{"id":"5d55fde9-2811-40ba-97c4-6f85332eac1d","arxiv_id":"2506.14738","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For radially symmetric potentials at beta = 2, the log N coefficient in the hard-wall partition function is -1/4 for an annulus and -1/3 for a disk when the wall lies strictly inside the droplet, instead of the usual -1/12.","lead":"This paper computes the exact large-N expansion of the partition function for a two-dimensional Coulomb gas with a radially symmetric potential and a hard wall in the solvable beta = 2 case. It finds that the coefficient of the log N term changes depending on where the hard wall sits relative to the droplet, a new universality result.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unproven relation (5.1) is the only bridge from the annulus theorem to the disk theorem; the advertised -1/3 log N coefficient is not established.","rationale":"The reader identified the weakest assumption as Eq. (5.1), and the stress-test pass confirms that this is the single most load-bearing concern in the paper. The annulus results (Theorem 2.1) are supported by detailed Laplace-method and Euler-Maclaurin computations, with explicit special-function cross-checks. However, the disk result (Theorem 2.3) -- which supplies the headline -1/3 log N coefficient when 1>eta>0 -- is obtained purely from the asserted equality of disk/annulus differences in Eq. (5.1). The paper provides no derivation of this equality, and the short proof in Section 5 merely asserts that the disk-specific alteration is independent of tau0. That independence is not obviously true, because tau0 controls the summation range over which the disk-specific asymptotics of u_j apply. A direct derivation would settle the matter; the known quadratic case provides a consistency check but not a proof of universality. Therefore the conditional verdict is appropriate and should not be changed: the annulus part may be accepted, but the disk part requires either a proof of (5.1) or a direct computation. No ad hominem or stylistic objection is raised; the concern is purely about the validity of the central disk claim.","tokens_in":29096,"tokens_out":4230,"duration_ms":40086,"concrete_test":"Derive Theorem 2.3 without invoking (5.1): for r0=0, split the sum (1.18) at j roughly N^{1-epsilon}, use Proposition 3.11 for the small-j block and the standard interior expansion (with the Section 3.3 modification) for the rest, apply the Euler-Maclaurin formula, and verify that the total equals (2.17). Independently, test numerically with a non-quadratic radial potential having q'(0)=0 and r1>1, e.g. q(r)=r^2 + lambda r^4 with lambda chosen so eta=1-q'(1)/2 is in (0,1); compute u_j via quadrature for N=100, 200, 400, 800, sum log u_j, and extract the log N coefficient; if it is not -1/3 within fit error, Eq. (5.1) is false.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 5 derives Theorem 2.3 entirely from Eq. (5.1): Z_s^disk - Z_s^annulus = Z_h^disk - Z_h^annulus, justified only by 'Since this common alteration is independent of tau0'. This is the sole bridge from the proven annulus expansion to the disk, and it is not a proof. First, as written the identity is dimensionally suspect: Z_s and Z_h are partition functions, not logarithms, and the left-hand side is used as if it were the log-expansion difference read off from [11]. Second, even interpreting both sides as log Z differences, tau0-independence is not automatic: tau0 determines the upper endpoint N*tau0 of the j-sum in Section 4, and the disk-specific small-tau expansion of u_j (Proposition 3.11) contributes only for j up to O(N^{1-epsilon}); the contribution of that block to the Euler-Maclaurin summation must be computed explicitly to see whether it cancels or reproduces the annulus terms. No such computation is supplied. The quadratic example q(r)=a^2 r^2 checks the final coefficient -1/3, but that is one special case; the universal claim for all radial potentials satisfying (1.16) remains unsupported. Hence the disk log N coefficient -1/3, and the constant terms zeta'(-1) and F_D - F_{S cap D}, are conditional on (5.1).","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the large-N asymptotics of the partition function of the beta=2 two-dimensional Coulomb gas in the unit disk with a radially symmetric potential and a hard wall. Using the exact factorization log(Z_N^h/(2 pi)^N) = sum_{j=0}^{N-1} log u_j, the authors apply Laplace's method and Euler-Maclaurin summation to obtain expansions in several configurations. Theorem 2.1 gives the