{"id":"cce7f5cb-2a63-4c2c-8928-72af8c50f3c4","arxiv_id":"2608.07608","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"At large tilt angles, the homogeneous ground state of dipolar lattice bosons supports only empty and full fillings, and thermal fluctuations turn the resulting instability into a phase-separated state that mimics the experimental self-bound insulator.","lead":"Using quantum Monte Carlo simulations, this paper maps the phases of hard-core dipolar bosons on a square lattice and finds a region where only the empty and fully filled states are stable. It argues that the 'self-bound insulator' reported in recent experiments is not a zero-temperature phase but a finite-temperature, phase-separated state.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ground-state instability is inferred solely from hysteretic grand-canonical µ-sweeps, without canonical fixed-density verification or proper long-range interaction extrapolation; the thermodynamic limit is not established.","rationale":"The reader's weakest assumption is that the QMC sizes 10–48 with β=L/J capture the thermodynamic limit and true zero-temperature behavior. This is indeed the central weak point: the ground-state instability at θ≳62° is the load-bearing element, because the finite-temperature explanation of the experimental self-bound insulator depends on the ground state lacking a half-filled phase. I agree with the reader that finite-size and finite-temperature effects could alter this conclusion, but I sharpen the concern in two ways. First, the evidence for the instability is purely the hysteretic µ-sweep; no canonical fixed-density simulation at T≈0 is shown, so the observed 'jump' may reflect the inability of the worm algorithm to sample both phases rather than a genuine absence of a homogeneous state. Second, the dipolar interaction cutoff is set to the system size, so the Hamiltonian is not fixed as L grows; without Ewald summation or a proper extrapolation of the long-range tail, consistency across L does not guarantee convergence to the infinite-lattice limit. These issues do not demonstrate that the central claim is wrong—the finite-temperature results and previous studies on related models lend plausibility—but they leave the thermodynamic-limit statement insufficiently supported. The proposed canonical test at several L with a controlled interaction truncation would settle the question directly. The reader's conditional verdict therefore remains appropriate; I would not change it.","tokens_in":9692,"tokens_out":11740,"duration_ms":119546,"concrete_test":"Run canonical-ensemble QMC at fixed n=0.5 for L=12, 24, 36, 48 at θ=75° and V/J=10, with β=L/J and β=2L/J, and with an Ewald-summed (or substantially larger-cutoff) dipole interaction. Measure the largest filled-domain size and the diagonal structure factor S(π/2,π/2). Phase separation into two domains with the largest filled-domain size scaling as L²/2 and an interface width independent of L would confirm the ground-state instability; a homogeneous state or a stripe of fixed width with increasing L would indicate that the observed jump is a finite-size artifact.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The key ground-state claim—that for θ ≳ 62° only n=0 and n=1 are stable, with no homogeneous half-filled phase—is inferred from grand-canonical µ sweeps that show a hysteretic jump with no stable intermediate densities (Fig. 4). In finite-size QMC, a first-order transition can display such a jump even when the true thermodynamic behavior is different, because the system can remain trapped in a metastable state. Establishing the instability requires either canonical fixed-density simulations at T≈0 or a Maxwell construction from E(n); neither is reported. A further technical issue is that the dipolar interaction is truncated at the system size L (Section II) without Ewald summation, so the Hamiltonian itself changes with L and the long-range tail is not properly extrapolated. The claim that the jump is 'consistently observed for all system sizes explored' does not establish convergence, since the dipolar sum is conditionally convergent and shape-dependent. If the observed instability is a finite-size or metastability artifact, the central conclusion that the experimental self-bound insulator is a finite-temperature effect loses its foundation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript studies hard-core dipolar bosons on a two-dimensional square lattice at fixed azimuthal angle φ=45°, using path-integral quantum Monte Carlo with the worm algorithm. The authors map a zero-temperature phase diagram at half filling as a function of polar angle θ and interaction strength V/J, identifying superfluid, checkerboard, double diagonal stripe, and incompressible regions. Their central claim is that for θ≳62° half filling becomes unstable in the homogeneous ground state, so that only the empty state n=0 and the fully filled state n=1 are stable, connected by a first-order transition. At finite temperature, specifically k_B T=5J, they report that thermal fluctuations stabilize intermediate fillings around n≈0.5 as phase-separated states of empty and fully filled domains, and that similar structures appear in a harmonic trap. They conclude that the experimentally reported self-bound insulator at half filling is not a ground-state phase of the homogeneous model but rather a finite-temperature phase-separated state or a trap-induced coexistence phenomenon.","tokens_in":9829,"tokens_out":5154,"duration_ms":56502,"significance":"If the central claim holds, the paper provides a direct theoretical resolution of an apparent discrepancy between the experimental observation of a self-bound insulator and the homogeneous ground state of an extended