{"id":"7d3429de-bf07-4a9d-907c-4a51b4ca57bc","arxiv_id":"1908.05950","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Rigorous singular asymptotics and connection formulas are proved for inhomogeneous Painlevé II solutions with infinitely many poles on the negative axis, for all real alpha.","lead":"This paper proves the precise oscillatory blow-up behavior of a family of solutions to the inhomogeneous Painlevé II equation that have infinitely many poles as x goes to negative infinity. It also completes the rigorous asymptotic description for all real parameter values at positive infinity, closing a gap left by earlier work.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Equation (4.23) is algebraically false as written: substituting (4.12) and (4.22) gives a factor t^{ν0-1}, so the local parametrix matching and residue conditions (4.37)-(4.38) that drive the connection formulas are not justified.","rationale":"After reading the paper in good faith, I find the reader's concern is correct and load-bearing. The paper's goal is to rigorously prove the singular asymptotics and connection formulas for PII with α≠0 and |k|>|cos πα|. The Deift-Zhou analysis is standard, but the new difficulty is the local parametrix near z=1/2 for the singular case ν = ν0 - 1/2. Equation (4.23) is the key step that asserts the t-dependence drops out of β^2/(√(tζ)); direct substitution refutes this. Moreover, the residue conditions (4.37)-(4.38) appear to have the reciprocal of ~β(z+) compared to what a direct computation from E_r gives. Since these residues determine the connection formulas, the proof has a genuine gap. However, the claimed results are consistent with known formulas from Kapaev and Bothner & Its (α=0), and I found no evidence of fitted parameters or circular reasoning. Therefore the appropriate verdict is unchanged: CONDITIONAL, pending correction of the algebra in Section 4.2.","tokens_in":21229,"tokens_out":20719,"duration_ms":154842,"concrete_test":"Independently re-derive Eq. (4.23) by substituting (4.12) and (4.22) with ν=ν0-1/2, computing β^2/(√(tζ)) and comparing to ~β/~α; if they differ by the factor t^{ν0-1} ζ^{-ν0-1}, then recompute the residue conditions (4.37)-(4.38) from the corrected Q^(r)=E_r P^(r) and test whether the resulting p in (4.49) still matches (4.53). If p changes, the claimed connection formulas (1.11)-(1.12) require revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Substituting the definitions (4.12) and (4.22) into the left side of (4.23) yields β^2/(√(tζ)) = t^{ν0-1} ζ^{ν0-1} α̃^{2ν0-1}, whereas the claimed right side ~β/~α equals ζ^{2ν0} α̃^{2ν0-1}. The two differ by the factor t^{ν0-1} ζ^{-ν0-1}, which is of order t^{-1} on ∂U(1/2,δ) and is not identically 1 for ν0∈iR\\{0}. Thus the displayed equality cannot hold for large t; the left side is O(t^{-1}) while the right side is O(1). This matters because (4.23) is used to simplify P^(r) to the lower-triangular form in (4.24), from which E_r is defined in (4.25)-(4.26) and the residue conditions (4.37)-(4.38) are extracted. A direct residue computation of E_r = (1 0; h1/s3 e^{-2it/3} ~α/~β 1) gives a residue proportional to 1/~β(1/2), not ~β(1/2) as stated in (4.37). Since (4.37)-(4.38) determine the matrix D in (4.48) and hence the pole spacing and phase φ in (5.6)-(5.9), the singular asymptotics and connection formulas (1.10)-(1.12) rest on an unproven identity. The final formulas match Kapaev's isomonodromy results, so the gap is likely fixable, but the proof as written is incomplete.