annulus case with hard wall inside the droplet (eta in (0,1)) or at the outer edge (eta=0), including log N coefficients -1/4 or 0 and universal constants alpha, beta, gamma. Theorem 2.3 asserts the disk case follows from the annulus case by adding -1/12 log N + zeta'(-1) + F_D - F_{S cap D}; this transfer rests on the unproven identity (5.1). Theorem 2.5 treats an annulus whose inner edge lies outside the hard wall.","tokens_in":29371,"tokens_out":6018,"duration_ms":56130,"significance":"If fully correct, the paper would give the first systematic hard-wall corrections to the free energy expansion for non-quadratic radial potentials, confirming and extending predicted universal log N and sqrt N terms. The annulus analysis is detailed and self-contained, with explicit universal constants (2.3)-(2.4), incomplete-gamma cross-checks in Examples 3.3, 3.5, 3.10, and agreement with the known quadratic example (1.13). However, the advertised disk result, in particular the -1/3 log N coefficient, is not established because Eq. (5.1) is unproven and appears dimensionally inconsistent. Thus the paper's central novelty for the disk is conditional. The strengths of the annulus part, and the clarity of the universal constants, make this a worthwhile contribution once the disk bridge is proved or replaced.","major_comments":[{"comment":"The disk result Theorem 2.3 rests entirely on the unproven identity Z_s^disk - Z_s^annulus = Z_h^disk - Z_h^annulus. As written this identity is dimensionally inconsistent: the left and right sides are partition functions, while the following sentence reads off a logarithmic-expansion difference from [11]. Even after replacing Z by log Z, the assertion that the disk/annulus difference is independent of the hard wall is not justified: tau0 determines the upper endpoint floor(N tau0) of the j-sum, and the small-tau block j = O(N^{1-epsilon}) governed by Proposition 3.11 changes when r0 changes from 0 to >0; its contribution to the Euler-Maclaurin summation must be computed explicitly to decide whether it cancels the annulus terms. Since no such computation is supplied, the advertised -1/3 log N coefficient, and the constant terms zeta'(-1) + F_D - F_{S cap D}, are conditional. The quadratic check in Example 2.4 is one special case and cannot establish the universal claim for all radial potentials satisfying (1.16).","section":"Section 5, Eq. (5.1)"},{"comment":"The proofs of Theorem 2.1 combine asymptotic formulas for u_j with Euler-Maclaurin summation in regimes separated by the cutoffs Delta_N and delta_N. The final results are stated with o(1) remainders after cancellation of terms depending on Delta_N, xi_N, xi*_N and {N tau0}. Uniform error bounds are not supplied, and the o(1) errors from Corollaries 3.2, 3.4, 3.7 and Proposition 3.9 are not tracked through summations over O(N) terms. The incomplete-gamma examples are reassuring, but the theorem as stated needs either explicit remainder control or a precise citation of where such control is established.","section":"Section 4, Propositions 4.1-4.2"}],"minor_comments":[{"comment":"In the discussion after Eq. (2.25), 'eliminate alpha_in, alpha_in' should read 'eliminate alpha_in, alpha_out'.","section":"Section 2.4"},{"comment":"The numerical verification statements are not accompanied by any displayed data; adding a table of computed versus predicted values would let the reader assess the order of the remainder and the claimed agreement.","section":"Examples 2.2, 2.4, 3.5"},{"comment":"The displayed formula (3.14) is extremely difficult to read in its current typesetting; restructuring it with named auxiliary quantities would improve usability.","section":"Proposition 3.1, Eq. (3.14)"},{"comment":"The notation Z_s^annulus and Z_h^annulus in (2.17) is confusing because the annulus partition function in Theorem 2.1 has r0>0 and r1>=1, while the disk setting has r0=0; the relation should specify exactly which parameters are held fixed in the comparison.","section":"Theorem 2.3"}],"recommendation":"major_revision","confidential_remarks":"The paper's main new result for the disk is conditional on Eq. (5.1), which is not a minor gap: it is the only bridge to the title's physical setting. If the authors can prove (5.1) or provide a direct derivation of Theorem 2.3, the paper will be suitable for publication. I would not recommend rejection because the annulus part is substantial and the claimed disk result is likely true, but the current manuscript oversells Theorem 2.3 as proven. Please ask the authors to either add a proof of (5.1) or downgrade the claim to a conjecture supported by numerics."