Bose-Hubbard model. The simulations are parameter-free, with V/J and θ as control parameters and no fitted constants, and the finite-temperature mechanism is a concrete, falsifiable prediction that can be tested experimentally. The paper also makes a useful methodological point: density configurations alone may not distinguish equilibrium phases from phase coexistence, and unbiased QMC benchmarks are valuable for interpreting experimental images. The main weakness is that the zero-temperature instability is inferred from hysteretic grand-canonical sweeps with truncated dipolar interactions and without canonical fixed-density verification, so the thermodynamic-limit ground-state statement is not yet established with the required rigor.","major_comments":[{"comment":"The central ground-state claim that for θ≳62° only n=0 and n=1 are stable is inferred exclusively from grand-canonical chemical-potential sweeps showing a hysteretic jump with no stable intermediate densities. In finite-size QMC, such a jump can reflect metastability rather than a true first-order transition, and a μ-sweep cannot exclude a homogeneous half-filled ground state that is simply not reached by the algorithm. The use of β=L/J is also not a controlled zero-temperature extrapolation. Please add canonical fixed-density simulations at N=L²/2 for several system sizes, or an explicit free-energy/Maxwell construction from E(N), and show that the half-filled state is thermodynamically unstable in the thermodynamic limit.","section":"Section III, Fig. 4"},{"comment":"The dipolar interaction is truncated at a distance equal to the system size L and used with periodic boundary conditions, but no Ewald summation or equivalent long-range treatment is described. The dipole-dipole sum for this angular dependence is conditionally convergent and shape-dependent, so the Hamiltonian itself changes with L. The statement that the jump is consistently observed for all system sizes explored (10≤L≤42) does not establish convergence of the interaction energy. Please demonstrate that the instability and the n=0/n=1 transition are robust to the interaction cutoff, for example by comparing truncated and Ewald-summed interactions or by extrapolating in L.","section":"Section II, Eq. (1)"},{"comment":"No statistical error bars are reported for any observable, and the phase boundaries in Fig. 2 are assigned widths δθ that appear to be hysteresis widths rather than statistical uncertainties. Without error bars and a finite-size scaling analysis, the quoted onset range θ_i∈[62°,68°] and the locations of the CB, double-DSS, and IP boundaries cannot be quantitatively assessed. Please include error bars for key quantities such as density, superfluid stiffness, and structure factor, and provide a scaling analysis for the onset angle of the instability.","section":"Section III and Fig. 2"},{"comment":"The finite-temperature stabilization of intermediate fillings is presented through grand-canonical n(μ) curves and bimodal histograms, which support coexistence, but the claim that these are thermodynamically stable phase-separated states would be strengthened by canonical simulations at fixed total filling for the same parameters, together with a demonstration that the phase-separated configuration is independent of initialization and system size. As written, the evidence for 'stable' intermediate fillings rests on the same grand-canonical framework used for the ground-state claim.","section":"Section IV, Fig. 5"}],"minor_comments":[{"comment":"The manuscript reports different ranges of system sizes in different places: Section III states 12≤L≤48, while Section IV states 10≤L≤42 for the ground-state instability. Please harmonize these statements and clarify which sizes were used for each data set.","section":"Section III versus Section IV"},{"comment":"The text acknowledges that the 'incompressible phase' does not have a unique ordering and that the particle arrangement depends on system size and initial conditions, yet Fig. 2 and the abstract label this as a single phase. Please clarify in the phase diagram that this region is an incompressible region with multiple near-degenerate stripe configurations rather than a uniquely ordered phase.","section":"Section III, IP description"},{"comment":"The caption of Fig. 3 states that the local occupation density is 'averaged over a single Monte Carlo configuration,' which is unclear. Please specify whether the maps show instantaneous snapshots, time-averaged densities, or averages over multiple configurations, and define the averaging procedure explicitly.","section":"Section II and Fig. 3"},{"comment":"No details are given about the number of Monte Carlo sweeps, equilibration criteria, binning, or statistical analysis. These details are standard for QMC studies and should be included so that the numerical claims are reproducible.","section":"General"},{"comment":"The trapped-system results are presented for a single set of parameters (L=30, W=0.003J, θ=80°, k_B T=5J). A brief discussion of how the qualitative picture depends on trap strength and system size would help connect these simulations to the experimental geometry.","section":"Section IV, trap simulation"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and addresses an experimentally relevant question. The finite-temperature phase-separation mechanism is plausible and testable, and the QMC approach is standard. My main concern is methodological: the central ground-state claim is currently supported only by grand-canonical hysteresis data with a truncated dipolar interaction, and the manuscript does not report statistical error bars. These issues are fixable within the manuscript's scope, so I recommend major revision rather than rejection. The comparison to the experimental self-bound insulator is valuable but should be framed as conditional on the thermodynamic-limit analysis being completed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Plainly: this is a credible QMC study with a genuinely new observation. For polar angles above about 62 degrees at fixed azimuthal angle 45 degrees, the homogeneous ground state of hard-core dipolar bosons on a square lattice refuses to stabilize half-filling; only empty and fully filled states are stable, connected by a first-order transition. At finite temperature, intermediate fillings reappear as phase-separated domains of empty and full regions, which look like the self-bound insulator seen in the 2023 Nature experiment. That is a real result, and the paper deserves a careful read.