{"id":"eda27762-a9c8-4083-a76e-773eca675403","arxiv_id":"2505.17411","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Condensate fraction scaling provides a more accurate and less finite-size-affected probe of the BKT transition in the 2D attractive Hubbard model, giving TBKT/t = 0.1420(7) for U/t = -4, μ/t = 0.25 and a specific-heat anomaly at about 1.10 TBKT.","lead":"A quantum Monte Carlo study of the 2D attractive Hubbard model shows that the condensate fraction, the fraction of fermions paired in the lowest Cooper-pair state, scales algebraically with system size below the Berezinskii-Kosterlitz-Thouless (BKT) transition and exponentially above it, with much smaller finite-size effects than the usual on-site pairing correlator.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that nc has negligible finite-size corrections for L≥20 is empirical and underpins all three TBKT estimates; no stability test under varying L_min is shown.","rationale":"The reader's weakest assumption concerns the heuristic inflection-point definition of TBKT(L) and the fixed b=1 in the logarithmic extrapolation. I agree that those are weaknesses, but I identify a more fundamental load-bearing issue: the finite-size scaling of nc itself. The inflection point is only one of three extraction methods; the more serious concern is that all three methods rely on the same assumption that nc follows the BKT scaling form without significant subleading corrections for L≥20. The paper provides strong empirical evidence for consistency among methods, and the exponents η from nc agree with those from ⟨Δ²⟩ in the large-L regime, which partially supports the assumption. However, no analytical derivation or systematic stability check (e.g., varying L_min) is given, so the systematic error in TBKT cannot be assessed. This does not invalidate the central claim — the agreement with superfluid-density results and the consistency of three methods are genuine evidence — but it does mean the claimed accuracy from moderate sizes should be treated as conditional. My concern does not push the verdict to REJECT or UNVERDICTED; it strengthens the need for the reproducibility and robustness analyses that the CONDITIONAL verdict already requests. Hence I recommend no change to the reader's verdict.","tokens_in":32116,"tokens_out":7678,"duration_ms":103360,"concrete_test":"For the central case U/t=-4, μ/t=0.25, recompute the power-law fits of nc(L) at each temperature using varying lower cutoffs L_min = 12, 16, 20, 24, 28, 32, and extract η(T; L_min). Then perform the linear η→ηc extrapolation for each L_min to obtain TBKT(L_min). If TBKT varies by more than twice the quoted statistical error (≈0.0014) between L_min=20 and L_min=32, or if η(T) shows a monotonic drift with L_min, the algebraic scaling of nc is not asymptotic in the studied range and the moderate-size accuracy claim is not supported. The same test can be applied to the data-collapse and TBKT(L) methods by repeating them with L≥32 only.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that condensate fraction nc provides accurate BKT temperatures from moderate system sizes rests on the assumption that nc obeys the same finite-size scaling as the on-site pairing correlator: nc ∝ L^{-η} below TBKT and the BKT form above, with negligible subleading corrections for L≥20. The only theoretical support is the SM derivation of a subleading L^{-2} term for ⟨Δ²⟩ (SM Eq. 13); no analogous derivation is given for nc. The main-text argument that nc includes Cooper pairs of all sizes explains why local-pair fluctuations are less severe, but it does not prove the absence of other subleading terms, e.g., from momentum-space discretization or from the normalization λmax/(N/2).\n\nAll three TBKT extraction methods — the η(T) linear extrapolation, the data collapse of Fig. 2, and the TBKT(L) logarithmic fit of Fig. 3 — use this scaling assumption. Therefore, if the extracted η(T) values from nc are biased by residual finite-size corrections, all three estimates shift together, and the quoted errors (e.g., TBKT/t = 0.1420(7)) do not include this systematic uncertainty. The paper itself notes finite-size effects in nc for low filling (SM Sec. IIIE, L≤28), yet no L_min-stability analysis is presented for the central parameter set. Without quantifying the dependence of η and TBKT on the chosen fitting range, the statement that L=20 suffices is not established.