpretation.","marker":"[33]"},{"why":"Supplies the magnetic-polaron framework for a mobile hole in an antiferromagnetic background used in the 2D experiment.","marker":"[45]"}],"fun_headline_variants":["Rydberg tweezers realize doped antiferromagnet with dipolar tunneling","Doped quantum antiferromagnet built with Rydberg tweezers","Tunable t-J model realized in Rydberg atom array","Rydberg array simulates doped antiferromagnet with hole pairs","Dipolar tunnelings tune holes in Rydberg antiferromagnet"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's explanation of the light and heavy hole pairs assumes that the sign-dependent interference between next-nearest-neighbor tunneling and second-order pair tunneling is the dominant cause of the measured pair-mobility asymmetry, even though the two angles compared also change $V$, $J_z$, and $J_\\perp$, and the spin-fluctuation prefactor $\\chi(T)$ is not derived in closed form.","fun_headline_variants_meta":{"raw":{"variants":["Rydberg tweezers realize doped antiferromagnet with dipolar tunneling","Doped quantum antiferromagnet built with Rydberg tweezers","Tunable t-J model realized in Rydberg atom array","Rydberg array simulates doped antiferromagnet with hole pairs","Dipolar tunnelings tune holes in Rydberg antiferromagnet"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000514,"raw_usage":{"total_tokens":2606,"prompt_tokens":1166,"completion_tokens":1440,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":782,"completion_tokens_details":{"reasoning_tokens":1342}},"tokens_in":782,"tokens_out":1440,"duration_ms":9984,"temperature":1.0,"reasoning_tokens":1342,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T20:29:32.981319+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the center-of-mass displacement of a bound hole pair at two angles with opposite signs of $t$ but with $V$, $J_z$, and $J_\\perp$ held fixed (by compensating the angle change with a change in lattice spacing), and compare with a simulation in which the next-nearest-neighbor tunneling $t'$ is set to zero; if the displacement asymmetry persists, the claimed NNN interference mechanism is not the cause.","supporting_citations":[{"cited_title":"Homeier, T","cited_arxiv_id":null,"evidence_quote":"Supplies the proposed three-Rydberg-level encoding of spins and holes as an antiferromagnetic bosonic t-J model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the bosonic t-J-V Hamiltonian form into which the Rydberg pair interactions are mapped."},{"cited_title":"Emperauger, M","cited_arxiv_id":null,"evidence_quote":"Benchmarks the dipolar and van der Waals interaction amplitudes used to validate the model and simulations."},{"cited_title":"Winkler, G","cited_arxiv_id":null,"evidence_quote":"Introduces repulsively bound pairs, the concept used to interpret the hole-pair bound states."},{"cited_title":"Boninsegni, Phase Separation in Mixtures of Hard Core Bosons, Physical Review Letters 87, 087201 (2001)","cited_arxiv_id":null,"evidence_quote":"Predicts phase separation in hard-core bosonic t-J models, which the dynamical phase-separation observation addresses."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the magnetic-polaron framework for a mobile hole in an antiferromagnetic background used in the 2D experiment."}],"review_version":1}