{"id":"a65bf8cc-5f74-4e21-b27b-564a0f470be7","arxiv_id":"2507.21592","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A Brownian SDE with merely measurable, Markovian drift is claimed to have a unique strong solution and H-C regular solution map whenever its Girsanov density has a finite L^{1+ε} moment.","lead":"This paper claims that Brownian equations with very rough, merely measurable drift still have unique strong solutions, as long as the associated Girsanov weight has a finite L^{1+ε} moment. It uses a new upper-floor construction on abstract Wiener space and claims the solution map is smooth in the Gross sense.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Variational identity in §3 is algebraically false: it identifies the H2 norm of the perturbation with the H norm of its terminal derivative, so the minimizer need not solve the claimed SDE.","rationale":"I read the paper in good faith. The central claim is strong existence and H-C regularity for a Markovian drift satisfying only the Girsanov L^{1+ε} condition. For the proof to work, the variational minimizer in §3 must satisfy the original SDE, and the calculation that extracts that SDE is the pivot of Theorem 3. I checked the algebra in the displayed chain after (3.8). The completion of square uses ||ξ||²_H2 = ∫|\\dot ξ_{1,s}|² ds, which contradicts the definition of the H2 norm given in the same section. With b=0, the assumptions hold trivially, yet a deterministic element with zero terminal slice has positive K and zero claimed right side. Hence the minimization over L²_a(P,H2) is not equivalent to the SDE as written. The reader located the main weakness in the H-C^1 and multiplicity-one steps of Theorem 1; I find a distinct, more elementary failure in §3. Both are load-bearing, and either one prevents the proof from going through. I am not claiming the theorem itself is false; a different argument might prove it. But the proof as written does not establish the central claim, so the rejection verdict is appropriate. No formal verification or reproducible code is present to offset the algebraic error.","tokens_in":6815,"tokens_out":25366,"duration_ms":318978,"concrete_test":"Run the paper's §3 displayed computation with d=1 and b=0 on the deterministic adapted shift k(t,s)=(t-1)² t s. Both sides of the claimed identity K(ξ)=E∫|\\dot ξ_{1,s}|² ds can be evaluated in closed form: the left side is 1/2∫_0^1 (t-1)²(3t-1)² dt >0, while the right side is 0 because k(1,·)=∂_t k(1,·)=0. If the author's notation intends a different derivative, repeat with their own definition of \\dot ξ_{1,s}: the equality must hold for all adapted ξ, and any deterministic ξ with nonzero H2 norm and zero terminal slice is a counterexample.","verdict_should_be":"REJECT","load_bearing_attack":"The central existence argument passes through the computation of the variational cost K in §3, immediately after (3.8). The displayed chain ends with the identity K(ξ)=E∫_0^1 |\\dot ξ_{1,s}+b(s,ξ_{1,s}+B_1(s))|² ds. For this identity to hold, the preceding cancellation must have used ||ξ||²_H2 = ∫_0^1 |\\dot ξ_{1,s}|² ds. But the definition of L²_a(P,H2) in §3 gives ||ξ||²_H2 = ∫_0^1 |\\dot ξ(t)|²_H dt, where \\dot ξ(t) is an H-valued derivative; in double-time notation this is ∫∫ |∂_t ∂_s ξ(t,s)|² ds dt. The terminal quantity \\dot ξ_{1,s} is at most one boundary slice, either ∂_t ξ|_{t=1} or ∂_s ξ_1, whose L²(ds) norm is not the H2 norm. The equality fails even for b=0: take the deterministic element k(t,s)=(t-1)² t s; then k(1,·)=0 and ∂_t k(1,·)=0, so the claimed right side is zero, while K=1/2||k||²_H2>0. Since this identity is the step that extracts the SDE dX=-b dt+dB_1 from the minimizer, the derivation of the strong solution is unsupported. This is an algebraic error in the main variational computation, not merely a missing regularity condition; the reader's H-C^1/multiplicity objection is a separate, also unproved step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims a proof of strong existence and uniqueness for the SDE dX_t = b(t,X_t)dt + dW_t on [0,1] for a measurable, Markovian drift b, under only the assumptions (1.1) that the Girsanov density ε(-δb̃) has expectation 1 and (1.2) that it lies in L^{1+c_b}. The method is to pass to an 'upper floor' abstract Wiener space (Ω,H_2,P), define an adapted perturbation of identity V^τ from the logarithmic gradient of the heat semigroup, assert its almost-sure invertibility via a change-of-variables formula with multiplicity one, identify its inverse as the strong solution of a related SDE, and finally use a variational calculation to show that the minimizer solves the original SDE and is adapted to the driving Brownian motion B_1.","tokens_in":7220,"tokens_out":11230,"duration_ms":117230,"significance":"If the result were correct, it would be a striking extension of strong-existence theory for SDEs with rough drifts, reducing the hypothesis to an integrability condition on the Girsanov density and introducing a genuinely new abstract-Wiener-space technique. The paper builds on the author's