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies real solutions of the inhomogeneous Painlevé II equation u''=2u^3+xu-α with α∈R\\{0}, k∈R, and |k|>|cos πα|. It claims to prove, via the Deift-Zhou nonlinear steepest descent method for the associated Riemann-Hilbert problem, the asymptotics as x→+∞ (algebraic expansion plus an exponentially small Airy term), the singular asymptotics as x→-∞ (an oscillatory √(-x)/sin(...) form away from poles), and connection formulas relating the phase and logarithmic shift to k and α. The x→+∞ part extends earlier work of Its and Kapaev to all real α, while the x→-∞ part is the main new contribution and extends Bothner-Its from α=0. The proof introduces an explicit model Riemann-Hilbert problem solved in terms of Hankel functions, constructs global and local parametrices near the stationary points ±1/2 and near the origin, and uses a dressing procedure to handle the poles of the error function.","tokens_in":21570,"tokens_out":20067,"duration_ms":174504,"significance":"If the proof were complete, this would be a valuable contribution: it gives a purely Riemann-Hilbert derivation of the singular asymptotics and connection formulas for a family of PII solutions, matching Kapaev's isomonodromy predictions, and it closes the half-integer gap in the x→+∞ asymptotics. The model problem in Section 2 is explicit, the connection formulas are not fitted to numerical data, and the x→+∞ extension covers all real α. However, the x→-∞ analysis has a load-bearing algebraic gap in the local parametrix at z=1/2, so the main theorem as written is not established.","major_comments":[{"comment":"Equation (4.23) is algebraically false as stated. From (4.12) and the definitions of tilde β and tilde α in (4.22), β^2(z)/√(t ζ(z)) = t^{ν-1/2} ζ^{ν-1/2} α̃(z)^{2ν-1}. Writing ν = ν0 - 1/2 with ν0∈iR gives t^{ν0-1} ζ^{ν0-1} α̃^{2ν0-1}. On ∂U(1/2,δ), ζ is bounded away from zero, so this is O(t^{-1}), whereas the asserted right-hand side tilde β/tilde α = ζ^{2ν0} α̃^{2ν0-1} is O(1). The two sides differ by the factor t^{ν0-1} ζ^{-ν0-1}, which is not identically 1. Thus the matching condition used to pass to the lower-triangular form (4.24) is not justified.","section":"Section 4.2, Eq. (4.23)"},{"comment":"The parametrix Q^(r) is built from the lower-triangular reduction in (4.24): E_r in (4.26) is chosen so that E_r P^(r) ≈ (I+O(1/t))N(z). Since (4.23) fails, the actual lower-left entry of P^(r) contains an additional t^{-ν0} ζ^{ν0} factor relative to tilde α/tilde β; consequently the residue conditions (4.37)-(4.38), the matrix D in (4.48), and the pole equation (4.54) are not established. The author should recompute the expansion of P^(r) from the Z expansion in (4.18) with the correct powers of t and ζ and adjust E_r, p, and the residue formulas accordingly.","section":"Section 4.2, Eqs. (4.24)-(4.26) and (4.37)-(4.38)"},{"comment":"The displayed transition from (4.19) to (4.24) is also not internally consistent: the (1,2) entry of the matrix in (4.24) contains a factor 1/(t ζ^2) that has no counterpart in (4.19), even after substitution of (4.23). Please derive the asymptotic expansion of P^(r) from the Z expansion in (4.18) in sufficient detail that the reader can verify the powers of t and ζ in all entries.","section":"Section 4.2, Eqs. (4.19) and (4.24)"}],"minor_comments":[{"comment":"The error terms O((-x)^{-3/2}) + O((-x)^{-1}) are redundant because the second term dominates; please state the intended uniformity statement more cleanly, for example by separating the error in the denominator approximation from the uniform remainder.","section":"Equation (1.10)"},{"comment":"The introduction states that Hastings-McLeod solutions exist for k = sgn(α) cos(πα) for all α, but for α = n±1/2 this gives k=0 and the table marks k=cos(πα) and k=-cos(πα) as D.N.E. Please clarify