\n\nWhat is done well: the methodology is standard path-integral QMC with the worm algorithm, system sizes up to L=48 are reasonable for a first study, and the authors are admirably honest about the incompressible region—they openly state that no unique ordering emerges and that metastability is present. The finite-temperature stabilization of phase-separated states is new and physically plausible. The comparison to experiment is measured: they do not claim the experimental state is impossible, only that it is likely a finite-T phase-separated equilibrium state or even phase coexistence, and they explicitly warn against misinterpreting density images.\n\nThe soft spots are real but not fatal. The central ground-state claim—no half-filled phase above theta~62 degrees—rests entirely on grand-canonical mu-sweeps that exhibit hysteresis and an abrupt n=0 to n=1 jump. In finite-size QMC, such a jump can be a metastability artifact. The paper does not provide canonical fixed-density simulations at T->0 or a Maxwell construction from E(n), which would nail the first-order character. In addition, the dipolar interaction is truncated at the system size L without Ewald summation, so the Hamiltonian itself changes with L; the authors say the jump is \"consistently observed\" for all sizes, but that does not establish thermodynamic-limit convergence for a long-range, conditionally convergent interaction. These two gaps are load-bearing because the experiment comparison depends on the ground-state claim. They are addressable, and the finite-T results give indirect support that the instability is real.\n\nMinor: no statistical error bars appear anywhere, which makes the hysteresis widths and phase boundaries hard to evaluate. The IP region is admittedly ill-defined, which is fine but limits the precision of the phase diagram.\n\nWho is this for? Anyone interpreting quantum-gas-microscope images of dipolar lattice gases, and people working on phase separation in long-range interacting lattice models. It deserves peer review. With canonical verification and a proper long-range extrapolation, this would be a solid contribution. Without those, the experimental interpretation remains suggestive rather than conclusive.","headline":"A credible QMC study with a new finite-T phase-separation claim, but the ground-state instability needs canonical verification and proper long-range extrapolation.","tokens_in":10395,"tokens_out":3620,"would_cite":false,"duration_ms":35642,"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":"For hard-core dipolar bosons past about 62 degrees of tilt, half-filling is unstable: only empty and fully filled states are stable, and finite temperature stabilizes the phase-separated 'self-bound' insulator.","keywords":["hard-core dipolar bosons","extended Bose–Hubbard model","quantum Monte Carlo","density instability","first-order phase transition","phase separation","self-bound insulator","optical lattice"],"falsifier":"Measure the ground state of the same model at, say, $\\theta=75^\\circ$ and $V/J=10$ with $k_{\\rm B}T$ much smaller than $J$: if a homogeneous half-filled state with uniform density and vanishing compressibility appears (rather than coexistence of empty and fully filled regions), the claim that only $n=0$ and $n=1$ are stable in the ground state is wrong. Conversely, if experiments see the self-bound-looking structure persist to temperatures far below $J$, the finite-temperature-stabilization explanation would be falsified.","tokens_in":9456,"feed_emoji":"🧲","tokens_out":13238,"duration_ms":105173,"temperature":0.7,"pith_summary":"Using quantum Monte Carlo simulations of hard-core dipolar bosons on a square lattice, this paper asks what really stabilizes the 'self-bound insulator' reported in recent experiments on dipolar atoms in optical lattices. It finds that for polar tilt angles $\\theta\\gtrsim 62^\\circ$ at fixed azimuthal angle $\\varphi=45^\\circ$, the half-filled state is thermodynamically unstable in the homogeneous ground state: only the empty state $n=0$ and the fully filled state $n=1$ are stable, with a first-order transition between them. At finite temperature, thermal fluctuations shift the instability threshold to larger $\\theta$ and stabilize intermediate fillings around $n\\approx 0.5$, realized as phase-separated domains of empty and fully filled regions that resemble the observed self-bound insulator. The paper's central conclusion is that the experimentally observed self-bound insulator is a finite-temperature phase-separated state rather than a zero-temperature equilibrium phase, which matters for interpreting microscope images of dipolar lattice gases.","feed_headline":"Steep dipole tilt kills half-filling; heat revives it as domains","feed_subtitle":"It means the experimental 'self-bound' insulator is a finite-temperature effect, not a ground-state phase.","key_machinery":"The central object is the angle-dependent dipole-dipole interaction in the extended Bose–Hubbard model, $V\\sum_{i