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the bottom line: the annulus expansion in Section 4 is careful, detailed, and probably right. The disk result in Section 5 is not established. The whole of Theorem 2.3 rests on Eq. (5.1), which is not a proof, and as written it mixes partition functions and logarithms.\n\nWhat is genuinely new: hard-wall configurations with the wall strictly inside the droplet, the -1/4 log N coefficient in the in/out annulus (0<eta<1), and the universal constants alpha/beta/gamma defined by convergent integrals. The authors are honest about what comes from prior work: the Euler-Maclaurin machinery is from [11], the gamma_out constant from [25]. The cross-checks against incomplete gamma examples (Examples 2.2, 3.5) are a real strength; the constants are computed, not fitted.\n\nThe problem: Eq (5.1) says Z_s^disk - Z_s^annulus = Z_h^disk - Z_h^annulus, and then the left side is \"read off from [11]\" as -1/12 log N + zeta'(-1) + F_D - F_{S∩D}. That is a difference of logarithms, not a difference of partition functions. Even if you silently read (5.1) as log differences, the justification \"this common alteration is independent of tau0\" is not a derivation. The small-tau block j up to O(N^{1-epsilon}) is handled differently when r0=0 (Proposition 3.11), and no computation shows that its contribution to the Euler-Maclaurin sum cancels in the same way. So the -1/3 log N coefficient and the disk constant terms are conditional. The quadratic example confirms one case, but universality for all radial potentials satisfying (1.16) remains unsupported.\n\nLess important: the error terms are o(1) with no explicit bounds, and the numerical checks are asserted, not displayed. That would be fine if the disk result were clean; here it adds to the need for a fix.\n\nWho gets value: the OCP/normal matrix model community. Section 4 deserves to be published. Section 5 needs either a real proof or a downgrade to a conjecture. I'd send it to peer review with the referee instructed to put most of the weight on (5.1). After that is fixed, this is a solid paper.","headline":"The annulus expansion is solid, careful work; the disk -1/3 log N result is a conjecture resting on an unproven and dimensionally sloppy relation (5.1).","tokens_in":29920,"tokens_out":3978,"would_cite":true,"duration_ms":35965,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60B20","82D05","41A60","60G55"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a hard wall strictly inside a two-dimensional Coulomb droplet changes the $\\log N$ coefficient of the partition function to $-\\frac{1}{4}$ for an annulus and $-\\frac{1}{3}$ for a disk, independent of the radial…","keywords":["normal matrix model","two-dimensional Coulomb gas","hard wall constraint","partition function asymptotics","log N coefficient","radially symmetric potential","Laplace method","Euler-Maclaurin summation"],"falsifier":"For a potential with a disk droplet, such as $q(r)=a^2r^2$ with $0<a<1$, every $u_j$ is an incomplete gamma function, so $\\log Z_N^h$ can be computed numerically at large $N$; comparing the $-\\frac{1}{3}\\log N$ coefficient and the constant term $\\zeta'(-1)+F_D[Q]-F_{S\\cap D}[Q]$ with Theorem 2.3 would settle the disk claim. Independently, evaluating the left and right sides of equation (5.1) at finite $N$ for a family of radial potentials and checking whether their difference tends to zero would test the transfer relation directly.","tokens_in":28881,"feed_emoji":"🧱","tokens_out":9565,"duration_ms":75213,"temperature":0.7,"pith_summary":"This paper establishes the large-$N$ asymptotic expansion of the partition function for a two-dimensional Coulomb gas (equivalently, the normal matrix model at $\\beta=2$) when a hard wall confines the particles to the unit disk and the external potential is radial. The coefficient of $\\log N$ in the expansion is shown to be universal, independent of the potential, but it changes when the hard wall lies strictly inside the droplet rather than at its boundary. For an annular droplet with the wall strictly inside ($0<\\eta<1$), the coefficient is $-\\frac{1}{4}$; when the wall coincides with the droplet's outer boundary ($\\eta=0$), the $\\log N$ term is absent. For a disk droplet with the wall strictly inside, the paper derives $-\\frac{1}{3}\\log N$. These results sharpen the topological prediction $\\chi/12$ for the free-energy expansion by showing how an interior constraint modifies