which convention is intended for half-integer α.","section":"Introduction and Table 1"},{"comment":"Please specify the branch of the square root in the definition of ζ(z) more explicitly; equation (4.9) fixes the local behavior, but the global branch choice in U(z_+,δ) is not stated.","section":"Equation (4.8)"},{"comment":"The affiliation line contains the misspelling 'Stastatics'; it should be 'Statistics'.","section":"Author affiliation"}],"recommendation":"major_revision","confidential_remarks":"The paper's x→-∞ section is the core novelty, and the gap is located in the local parametrix computation. Since the final formulas match Kapaev's known isomonodromy results, I expect the gap can be repaired locally, but the proof as submitted is incomplete. I would not reject; the authors should be asked to redo the parametrix expansion in Section 4.2 and to check all subsequent residue and dressing formulas."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a serious paper in the standard Deift–Zhou program for the inhomogeneous Painlevé II equation. It proves the x→+∞ asymptotics for all real α, closing the half-integer gap left by Its–Kapaev, and it gives the first rigorous singular asymptotics for α ≠ 0, with connection formulas matching Kapaev's 1992 isomonodromy predictions and the α=0 case of Bothner–Its. The new model problem M built from Hankel functions is the right tool: H_{α±1/2}^{(1,2)} stay linearly independent at the half-integer α values where Bessel J functions fail. The paper is self-contained, uses no fitted parameters, and cites the earlier work accurately.\n\nBut there is a real problem in Section 4.2 that the referee needs to dig into. Equation (4.23) claims β²(z)/√(tζ(z)) = β̃(z)/ᾱ(z) = O(1). Put in the definitions from (4.12) and (4.22), and with ν = ν0 − 1/2, ν0 ∈ iℝ, the left side equals t^{ν0−1} ζ^{ν0−1} ᾱ^{2ν0−1}. That is O(t^{−1}) as t→∞, not O(1). The right side is ζ^{2ν0} ᾱ^{2ν0−1}. The two differ by t^{ν0−1} ζ^{−ν0−1}, which is not 1. So (4.23) is algebraically false. The intended conclusion—that the (1,2) entry of P^(r) is O(1/t) and can be dropped—may still survive because the true left side is even smaller than O(1), but the displayed equality is wrong and the error propagates.\n\nThe bigger issue is the residue conditions. From the explicit definition of E_r in (4.26), the pole at z=1/2 comes from α̃(z)/β̃(z), whose residue is 1/β̃(1/2), not β̃(1/2). Yet (4.37)–(4.38) state residues proportional to β̃(1/2). That is a factor-of-β̃(1/2)² discrepancy (modulus one, so the |p|=1 normalization in (4.53) may survive, but the phase and the connection formula φ will not be justified). Since the final formulas match Kapaev's known results, I suspect the error is fixable, but the proof as written is incomplete.\n\nWho is this for? Someone working on Painlevé asymptotics or Riemann–Hilbert steepest descent will want to see the Hankel model problem, even before a corrigendum. The x→+∞ half is fine. The x→−∞ part should not be accepted until (4.23) and the residue derivation are corrected. Send it to a referee who knows parabolic cylinder parametrices; it deserves serious peer review.","headline":"The x→+∞ extension is solid, but the local parametrix near z=1/2 contains an algebraic error that leaves the singular asymptotics and connection formulas unproven as written.","tokens_in":22133,"tokens_out":12310,"would_cite":false,"duration_ms":98051,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["41A60","33C45"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every real α and every real k with |k|>|cos(πα)|, the inhomogeneous Painlevé II equation has a real solution whose +∞ decay is an Airy correction