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the 2D attractive Fermi-Hubbard model with finite-temperature auxiliary-field quantum Monte Carlo, reaching system sizes up to 64x64 (4096 sites). It defines the condensate fraction n_c from the largest eigenvalue of the momentum-space pairing matrix and shows numerically that n_c scales algebraically with L below the BKT transition and exponentially above it, with markedly smaller finite-size effects than the on-site pairing correlator. From the n_c scaling the paper extracts the BKT transition temperature for U/t=-4, mu/t=0.25 by three methods: linear extrapolation of the exponent eta(T) to eta_c=1/4, data collapse of n_c L^{1/4}, and extrapolation of finite-size transition temperatures T_BKT(L), yielding consistent values around T_BKT/t=0.142. A specific-heat anomaly with a peak near 1.10 T_BKT is also reported, and the methods are argued to be applicable to other 2D fermionic and bosonic superfluids.","tokens_in":32412,"tokens_out":5420,"duration_ms":46900,"significance":"If the central scaling claim holds, the condensate fraction provides a practical and accurate observable for determining BKT transitions in 2D correlated systems, with substantially smaller finite-size corrections than the widely used on-site pairing correlator. The paper's strengths include numerically exact simulations at unprecedented lattice sizes for this model, small statistical errors, extensive supplementary results for multiple fillings and interaction strengths, and a consistent cross-check of T_BKT across three independent fitting schemes. The claimed general applicability to other fermionic and bosonic systems is plausible but remains an extrapolation from the Hubbard-model evidence presented.","major_comments":[{"comment":"The central claim that L>=20 is sufficient to determine T_BKT accurately is not quantitatively tested. The supplemental material explicitly states that for U/t=-4, mu/t=1.25 the condensate fraction shows finite-size effects for L<=28 (SM Sec. IIIE), yet for the main parameter set no stability analysis with respect to the minimum system size L_min is shown. All three T_BKT estimates use the scaling assumption with a chosen L_min (L>=20 for n_c in Fig. 1(a), L=20-64 for the data collapse in Fig. 2(a), and the n_c(T) inflection points in Fig. 3), so a systematic bias from residual finite-size corrections would shift all three estimates together. The paper should report how eta and T_BKT vary as L_min is changed (for example L_min=16, 20, 24, 28) and include this spread in the final uncertainty of T_BKT=0.1420(7).","section":"Fig. 1(a); SM Sec. IIIE"},{"comment":"The identification of the inflection point of n_c(T) as the finite-size BKT transition temperature T_BKT(L) is heuristic and is not derived or independently justified. The agreement with the eta(T) scaling and data-collapse results is encouraging, but the logarithmic-correction extrapolation T_BKT(L)=T_BKT(infinity)+a/(ln bL)^2 and the associated claim that the logarithmic correction is 'confirmed' rest entirely on this mapping. The paper should either provide a derivation or a systematic comparison with alternative finite-size definitions of T_BKT(L), such as the crossing point of n_c L^{1/4} curves at eta=1/4. In addition, the reported error bars for T_BKT(infinity) from Fig. 3(b) exclude the fitting choice of fixing b=1; the inset with b free is mentioned as consistent, but the sensitivity should be quantified.","section":"Fig. 3(a) and (b)"},{"comment":"The paper derives the subleading L^{-2} correction for the on-site pairing correlator (SM Eq. 13), but no analogous derivation is given for the condensate fraction n_c. The main-text argument that n_c includes Cooper pairs of all sizes explains physically why local-pair fluctuations are reduced, but it does not rule out subleading corrections from momentum-space discretization or from the normalization lambda_max/(N/2). Because the central advantage of n_c over the pairing correlator is precisely its small finite-size corrections, this assumption is load-bearing for all three T_BKT estimates. The authors should either supply a derivation of the leading and subleading finite-size behavior of n_c or perform a direct multi-term fit (for example n_c = A L^{-eta} (1 + c L^{-2} + ...) with varying L_min) to demonstrate the absence of significant subleading terms for L>=20.","section":"SM Eq. (13) and main text around Fig. 1"},{"comment":"The linear extrapolation of eta(T) to eta_c=1/4 is described as a limit approached 'from below', but the text reports eta=0.304(2) at T/t=0.145, which is above the stated transition interval 0.140<T_BKT/t<0.145. If T/t=0.145 is included in the linear fit, the result would be biased by data that the paper itself says should be in the crossover regime to exponential decay. The manuscript should state exactly which