it.","feed_headline":"The log N term becomes -1/4 when a hard wall cuts inside a droplet","feed_subtitle":"For radial 2D Coulomb gases, the coefficient is universal and the disk case yields -1/3.","key_machinery":"The load-bearing object is the factorization $\\log(Z_N^h/(2\\pi)^N)=\\sum_{j=0}^{N-1}\\log u_j$, with $u_j=\\int_0^1 r^{2j+1}e^{-Nq(r)}dr$, which radial symmetry produces from the $N$-particle integral. Each $u_j$ is analyzed by Laplace's method with the critical point $r_\\tau$ fixed by $rq'(r)=2\\tau$, $\\tau=j/N$, and the hard wall enters through $\\tau_0$ with $r_{\\tau_0}=1$ and $\\eta=1-q'(1)/2$. The delicate part is the boundary layer $|\\tau-\\tau_0|$ of width $N^{-1/2}$, where the $u_j$ expansion is expressed through complementary error functions; those expressions are converted into sums by Euler-Maclaurin summation, producing the universal constants $\\gamma_{\\rm in},\\gamma_{\\rm out},\\alpha_{\\rm in},\\alpha_{\\rm out},\\beta_{\\rm in},\\beta_{\\rm out}$ as definite integrals.","core_discovery":"The central claim is Theorem 2.1: in the in/out annulus case $0<r_0<1\\le r_1$ with $\\eta=1-q'(1)/2\\in(0,1)$, one has $\\log(Z_N^h/(2\\pi)^N)= -N^2[I_{S\\cap D}[\\mu_Q]+\\eta q(1)] -\\frac{(1+\\eta)}2 N\\log N - N(\\cdots) -\\sqrt N(\\gamma_{\\rm in}+\\gamma_{\\rm out})\\sqrt{\\Delta Q(1)} -\\frac{1}{4}\\log N + F_{S\\cap D}[Q] -(\\alpha_{\\rm in}+\\alpha_{\\rm out})+(\\beta_{\\rm in}+\\beta_{\\rm out})\\frac{\\partial_r\\Delta Q(1)}{\\Delta Q(1)} + \\frac{\\Delta Q(1)}{4\\eta} -\\frac{1}{2}\\log\\eta+\\frac{1}{4}\\log(2\\pi\\Delta Q(1))+o(1)$. The $-\\frac{1}{4}\\log N$ coefficient is independent of $q(r)$; when $r_1=1$ ($\\eta=0$) no $\\log N$ term appears. Theorem 2.3 transfers the annulus result to the disk, giving $-\\frac{1}{3}\\log N$ for $1>\\eta>0$, together with a constant-term shift $\\zeta'(-1)+F_D[Q]-F_{S\\cap D}[Q]$. The expansions also identify universal constants at order $\\sqrt N$ and at $O(1)$ as definite integrals involving the complementary error function.","pith_inferences":["The pattern $\\{0,-\\frac{1}{4},-\\frac{1}{3}\\}$ for the $\\log N$ coefficient looks like one plus the number of hard edges that intersect the droplet; the authors leave open how to anticipate it, and a testable extension would be to compute the same coefficient for $\\beta\\neq2$ radial hard-wall gases to see whether the fractions become $\\beta$-dependent while remaining potential-independent.","Equation (5.1) suggests a transfer principle stronger than the paper's radial setting: the disk-annulus difference in the partition function may be independent of the confining wall for general, not necessarily radial, potentials. If so, the $-\\frac{1}{3}\\log N$ coefficient would extend to non-radial disk droplets with an interior hard wall.","The universal constants $\\gamma_{\\rm in},\\gamma_{\\rm out}$ are defined by erfc integrals that also encode surface tension; one might look for closed-form evaluations or relations connecting them to $\\alpha$ and $\\beta$, which would make the constant term fully explicit without numerical integration.","The breakdown of the $\\eta\\to0^+$ limit at order $\\sqrt N$ noted in the paper implies that small changes in wall position have non-perturbative effects in $N$; a natural check is to examine the crossover regime where $\\eta$ is taken to zero simultaneously with $N$, for instance $\\eta\\sim N^{-1/2}$."],"forward_implications":["In an annulus with a hard wall strictly inside the droplet, the $\\log N$ coefficient is $-\\frac{1}{4}$ for every admissible radial potential, whereas it vanishes when the wall sits exactly at the droplet boundary.","In a disk with the wall strictly inside, the $\\log N$ coefficient is $-\\frac{1}{3}$, and the disk-annulus difference at order $1$ is $\\zeta'(-1)+F_D[Q]-F_{S\\cap D}[Q]$, independent of the wall position.","At order $\\sqrt N$, the coefficient is universal up to the factor $\\sqrt{\\Delta Q(1)}$: it is $(\\gamma_{\\rm in}+\\gamma_{\\rm out})\\sqrt{\\Delta Q(1)}$ when the wall is inside, and $\\gamma_{\\rm out}\\sqrt{\\Delta Q(1)}$ when the wall is at the outer boundary, with the same constants appearing in known gap-probability expansions.","At order $1$, the constants $\\alpha_{\\rm in},\\alpha_{\\rm out},\\beta_{\\rm in},\\beta_{\\rm out}$ coincide with the constants in Mittag-Leffler-type large-gap asymptotics through the relations $\\beta_{\\rm in}=-\\frac{1}{2}\\alpha_{\\rm in}+\\frac{1}{2}\\log 2$ and $\\beta_{\\rm out}=-\\frac{1}{2}\\alpha_{\\rm out}+\\frac{1}{4}\\log\\pi$.","In the out-annulus case