{"id":"03abdc73-acce-4194-be92-c963387388b3","arxiv_id":"2501.04452","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Partition function zeros show that finite-size Blume-Capel systems on complete graphs converge very slowly to mean-field critical behavior, with effective exponents still below asymptotic values at N=1500.","lead":"This paper studies where the partition function of the Blume-Capel spin model on a complete graph vanishes in complex temperature, magnetic field, and crystal field planes. It finds that even for systems with up to 1500 spins, the effective critical exponents remain far from the infinite-size mean-field values.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Concern: the non-asymptotic claim rests on pure-power-law fits to the first Fisher zero at N≤1500 without corrections to scaling, logarithmic terms, or error bars; in a mean-field/tricritical setting these fitted exponents can be pre-asymptotic artifacts.","rationale":"The reader's weakest assumption correctly identifies the key vulnerability: the central claim that finite-size effective exponents differ from mean-field asymptotics is inferred from pure-power fits over a limited range of N, with no corrections-to-scaling, no logarithmic terms, and no error bars. My stress-test pass confirms this is the most load-bearing concern. The paper's own evidence is partly self-qualifying: §5.1.3 reports that only a few Fisher zeros are visible near the tricritical point, so the first zero may be affected by numerical resolution, while other observables (crystal-field angles in Fig. 16, Lee-Yang exponents in Fig. 19) converge more convincingly to their mean-field values. This asymmetry makes the Fisher-zero exponent interpretation especially fragile. The proposed test directly checks whether the observed 0.35–0.46 exponents persist when higher-order and logarithmic corrections are included and when the system size is extended. If they do, the paper's qualitative message is supported; if they do not, the central claim should be softened. Other issues noted by the reader—the missing expanded-function approximation and the unfulfilled promise of an analytic confirmation of Fisher/crystal-field equivalence—are real but secondary; they do not change the conditional verdict. No stronger verdict movement is warranted because the exact integral representation is solid and the qualitative phenomenon of slow convergence is plausible.","tokens_in":13969,"tokens_out":4844,"duration_ms":52600,"concrete_test":"Use arbitrary-precision evaluation of Eq. (4) to compute the first Fisher zero at (T_t, Delta_t) for N = 2000, 3000, 5000, 10000, and at (T_c, Delta=0) for N = 2000, 5000. Fit log Im T1 versus log N with three forms: (i) pure power; (ii) power times correction A N^{-g}(1 + B N^{-theta}); (iii) power times logarithm A N^{-g}(ln N)^p, with bootstrap-based error bars. Repeat for the three largest original sizes. If the leading g from corrected forms approaches 0.5/0.66, the pure-power fits in §5.1.1/5.1.3 are unreliable evidence for non-asymptotic exponents; if g stays near 0.44/0.41, the claim is confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central assertion is supported mainly by the fits in §5.1.1 and §5.1.3: Im T1(N) ~ N^{-0.44} at the critical point and N^{-0.35}/N^{-0.41} at the tricritical point, compared with asymptotic exponents 1/2 and 2/3. The load-bearing assumption is that a single pure power law over N=50–1500 (or the last three points) captures the leading size dependence. This is insecure. At the tricritical point the quartic Landau term vanishes, so the finite-N expansion of Eq. (4) contains strong higher-order and possibly logarithmic corrections; Ref. [8] (cited by the authors) is precisely about logarithmic corrections to mean-field scaling. Fitting a pure power over a short N-window mixes the true exponent with these corrections, and no fits of the form T1 = A N^{-g}(1 + B N^{-theta}) or A N^{-g}(ln N)^p are reported. No error bars or stability checks are given. Moreover, unlike the Lee-Yang and crystal-field observables (Figs. 16, 19), very few Fisher zeros can be resolved near the tricritical point, so T1 from the Re/Im intersection search may carry an unidentified systematic offset. If a corrected fit still gives g about 0.41, the paper's conclusion stands; if g moves toward 2/3, the claim reduces to 'corrections are large,' not 'criticality is not asymptotic.' The promised expanded-function approximation in the Introduction is also never delivered, adding to the need for an independent numerical check.