temperatures enter the linear fit and justify the exclusion or inclusion of T/t=0.145.","section":"Inset of Fig. 1(a)"}],"minor_comments":[{"comment":"There are numerous typographical errors and misspellings, including 'attracive', 'diffrent', 'the the', 'inlcuding', 'questiones', 'capabilty', 'seperated', and 'tunning'. A careful proofreading pass is needed.","section":"Throughout"},{"comment":"The main text fixes b=1 in the formula T_BKT(L)=T_BKT(infinity)+a/(ln bL)^2, while the inset shows a fit using c+a/(ln bL)^2. The relationship between these parametrizations and why b=1 is the preferred choice should be stated more clearly.","section":"Fig. 3(b) and inset"},{"comment":"In the fixed-filling results, some statements such as 'the result of T_BKT/t ~ 0.10 should be obtained' for L=20 are not tied to a specific figure or table; giving the actual fitted value and its uncertainty would make the comparison quantitative.","section":"Sec. IV (SM)"},{"comment":"The phrase 'condensate fractionscaling' is broken across a line in the abstract and appears without a space in several places; this is a formatting issue, not a substantive one.","section":"Abstract and introduction"}],"recommendation":"major_revision","confidential_remarks":"The paper is well within the scope of cond-mat.str-el and presents high-quality numerical data. The main concern is that the central claim of negligible finite-size effects in the condensate fraction is supported by observation but lacks a systematic stability analysis and a derivation of the subleading corrections. These issues are fixable with additional analysis and should not require new physics. I do not see grounds for rejection, but the current uncertainty estimates for T_BKT do not yet account for the systematic choices in the fitting procedure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid, careful numerical study. The core claim — that the condensate fraction nc has much weaker finite-size effects than the on-site pairing correlator across the BKT transition — is well supported by the data. The AFQMC simulations reach 64×64, an order of magnitude beyond earlier work, and the algebraic-versus-exponential scaling crossover in nc is clean, with exponents η consistent between nc and ⟨Δ²⟩ in the large-L regime. The agreement of TBKT from the η(T) extrapolation (0.1420(7)), data collapse (0.1413(6)), and TBKT(L) log-correction (0.1418(40)) is reassuring, and the specific-heat anomaly at ~1.1 TBKT is a useful by-product consistent with XY-model expectations.\n\nThe conditional verdict is fair. My main reservation matches the stress-test note: the claim that L=20 suffices is empirical and never stress-tested by varying the L_min of the fits for the central parameter set. The paper's own SM shows nc has some finite-size effects at low filling (μ/t=1.25, L≤28), so the headline statement is parameter-dependent and should be softened. That matters because all three extraction methods lean on the same scaling assumption; residual corrections would shift all TBKT estimates together and the quoted errors would miss the systematic bias.\n\nThe inflection-point definition of TBKT(L) is heuristic — no derivation that the peak of -dnc/dT tracks the finite-size transition — and fixing b=1 in the log-correction fit is a choice that affects the extrapolation. Since the other two methods agree, this is not load-bearing. The paper would be stronger with an explicit L_min-stability analysis, fuller specification of the data-collapse fits (polynomial degree, x-range), and released code and data. The SM is extensive, and the superfluid-density cross-check at μ/t=0.60 helps.\n\nThe circularity worry is mild: using BKT scaling forms to fit TBKT is standard practice, and the algebraic/exponential observation is the genuinely new result.\n\nThis paper is for people running numerics on 2D superconductivity and superfluidity, and it deserves a serious referee. Send it to peer review with a request for the robustness analysis and reproducibility details. I would cite it if I were extracting BKT temperatures in 2D fermionic models.","headline":"Solid AFQMC study showing condensate fraction is a far better finite-size probe of the BKT transition than the on-site pairing correlator; central claim holds up, though the L≥20 shortcut needs an explicit L_min robustness test.","tokens_in":32940,"tokens_out":4674,"would_cite":true,"duration_ms":36086,"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":"Condensate fraction finds the BKT transition from small lattices","keywords":["condensate fraction","Berezinskii-Kosterlitz-Thouless transition","attractive Hubbard model","auxiliary-field