with the wall inside the inner radius ($\\eta>1$), the expansion has no $\\sqrt N$ or $\\log N$ term, while at $\\eta=1$ (wall exactly at $r_0=1$) the coefficient $-\\frac{1}{4}\\log N$ reappears with a $\\sqrt N$ term."],"supporting_citations":[{"why":"Supplies the soft-wall radial partition function expansion and the functionals $F_D[Q]$, $F_{S\\cap D}[Q]$ from which the disk-annulus difference is read off.","marker":"[11]"},{"why":"Source of the $\\chi/12$ log-term prediction and of the surface-tension constant $\\gamma_{\\rm in}$ for a confined disk plasma.","marker":"[28]"},{"why":"Provides the large-gap asymptotics whose order-one constants $\\tilde\\beta_{\\rm in}$, $\\tilde\\beta_{\\rm out}$ are identified with $\\beta_{\\rm in}$, $\\beta_{\\rm out}$.","marker":"[19]"},{"why":"First implementation of the Laplace-plus-Euler-Maclaurin strategy for hard-wall gap probabilities, giving the combination $\\gamma_{\\rm in}+\\gamma_{\\rm out}$ at order $\\sqrt N$.","marker":"[25]"},{"why":"Exact solution of the two-dimensional one-component plasma on a disk at $\\beta=2$, the baseline expansion (1.4) this paper tests against.","marker":"[4]"},{"why":"Provides the incomplete-gamma asymptotics and the Euler-Maclaurin formula used to sum the $u_j$ expansions.","marker":"[22]"}],"fun_headline_variants":["Hard wall inside droplet: log N coefficient becomes -1/4","2D Coulomb gas: interior hard wall yields universal -1/4 log N","Annulus hard wall: -1/4 log N; disk: -1/3","Radial 2D Coulomb gas: hard wall shifts log N term to -1/4"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The disk result rests on equation (5.1), an unproved equality stating that replacing an annular droplet by a disk changes the partition function by the same amount whether or not a hard wall is present; if that relation fails, the $-\\frac{1}{3}\\log N$ coefficient for the disk does not follow.","fun_headline_variants_meta":{"raw":{"variants":["Hard wall inside droplet: log N coefficient becomes -1/4","2D Coulomb gas: interior hard wall yields universal -1/4 log N","Annulus hard wall: -1/4 log N; disk: -1/3","Radial 2D Coulomb gas: hard wall shifts log N term to -1/4"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00062,"raw_usage":{"total_tokens":2916,"prompt_tokens":1028,"completion_tokens":1888,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":644,"completion_tokens_details":{"reasoning_tokens":1799}},"tokens_in":644,"tokens_out":1888,"duration_ms":12566,"temperature":1.0,"reasoning_tokens":1799,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:09:36.628686+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a potential with a disk droplet, such as $q(r)=a^2r^2$ with $0<a<1$, every $u_j$ is an incomplete gamma function, so $\\log Z_N^h$ can be computed numerically at large $N$; comparing the $-\\frac{1}{3}\\log N$ coefficient and the constant term $\\zeta'(-1)+F_D[Q]-F_{S\\cap D}[Q]$ with Theorem 2.3 would settle the disk claim. Independently, evaluating the left and right sides of equation (5.1) at finite $N$ for a family of radial potentials and checking whether their difference tends to zero would test the transfer relation directly.","supporting_citations":[{"cited_title":"Partition functions of determinantal and Pfaffian Coulomb gases with radially symmetric potentials","cited_arxiv_id":"2210.02799","evidence_quote":"Supplies the soft-wall radial partition function expansion and the functionals $F_D[Q]$, $F_{S\\cap D}[Q]$ from which the disk-annulus difference is read off."},{"cited_title":"Large gap asymptotics on annuli in the random normal matrix model","cited_arxiv_id":"2110.06908","evidence_quote":"Provides the large-gap asymptotics whose order-one constants $\\tilde\\beta_{\\rm in}$, $\\tilde\\beta_{\\rm out}$ are identified with $\\beta_{\\rm in}$, $\\beta_{\\rm out}$."},{"cited_title":"PhD thesis, Queen Mary, University of London (January 2013)","cited_arxiv_id":null,"evidence_quote":"First implementation of the Laplace-plus-Euler-Maclaurin strategy for hard-wall gap probabilities, giving the combination $\\gamma_{\\rm in}+\\gamma_{\\rm out}$ at order $\\sqrt N$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Exact solution of the two-dimensional one-component plasma on a disk at $\\beta=2$, the baseline expansion (1.4) this paper tests against."},{"cited_title":"https://dlmf.nist.gov/, Release 1.2.4 of 2025-03-","cited_arxiv_id":null,"evidence_quote":"Provides the incomplete-gamma asymptotics and the Euler-Maclaurin formula used to sum the $u_j$ expansions."}],"review_version":1}