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies finite-size effects in the Blume-Capel model on a complete graph by locating zeros of the exact integral representation of the partition function, Eq. (4), in the complex temperature, crystal-field, and magnetic-field planes for system sizes up to N=1500. The thermodynamic-limit mean-field exponents are recalled, and the finite-size scaling of the first Fisher, crystal-field, and Lee-Yang zeros is compared with those asymptotic exponents. The central claim is that even at the largest accessible sizes the effective critical behaviour read from Fisher zeros is not asymptotic, e.g. T_1 ~ N^{-0.35} or N^{-0.41} at the tricritical point instead of N^{-2/3}. The paper also reports angle estimates, crystal-field-zero scaling along the critical line, and running Lee-Yang gap exponents.","tokens_in":14303,"tokens_out":3039,"duration_ms":34096,"significance":"If the central claim is correct, the paper provides a useful example in which partition-function-zero scaling at accessible finite sizes is governed by effective exponents that differ substantially from the known mean-field asymptotics, which matters for interpreting finite-size data in mean-field and tricritical systems. The work has clear strengths: the starting point is an exact integral representation, the thermodynamic-limit phase diagram is derived analytically, the standard zero-scaling relations are applied, and three different complex-field planes are compared systematically. However, the central quantitative conclusion rests on pure-power-law fits without corrections to scaling, logarithmic terms, or error bars, so the distinction between 'non-asymptotic effective behaviour' and 'large corrections to scaling' is not yet established.","major_comments":[{"comment":"The main claim that criticality is not asymptotic at the tricritical point is based on fitting Im T_1(N) to a pure power law, giving N^{-0.35} or N^{-0.41} versus the expected N^{-2/3}. No corrections to scaling, no logarithmic factors, and no error bars are reported. At a tricritical point the quartic Landau coefficient vanishes, so logarithmic and higher-order corrections are expected; Ref. [8] is precisely about such logarithmic corrections to mean-field scaling. The authors should test fits of the form T_1 = A N^{-g}(1 + B N^{-theta}) or T_1 = A N^{-g}(ln N)^p over the available N range, and report the resulting g-values with uncertainties. If g moves toward 2/3 under such fits, the conclusion reduces to 'corrections are large' rather than 'the asymptotic exponent is not approached.'","section":"§5.1.3, Eq. (4), Fig. 9"},{"comment":"The analogous pure-power-law analysis at the Ising critical point gives g=0.44 or 0.46 versus the expected 0.5, but the fitted range is limited (N=50 to 500, or only the three largest sizes) and no error bars or corrections-to-scaling fits are shown. The statement that the scaling 'approaches the theoretical value, even though slowly' is plausible, but the numerical support is not quantitative. Please provide stability checks under different N-windows and fits with corrections, and report confidence intervals for the effective exponents.","section":"§5.1.1, Fig. 6"},{"comment":"The running Lee-Yang exponent g_h is computed by repeated fits that add one system size at a time, but no uncertainties are quoted, and the conclusion that the exponents 'finally converge' toward 3/4 or 5/6 is drawn from trajectories without error bars. The qualitative trend is visible, but a quantitative statement about convergence requires either error bars or a stated criterion for convergence. This is important because the conclusions section asserts that Lee-Yang scaling 'matched theoretical predictions.'","section":"§5.3.1, Figs. 18-19"}],"minor_comments":[{"comment":"The Introduction promises an 'expanded function approximation' that 'simplifies the calculations without sacrificing accuracy,' but this method is never defined or used anywhere in the paper. The authors should either deliver it or remove the promise from the Introduction.","section":"Introduction, Section 5"},{"comment":"The claim that crystal-field zeros at the Ising critical point are not well defined because the scaling field is perpendicular to the tricritical point is stated very briefly; a short explanation of why this makes the zeros poorly defined would improve readability.","section":"§5.2.1"},{"comment":"The conclusion states that 'the equivalence between Fisher and crystal field zeros was confirmed analytically,' but the body of the paper only establishes that both obey the same expected scaling relation, Eq. (13) and Eq. (20). The wording is stronger than the presented derivation and should be adjusted.","section":"Conclusions"},{"comment":"The text writes T_1 ~ N^{0.35} in one place and T_1 ~ (1/N)^{0.35} in another. Since the quantity is the imaginary part of the first zero decreasing with N, the notation should be made uniform to avoid sign ambiguity.","section":"Notation in §5.1.3, Fig. 9"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and addresses a legitimate question about effective versus asymptotic finite-size scaling. The core issue is not circularity or a wrong exact calculation, but the lack of a robust scaling analysis behind the central 'not asymptotic' claim. The promised expanded-function approximation is absent, and the fit methodology needs error bars and corrections-to-scaling tests. I recommend major revision rather than rejection because the underlying exact representation and the qualitative observations are sound and the central claim could become convincing with additional analysis."