quantum Monte Carlo","finite-size scaling","two-dimensional superconductivity","specific heat anomaly"],"falsifier":"Compute the condensate fraction for a model with an exactly known BKT transition, such as the 2D XY model or the 2D Bose–Hubbard model at a known filling, and test whether the inflection-point scheme with $b=1$ reproduces the accepted $T_\\mathrm{BKT}$ from lattices of size $L=20$ to $64$; a systematic offset would indicate the heuristic mapping is biased.","tokens_in":31870,"feed_emoji":"🌀","tokens_out":5181,"duration_ms":32864,"temperature":0.7,"pith_summary":"This paper establishes the condensate fraction—the fraction of fermions that form condensed pairs—as a sharp probe of the Berezinskii–Kosterlitz–Thouless (BKT) transition in two-dimensional superconductors and superfluids. Using the 2D attractive Fermi–Hubbard model and numerically exact auxiliary-field quantum Monte Carlo simulations on lattices up to 64×64, the author shows that the condensate fraction decays algebraically with system size below the transition and exponentially above it. Because this quantity has far weaker finite-size effects than the commonly used on-site pairing correlator, the BKT transition temperature can be determined accurately from lattices as small as L=20. The paper reports $T_\\mathrm{BKT}/t = 0.1420(7)$ for $U/t=-4$, $\\mu/t=0.25$, and finds a specific-heat peak at about $1.10\\,T_\\mathrm{BKT}$.","feed_headline":"Condensate fraction resolves 2D superfluid transition at L=20","feed_subtitle":"Algebraic-to-exponential scaling in condensate fraction yields BKT temperature with far smaller finite-size errors than standard pairing…","key_machinery":"The load-bearing object is the condensate fraction $n_c = \\lambda_\\mathrm{max}/(N/2)$, the largest eigenvalue of the momentum-space pairing matrix $M_{kk'} = \\langle \\hat{\\Delta}_k^\\dagger \\hat{\\Delta}_{k'} \\rangle - \\langle c^\\dagger_{k\\uparrow}c_{k\\uparrow}\\rangle\\langle c^\\dagger_{-k\\downarrow}c_{-k\\downarrow}\\rangle$ divided by the number of pairs. Its eigenvector encodes the full pair wave function, including nonlocal Cooper pairs. The argument runs on the algebraic-to-exponential crossover of $n_c$ with linear system size $L$: $n_c \\propto L^{-\\eta}$ below the BKT temperature with $\\eta$ reaching $1/4$ at the transition, and $n_c \\propto e^{-\\alpha L}$ above it. That crossover, rather than the noisy pairing correlator, is what locates the transition.","core_discovery":"The central discovery is that the condensate fraction $n_c$, defined from the largest eigenvalue of the momentum-space pairing matrix, obeys the same algebraic size scaling as the pairing correlator in the quasi-ordered phase, $n_c \\propto L^{-\\eta}$ with $\\eta \\to 1/4$ at the transition, but with much smaller subleading corrections. Formally, the on-site pairing correlator carries a subleading $L^{-2}$ correction that dominates at the moderate sizes accessible to simulation, whereas $n_c$ includes Cooper pairs of all sizes and therefore scales cleanly from $L \\approx 20$ upward. The crossover between algebraic and exponential scaling brackets the transition, and extrapolating the extracted exponent $\\eta(T)$ to $\\eta_c=1/4$ yields the transition temperature; the same $T_\\mathrm{BKT}$ is obtained from data collapse and from the logarithmic finite-size scaling of $T_\\mathrm{BKT}(L)$. The specific heat shows a peak slightly above the transition, at about $1.10\\,T_\\mathrm{BKT}$.","pith_inferences":["Because $n_c$ collects pairing weight at all momenta, the small finite-size corrections observed here may be generic for order parameters defined from the leading eigenvalue of a reduced density matrix, not special to the Hubbard model; testing this on the 2D XY model would separate the two possibilities.","In dilute 2D Fermi gases, where both the pairing correlator and superfluid density vanish in finite systems, $n_c$ may be the only practical observable; a two-step extrapolation in $L$ and particle number, as outlined in the paper, could give the first unbiased $T_\\mathrm{BKT}$ in the continuum limit.","The inflection-point method, if validated against exactly solvable BKT models, could be automated as a black-box estimator for transition temperatures in future tensor-network and Monte Carlo studies of 2D superconducting and superfluid models."],"forward_implications":["Accurate BKT temperatures can be extracted from lattices as small as $L=20$, roughly an order of magnitude cheaper than the sizes needed for the on-site pairing correlator.","The scheme transfers directly to other 2D fermionic systems, including spin-orbit-coupled models