and whose −∞ behaviour is a singular oscillation with explicit connection formulas.","keywords":["Painlevé II equation","singular asymptotics","Riemann-Hilbert problem","connection formulas","inhomogeneous Painlevé II","nonlinear steepest descent","Stokes multipliers","Airy function"],"falsifier":"Evaluate $\\beta(z)^2/(\\sqrt{t\\,\\zeta(z)})$ on the circle $|z-1/2|=\\delta$ with $\\beta(z)=(\\sqrt{t\\,\\zeta(z)}\\,(z+1/2)/(z-1/2))^{\\nu}$ and $\\nu=\\nu_0-\\tfrac12$, $\\nu_0\\in i\\mathbb{R}$, letting $t=(-x)^{3/2}\\to\\infty$. If the absolute value is proportional to $t^{-1}$ rather than $O(1)$, the matching identity (4.23) fails and the residue equations (4.37)--(4.38), and hence the connection formulas (1.11)--(1.12), would need to be re-derived.","tokens_in":20971,"feed_emoji":"🧮","tokens_out":16380,"duration_ms":130151,"temperature":0.7,"pith_summary":"This paper proves that the inhomogeneous Painlevé II equation $u''=2u^3+xu-\\alpha$ has, for every real $\\alpha$ and every real $k$ with $|k|>|\\cos(\\pi\\alpha)|$, a real solution that decays at $+\\infty$ as an algebraic series plus $k$ times the Airy function, and that the same solution oscillates with growing amplitude and infinitely many poles as $x\\to-\\infty$. It supplies rigorous connection formulas for this regime, expressing the amplitude $d$ and phase $\\phi$ of the $-\\infty$ oscillations directly in $k$ and $\\alpha$: $d=\\pi^{-1/2}\\sqrt{\\ln(k^2-\\cos^2(\\pi\\alpha))}$ and $\\phi=(3\\ln2/2)d^2-\\arg\\Gamma(1/2+id^2/2)-\\arg(-\\sin(\\pi\\alpha)-ki)$. The derivation is a nonlinear steepest-descent analysis of the associated Riemann-Hilbert problem, and it extends the known $+\\infty$ asymptotics from the previously accessible cases $\\alpha-\\tfrac12\\notin\\mathbb{Z}$ to all real $\\alpha$. The result completes the classification of real Painlevé II solutions with decaying boundary condition in this parameter range and makes the pole locations at $-\\infty$ explicit.","feed_headline":"Exact connection formulas found for singular Painlevé II solutions","feed_subtitle":"For every real α and |k| > |cos(πα)|, the +∞ Airy tail fixes the −∞ pole oscillations.","key_machinery":"The load-bearing object is a model Riemann-Hilbert problem for a $2\\times2$ matrix $M(\\eta)$ with jumps on four rays and an algebraic singularity at $\\eta=0$, solved explicitly by (2.6)--(2.8) in terms of the two Hankel functions $H^{(1)}_{\\alpha\\pm1/2}$ and $H^{(2)}_{\\alpha\\pm1/2}$. Because those two functions are linearly independent for every real $\\alpha$, the same model works as the local parametrix at the origin when $\\alpha-\\tfrac12$ is an integer, a case where Bessel functions of the first kind would become linearly dependent. At the stationary point $z=\\tfrac12$, the parametrix $Q^{(r)}(z)=E_r(z)P^{(r)}(z)$ is built from parabolic cylinder functions plus an explicitly singular prefactor $E_r$ with a simple pole; the resulting residue conditions (4.37)--(4.38) are fed into a dressing step $R(z)=(zI+D)\\,W(z)\\,\\mathrm{diag}(1/(z-z_+),1/(z-z_-))$ that turns the pole problem into a solvable singular integral equation for $W$. The $(1,2)$ entry of the constant matrix $D$ gives the leading term of $u(x;\\alpha)$, and the condition $1+p^2=0$ locates the poles.","core_discovery":"The central claim is Theorem 1: given $\\alpha\\in\\mathbb{R}$ and $k\\in\\mathbb{R}$ with $|k|>|\\cos(\\pi\\alpha)|$, there exists a real solution $u(x;\\alpha)$ of $u''=2u^3+xu-\\alpha$ such that $u(x;\\alpha)=B(x;\\alpha)+k\\,\\mathrm{Ai}(x)(1+O(x^{-3/4}))$ as $x\\to+\\infty$, where $B$ has the full expansion $B(x;\\alpha)\\sim(\\alpha/x)\\sum_{n\\ge0}a_n x^{-3n}$ with $a_0=1$ and the recursion $a_{n+1}=(3n+1)(3n+2)a_n-2\\alpha^2\\sum_{k+l+m=n}a_k a_l a_m$; and as $x\\to-\\infty$, away from the zeros of the denominator, $$u(x;\\$\\alpha$)=\\frac{\\sqrt{-x}}{\\sin\\!