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The thing to know: this paper maps out finite-size effective exponents from Fisher, Lee-Yang, and crystal-field zeros for the Blume-Capel model on a complete graph, and the qualitative message—that effective behavior is far from mean-field even at N=1500—is plausible and worth taking seriously. The exact integral representation of the partition function is clean, the numerical zero-finding is systematic, and the comparison of the three zero types at the same points in the phase diagram is genuinely useful. The observation about the pseudo-transition point at T=1/2, Delta=(ln2)/2, where the Lee-Yang scaling is unexpectedly good, is a nice detail.\n\nThat said, the central quantitative claim is not as solid as the text implies. The effective exponents are extracted from pure power-law fits over N=50–1500, with no corrections-to-scaling, no logarithmic terms, and no error bars. For a mean-field/tricritical model with known log corrections, that's a real gap. The fitted value g=0.41 at the tricritical point could move toward the asymptotic 2/3 once corrections are accounted for; the paper does not test that. I don't think the claim collapses—the systematic pattern across different zero types is suggestive—but it needs a stability analysis.\n\nTwo smaller problems. The introduction promises an \"expanded function approximation\" that never appears anywhere in the paper; either it's in a supplementary file or it should be cut. And the conclusion says the equivalence between Fisher and crystal-field zeros was \"confirmed analytically,\" but reading Sections 5.1 and 5.2, I see the equivalence assumed from the scaling-field argument, not derived or verified analytically. That overstatement should be fixed. No code or data files are shipped, which makes independent checks harder.\n\nNone of this is fatal. The paper's core contribution—a careful finite-size study of zeros in three complex planes for the complete-graph Blume-Capel model—is real and within the scope of a specialist journal. My own verdict is close to the reader's: conditional. If the authors add corrections-to-scaling fits, report uncertainty, and either deliver or remove the promised expansion, the results would be a solid reference for people doing zeros-based finite-size studies.\n\nI would send this to peer review. A good referee can push for the fixes without needing a re-derivation of the whole method.","headline":"A useful finite-size zeros map for the complete-graph Blume-Capel model, but the headline claim rests on power-law fits that need corrections-to-scaling and error bars before I'd trust the exponents.","tokens_in":14859,"tokens_out":1191,"would_cite":false,"duration_ms":13544,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["82B20","82B27","82B26"],"pacs":["05.70.Jk","05.50.+q"],"model":"deepseek-v4-flash","headline":"For the Blume-Capel model on a complete graph, partition-function zeros show that finite-size systems up to N=1500 still display effective critical exponents far from the thermodynamic-limit mean-field values.","keywords":["Blume-Capel model","partition function zeros","Fisher zeros","Lee-Yang zeros","crystal field zeros","complete graph","finite-size scaling","tricritical point"],"falsifier":"A concrete check: compute the first Fisher zero at the tricritical point ($T_t=1/3$, $\\Delta_t=2\\ln 2/3$) for $N=10^4$ to $10^6$ using the exact integral representation and fit $T_1(N)$ with the asymptotic form $a N^{-2/3}$ plus a logarithmic-correction term. If the effective exponent moves toward $2/3$ once those corrections are included, the paper's claim of non-asymptotic criticality at $N=1500$ is a finite-size fitting artifact; if the exponent remains near $0.4$ even at $N=10^6$, the claim is supported.","tokens_in":13735,"feed_emoji":"🧲","tokens_out":12564,"duration_ms":97242,"temperature":0.7,"pith_summary":"The paper examines the Blume-Capel model on a complete graph, where every spin interacts with every other spin, and asks how the finite-size system's critical behaviour is seen through the zeros of its partition function. It establishes that, even for graphs of 1500 sites, the scaling of the first Fisher zero gives effective exponents $T_1 \\sim N^{-0.35}$ or $T_1 \\sim N^{-0.41}$ near the tricritical point, whereas the exact thermodynamic limit predicts $T_1 \\sim N^{-2/3}$. At the ordinary critical point the effective exponent is about 0.44 rather than the mean-field value 