and dilute Fermi gases, where the pairing correlator and superfluid density become vanishingly small.","For 2D bosonic systems, the condensate fraction from the single-particle density matrix provides a competitive alternative to superfluid-density-based methods.","The specific-heat anomaly at about $1.1\\,T_\\mathrm{BKT}$ gives experimentalists a precursor signature for locating the BKT transition in optical lattices.","The logarithmic correction $T_\\mathrm{BKT}(L) = T_\\mathrm{BKT}(\\infty) + a/(\\ln bL)^2$ is confirmed for a correlated fermion system, supporting the BKT finite-size scenario."],"supporting_citations":[{"why":"BKT theory supplies the algebraic-decay exponent $\\eta_c=1/4$ and the vortex-binding picture that the scaling analysis leans on.","marker":"[1–3]"},{"why":"Prior AFQMC studies of the attractive Hubbard model that used the on-site pairing correlator and reported elevated $T_\\mathrm{BKT}$ from small lattices; this paper compares against those baselines.","marker":"[33–41]"},{"why":"Provides the superfluid-density formula from dynamic current-current correlations used as an independent check of $T_\\mathrm{BKT}$ up to $L=32$.","marker":"[44]"},{"why":"Establishes the logarithmic finite-size scaling $T_\\mathrm{BKT}(L) = T_\\mathrm{BKT}(\\infty)+a/(\\ln bL)^2$ used for the thermodynamic-limit extrapolation.","marker":"[24]"},{"why":"Supplies the data-collapse scaling form $x = L \\exp[-A(T/T_\\mathrm{BKT} - 1)^{-1/2}]$ used for the collapse fits.","marker":"[30,36,41]"},{"why":"Prior precision many-body study of the 2D Fermi gas that provides the context and two-step extrapolation strategy for dilute systems.","marker":"[43]"},{"why":"Direct experimental observation of nonlocal fermion pairing, the physical motivation for why the on-site correlator is incomplete.","marker":"[73]"}],"fun_headline_variants":["Condensate fraction scaling sharpens BKT transition detection","Algebraic-exponential crossover in condensate fraction reveals BKT point","Less finite-size noise: condensate fraction determines 2D BKT transition","Efficient BKT transition extraction via condensate fraction scaling","Condensate fraction: a superior probe for 2D superfluid critical point"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper identifies the inflection point of the condensate-fraction temperature curve as the finite-size BKT transition temperature and extrapolates it using $T_\\mathrm{BKT}(L) = T_\\mathrm{BKT}(\\infty) + a/(\\ln bL)^2$ with $b$ fixed to 1; this mapping and fitting choice are heuristic and, if inappropriate, would shift the reported $T_\\mathrm{BKT}$.","fun_headline_variants_meta":{"raw":{"variants":["Condensate fraction scaling sharpens BKT transition detection","Algebraic-exponential crossover in condensate fraction reveals BKT point","Less finite-size noise: condensate fraction determines 2D BKT transition","Efficient BKT transition extraction via condensate fraction scaling","Condensate fraction: a superior probe for 2D superfluid critical point"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000631,"raw_usage":{"total_tokens":2953,"prompt_tokens":1025,"completion_tokens":1928,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":641,"completion_tokens_details":{"reasoning_tokens":1836}},"tokens_in":641,"tokens_out":1928,"duration_ms":8274,"temperature":1.0,"reasoning_tokens":1836,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:47:46.835494+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the condensate fraction for a model with an exactly known BKT transition, such as the 2D XY model or the 2D Bose–Hubbard model at a known filling, and test whether the inflection-point scheme with $b=1$ reproduces the accepted $T_\\mathrm{BKT}$ from lattices of size $L=20$ to $64$; a systematic offset would indicate the heuristic mapping is biased.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the superfluid-density formula from dynamic current-current correlations used as an independent check of $T_\\mathrm{BKT}$ up to $L=32$."},{"cited_title":"Filinov, N","cited_arxiv_id":null,"evidence_quote":"Establishes the logarithmic finite-size scaling $T_\\mathrm{BKT}(L) = T_\\mathrm{BKT}(\\infty)+a/(\\ln bL)^2$ used for the thermodynamic-limit extrapolation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prior precision many-body study of the 2D Fermi gas that provides the context and two-step extrapolation strategy for dilute systems."},{"cited_title":"Hartke, B","cited_arxiv_id":null,"evidence_quote":"Direct experimental observation of nonlocal fermion pairing, the physical motivation for why the on-site correlator is incomplete."}],"review_version":1}