\\big(\\tfrac23(-x)^{3/2}+\\tfrac34 $d^{2}$\\ln(-x)+\\phi\\big)}+O((-x)^{-1}),$$ with $d=\\pi^{-1/2}\\sqrt{\\ln(k^2-\\cos^2(\\pi\\alpha))}$ and $\\phi=(3\\ln2/2)d^2-\\arg\\Gamma(\\tfrac12+\\tfrac i2 d^2)-\\arg(-\\sin(\\pi\\alpha)-ki)$. The formulas are derived by constructing explicit local parametrices: a parabolic-cylinder model at the two stationary points and a model problem at the origin solved in closed form using Hankel functions, which remain linearly independent for every real $\\alpha$; this is what removes the earlier restriction $\\alpha-\\tfrac12\\notin\\mathbb{Z}$. The pole locations arise from the condition $1+p^2=0$ on an explicit phase parameter $p$, producing the quantization equation (4.54).","pith_inferences":["Editorial inference: the explicit Hankel-function model should also work at the critical value $|k|=|\\cos(\\pi\\alpha)|$ by taking the limit $d\\to0$, which would connect the finitely-poled monotonic branch at criticality to the infinitely-poled family treated here, with the phase formula degenerating in a predictable way.","Editorial inference: the boundedness assumption behind the turning-point matching can be settled by a direct asymptotic calculation; if the ratio $\\beta^2/(\\sqrt{t\\,\\zeta})$ is in fact of order $t^{-1}$, the residue conditions would acquire extra $t$-dependence, shifting the pole quantization without necessarily destroying the leading sine formula.","Editorial inference: the same parametrix strategy, based on Hankel functions with integer-shifted order, is likely to remove analogous exceptional-parameter restrictions in other Riemann-Hilbert problems where Bessel-function parametrices degenerate at discrete parameter values."],"forward_implications":["The $+\\infty$ asymptotic $u(x;\\alpha)=B(x;\\alpha)+k\\,\\mathrm{Ai}(x)(1+O(x^{-3/4}))$ now holds for every real $\\alpha$, including the previously excluded half-integer shifts $\\alpha-\\tfrac12\\in\\mathbb{Z}$.","For $|k|>|\\cos(\\pi\\alpha)|$, every such solution has infinitely many poles on $(-\\infty,0)$, with positions governed by the explicit quantization condition (4.54) involving the same parameters $k$ and $\\alpha$.","The connection formulas give a complete dictionary: the decay parameter $k$ at $+\\infty$ determines both the logarithmic phase-shift amplitude $d$ and the phase $\\phi$ at $-\\infty$, so no free data is lost across the real line.","The singular asymptotic (1.10) is uniform on intervals bounded away from poles, so the leading oscillatory envelope $\\sqrt{-x}$ and the logarithmic phase $\\tfrac34 d^2\\ln(-x)$ are robust predictions that can be compared numerically.","Combined with the earlier pole-free and finitely-poled cases, the classification table is now complete for all real $\\alpha$ and all real $k$: pole-free solutions for $|k|\\le|\\cos(\\pi\\alpha)|$ and singular solutions with infinitely many poles otherwise."],"supporting_citations":[{"why":"establishes the meromorphic solvability of the Riemann-Hilbert problem and the correspondence between the matrix expansion and the Painlevé solution","marker":"[4]"},{"why":"supplies the singular-asymptotics machinery and the homogeneous (α=0) connection formulas that the present paper generalizes","marker":"[5]"},{"why":"defines the Riemann-Hilbert problem for the inhomogeneous equation and the local behavior at the origin used in the model problem","marker":"[11]"},{"why":"provides the steepest-descent framework and the parabolic-cylinder parametrix near the stationary points that the paper