0.5. Crystal-field zeros along the critical line are even more strongly skewed toward tricritical behaviour, with exponents between 0.62 and 0.69. If the paper is right, finite-size data at accessible sizes cannot be read directly as mean-field asymptotics; corrections or larger sizes are needed.","feed_headline":"Finite-size Blume-Capel criticality is not yet asymptotic at N=1500","feed_subtitle":"Fisher, crystal-field, and Lee-Yang zeros all show effective exponents that drift slowly toward mean-field values.","key_machinery":"The load-bearing object is the exact integral representation of the partition function, Eq. (4), which expresses $Z_N$ as a single integral over a collective variable $x$ for any finite $N$. On top of this, the paper uses partition-function zeros in three complex planes: Fisher zeros (complex temperature), crystal-field zeros (complex $\\Delta$), and Lee-Yang zeros (complex magnetic field). These zeros are found by solving the simultaneous equations $\\mathrm{Re}\\,Z=0$ and $\\mathrm{Im}\\,Z=0$ (for Lee-Yang zeros, $\\mathrm{Re}\\,Z=0$ alone). The argument is carried by scaling relations of the form $T_j \\sim (j/N)^{1/(2-\\alpha)}$, $H_j \\sim (j/N)^{\\beta\\delta/(2-\\alpha)}$, and the impact-angle relation (12), which connect the first zero's distance from the real axis to the critical exponents. By fitting the size dependence of the first zero over $N=20$ to $1500$, the paper extracts effective gap exponents and compares them with the mean-field expectations $1/2$, $2/3$, $3/4$ and $5/6$.","core_discovery":"The central claim is that finite-size criticality in this exactly solvable mean-field model is not asymptotic for the sizes studied. The authors analyse the exact integral representation of the partition function and locate its zeros in the complex temperature, crystal-field, and magnetic-field planes. Near the tricritical point at $(T_t,\\Delta_t)=(1/3,2\\ln 2/3)$, the first Fisher zero scales as $T_1 \\sim N^{-0.35}$ for all sizes up to $N=1500$, or $N^{-0.41}$ using only the three largest sizes, compared with the asymptotic mean-field exponent $2/3$. Near the Ising critical point $(T=2/3,\\Delta=0)$ the effective exponent is about 0.44, not 0.5. Crystal-field zeros behave well only near the tricritical point and scale with effective exponents close to $2/3$ even when the thermodynamic point is critical ($1/2$). Lee-Yang zeros in the complex magnetic field obey the Lee-Yang circle theorem and their gap exponent approaches $3/4$ at the critical point and $5/6$ at the tricritical point, but along most of the critical line the fitted exponents are skewed toward the tricritical value.","pith_inferences":["An implication the authors leave implicit is that the fitted pure power laws may be reconciled with the true mean-field asymptotics by including logarithmic corrections, which are known to appear in mean-field models above the upper critical dimension; if so, the effective exponents would be a crossover phenomenon rather than evidence of a new universality class.","A testable extension would be to repeat the zero analysis at substantially larger $N$ (for example $10^4$ to $10^5$) and check whether the effective Fisher-zero exponent increases monotonically toward $2/3$; the paper does not report such data.","The observation that crystal-field zeros are skewed toward tricritical behaviour even along the critical line suggests that fits of thermodynamic quantities near tricritical points in other spin-1 models may similarly be contaminated by tricritical scaling, a caution that could be tested on related models.","One could also fit the first Fisher zero with explicit correction terms, such as $T_1 = a N^{-2/3}(1 + b N^{-\\omega})$, to see whether the effective exponent 0.35 to 0.41 at $N \\le 1500$ is a leading-order artefact or a genuine slow crossover."],"forward_implications":["Finite-size scaling analyses that use only a few hundred or thousand sites for this model will report effective exponents that are not the true mean-field values; the gap is especially large at the tricritical point.","Crystal-field zeros are a reliable probe near the tricritical point but a poor one near the Ising critical point, because the scaling field is perpendicular to the tricritical point rather than to the critical point.","The point $(T=1/2,\\Delta=\\ln 2/2)$, which lies on the pseudo-transition line $\\Delta=T\\ln 2$, shows unusually fast convergence to the critical exponent and may be a useful reference point in finite-size studies.","Lee-Yang zeros confirm the Lee-Yang circle theorem on the complete graph: all zeros lie on the imaginary magnetic-field axis at the critical line and the tricritical point.","Effective exponents extracted from the first Fisher zero depend on which system sizes are included (0.35 versus 0.41 at the tricritical point), so reported exponents should carry this dependence."],"supporting_citations":[{"why":"Introduces