adapts by adding a pole","marker":"[15]"},{"why":"gives the standard Riemann-Hilbert formulation and the parabolic-cylinder parametrix construction behind the local analysis","marker":"[21]"},{"why":"proved the +∞ asymptotics for α−1/2 not in Z, the case extended here to all real α","marker":"[25]"},{"why":"first derived the connection formulas that the paper reproves by steepest descent","marker":"[27]"}],"fun_headline_variants":["Painlevé II singular asymptotics: connection formulas for all real α","All real α: Painlevé II pole asymptotics solved","Closed-form parametrices yield Painlevé II pole laws","Removing the α barrier in Painlevé II connection formulas","Painlevé II: exact pole asymptotics for every real parameter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The derivation leans on an unverified balance: the matching of the local model near the turning point z=1/2 is assumed to stay finite as x→−∞, and if that balance fails, the formulas connecting the +∞ and −∞ behaviors would need to change.","fun_headline_variants_meta":{"raw":{"variants":["Painlevé II singular asymptotics: connection formulas for all real α","All real α: Painlevé II pole asymptotics solved","Closed-form parametrices yield Painlevé II pole laws","Removing the α barrier in Painlevé II connection formulas","Painlevé II: exact pole asymptotics for every real parameter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000262,"raw_usage":{"total_tokens":1656,"prompt_tokens":1064,"completion_tokens":592,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":680,"completion_tokens_details":{"reasoning_tokens":504}},"tokens_in":680,"tokens_out":592,"duration_ms":5843,"temperature":1.0,"reasoning_tokens":504,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:01:21.691477+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate $\\beta(z)^2/(\\sqrt{t\\,\\zeta(z)})$ on the circle $|z-1/2|=\\delta$ with $\\beta(z)=(\\sqrt{t\\,\\zeta(z)}\\,(z+1/2)/(z-1/2))^{\\nu}$ and $\\nu=\\nu_0-\\tfrac12$, $\\nu_0\\in i\\mathbb{R}$, letting $t=(-x)^{3/2}\\to\\infty$. If the absolute value is proportional to $t^{-1}$ rather than $O(1)$, the matching identity (4.23) fails and the residue equations (4.37)--(4.38), and hence the connection formulas (1.11)--(1.12), would need to be re-derived.","supporting_citations":[{"cited_title":"Bolibruch, A","cited_arxiv_id":null,"evidence_quote":"establishes the meromorphic solvability of the Riemann-Hilbert problem and the correspondence between the matrix expansion and the Painlevé solution"},{"cited_title":"Bothner and A","cited_arxiv_id":null,"evidence_quote":"supplies the singular-asymptotics machinery and the homogeneous (α=0) connection formulas that the present paper generalizes"},{"cited_title":"Claeys, A","cited_arxiv_id":null,"evidence_quote":"defines the Riemann-Hilbert problem for the inhomogeneous equation and the local behavior at the origin used in the model problem"},{"cited_title":"Dai and W","cited_arxiv_id":null,"evidence_quote":"provides the steepest-descent framework and the parabolic-cylinder parametrix near the stationary points that the paper adapts by adding a pole"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"gives the standard Riemann-Hilbert formulation and the parabolic-cylinder parametrix construction behind the local analysis"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"proved the +∞ asymptotics for α−1/2 not in Z, the case extended here to all real α"},{"cited_title":"Kapaev, Global asymptotics of the second Painlev´ e transcendent,Phys","cited_arxiv_id":null,"evidence_quote":"first derived the connection formulas that the paper reproves by steepest descent"}],"review_version":1}