the three-state spin Hamiltonian whose complete-graph version is the object of study.","marker":"[1]"},{"why":"Adds the single-ion anisotropy term that defines the Blume-Capel model and its tricritical phase diagram.","marker":"[2]"},{"why":"Supplies the scaling relations connecting partition-function-zero coordinates and angles to critical exponents, used throughout the analysis.","marker":"[5]"},{"why":"Provides the scaling relations for logarithmic corrections that frame the mean-field expectations in Eq. (12).","marker":"[8]"},{"why":"Gives the partition-function-zero method for complete-graph Ising systems that this paper extends to the Blume-Capel model.","marker":"[10]"},{"why":"Documents the poor behaviour of crystal-field zeros near the Ising critical point in Blume-Capel models, a key comparison in Section 5.2.","marker":"[14]"},{"why":"States the Lee-Yang circle theorem that locates zeros in the complex magnetic field plane, used to interpret the Lee-Yang data.","marker":"[16]"},{"why":"Introduces Fisher zeros in the complex temperature plane, the central tool for the temperature-plane analysis.","marker":"[18]"}],"fun_headline_variants":["Blume-Capel zeros: finite-size criticality not yet asymptotic at N=1500","Large-N zeros still show pre-asymptotic critical scaling in Blume-Capel","Mean-field limit elusive: Blume-Capel critical exponents drift slowly at N=1500","Fisher zeros show: Blume-Capel criticality still pre-asymptotic at N=1500","Pre-asymptotic criticality in Blume-Capel zeros persists up to N=1500"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a pure power law $T_1 \\sim N^{-g}$, fitted over the available range of $N$, captures the asymptotic exponent; if the data instead contain logarithmically suppressed corrections, the fitted effective exponents could be artifacts of omitting them.","fun_headline_variants_meta":{"raw":{"variants":["Blume-Capel zeros: finite-size criticality not yet asymptotic at N=1500","Large-N zeros still show pre-asymptotic critical scaling in Blume-Capel","Mean-field limit elusive: Blume-Capel critical exponents drift slowly at N=1500","Fisher zeros show: Blume-Capel criticality still pre-asymptotic at N=1500","Pre-asymptotic criticality in Blume-Capel zeros persists up to N=1500"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001431,"raw_usage":{"total_tokens":5773,"prompt_tokens":947,"completion_tokens":4826,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":563,"completion_tokens_details":{"reasoning_tokens":4704}},"tokens_in":563,"tokens_out":4826,"duration_ms":30231,"temperature":1.0,"reasoning_tokens":4704,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:32:51.344630+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check: compute the first Fisher zero at the tricritical point ($T_t=1/3$, $\\Delta_t=2\\ln 2/3$) for $N=10^4$ to $10^6$ using the exact integral representation and fit $T_1(N)$ with the asymptotic form $a N^{-2/3}$ plus a logarithmic-correction term. If the effective exponent moves toward $2/3$ once those corrections are included, the paper's claim of non-asymptotic criticality at $N=1500$ is a finite-size fitting artifact; if the exponent remains near $0.4$ even at $N=10^6$, the claim is supported.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces the three-state spin Hamiltonian whose complete-graph version is the object of study."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Adds the single-ion anisotropy term that defines the Blume-Capel model and its tricritical phase diagram."},{"cited_title":"Itzykson, R.B","cited_arxiv_id":null,"evidence_quote":"Supplies the scaling relations connecting partition-function-zero coordinates and angles to critical exponents, used throughout the analysis."},{"cited_title":"Kenna, D","cited_arxiv_id":null,"evidence_quote":"Provides the scaling relations for logarithmic corrections that frame the mean-field expectations in Eq. (12)."},{"cited_title":"Partition function zeros for the Ising model on complete graphs and on annealed scale-free networks","cited_arxiv_id":null,"evidence_quote":"Gives the partition-function-zero method for complete-graph Ising systems that this paper extends to the Blume-Capel model."},{"cited_title":"Critical and tricritical singularities from small-scale Monte Carlo simulations: the Blume–Capel model in two dimensions","cited_arxiv_id":null,"evidence_quote":"Documents the poor behaviour of crystal-field zeros near the Ising critical point in Blume-Capel models, a key comparison in Section 5.2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"States the Lee-Yang circle theorem that locates zeros in the complex magnetic field plane, used to interpret the Lee-Yang data."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduces Fisher zeros in the complex temperature plane, the central tool for the temperature-plane analysis."}],"review_version":1}