substantial prior work on invertibility of adapted shifts and variational calculus, and the overall architecture is ambitious and worth serious study. However, the proof as written contains load-bearing algebraic and analytic gaps that prevent the central claims from being established; in particular, the variational identification in §3 is algebraically incorrect, and the invertibility step in §2 is asserted rather than proved under the stated hypotheses.","major_comments":[{"comment":"The variational computation contains a false algebraic identity. With N=||ξ||²_H2, M=∫|ξ̇_{1,s}|²ds, D=∫b·ξ̇_{1,s}ds and C=∫|b|²ds, the expression 1/2(N+∫|ξ̇_{1,s}+b|²ds)-1/2M equals 1/2N+D+1/2C, while the preceding line for K(ξ) is 1/2N+D+C. The displayed chain therefore already loses a factor 1/2C. Even if this were repaired, the final equality with EP∫|ξ̇_{1,s}+b|²ds would require N=M, which is false in general: for b=0 and the deterministic two-parameter function k(t,s)=(t-1)²ts, one has ξ̇_{1,s}=0 and hence the paper's final expression is 0, whereas K=1/2||k||²_H2>0. Consequently the extraction of the SDE dX_t=-b(t,X_t)dt+dB_{1,t} from the minimizer is unsupported.","section":"§3, after (3.8)"},{"comment":"The almost-sure invertibility of V^τ is asserted from the change-of-variables formula (2.6) without verifying the hypotheses of that formula. The estimates (2.4)-(2.5) are claimed to follow from H-analyticity and the strict positivity of the martingale, but no argument is given that these bounds are uniform over the H-balls needed for the H-C^1 property of v^τ on (Ω,H_2,P). Moreover, the conclusion N(·,V^τ)=1 P-a.s. from EP[G∘V^τ Λ]=EP[G] presupposes the area formula and that the multiplicity function is well-defined; this is exactly the kind of regularity that must be proved before use.","section":"§2, proof of Theorem 1"},{"comment":"The representation of the martingale EP[e^{-f}∘B_1|B_t] as the exponential martingale exp(-∫(v̇_s,dB_s)-1/2∫|v̇_s|²_H ds) requires an Itô formula for the W-valued cylindrical Brownian motion and sufficient regularity of the heat semigroup map w↦Q_{1-t}(e^{-f})(w). Under the sole assumption e^{-f}∈L^{1+ε}(μ), the H-differentiability and the integrability of the stochastic integral are not justified; the paper does not state or prove the precise conditions under which this representation holds.","section":"§2, martingale representation"},{"comment":"The vanishing of the term EP[∫ b(s, ξ_{1,s}+B_1(s)) dB_1(s)] is asserted from orthogonality of Brownian increments, but the integrand is not shown to be adapted to the filtration of B_1. The adaptedness of the minimizer is one of the main conclusions of the paper and cannot be assumed in the variational computation. The argument therefore contains a circularity: it uses the B_1-adaptation of ξ to cancel the stochastic integral before proving that adaptation.","section":"§3, stochastic integral term"},{"comment":"The approximation argument via e^{-f_n}=P_{1/n}E[e^{-f}|V_n] and the convergence v̇_n→v̇ P-a.s. are asserted without proof. The weak-convergence passage showing that U^1 is adapted to the filtration of B_1 relies on a limiting argument for which the necessary tightness and continuity in the SDE coefficients are not demonstrated; convergence of the pair (U_n,B_1) in the weak sense does not by itself preserve B_1-adaptedness of the limit. Thus the strong-solution claim of Theorem 3 is not rigorously established.","section":"Theorem 3"}],"minor_comments":[{"comment":"The phrase 'without no regularity hypothesis' should read 'without any regularity hypothesis'.","section":"Abstract"},{"comment":"The term 'Carlman-Fredholm determinant' is a typo; it should be 'Carleman-Fredholm determinant'.","section":"p. 3, proof of Theorem 1"},{"comment":"The conference proceedings 'Proc. 4 th Berkley Sym. Math. Stat. Prob.' misspells 'Berkeley'.","section":"p. 4, reference [5]"},{"comment":"The chain EP[sup_{t<1}||U_t||_W] = ... = EP[sup_{t≤1}||B_t||_W e^{-f∘B_1}] contains the typo 'ρ(-δB(v)' and the notation for the change of measure is inconsistent.","section":"p. 4, chain in Theorem 2 proof"}],"recommendation":"reject","confidential_remarks":"The manuscript relies heavily on the author's own previous results ([17], [19], [20], [21]), which is not by itself problematic, but the claimed extension to the present setting is not carried out with the necessary rigor. The variational error in §3 is a concrete algebraic failure, not a matter of presentation, and it invalidates the central mechanism by which the strong solution is identified. Given the strength of the claim, the paper would need a substantial rewrite and a correct variational argument before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline, in plain terms: this paper does not prove what it claims, but it is not a crank submission. Ustunel builds an ingenious 'upper floor' construction—an abstract Wiener space (Ω,H2,P) where the coordinate process is a Brownian sheet—and tries to derive a strong solution's regularity from the invertibility of an adapted perturbation of identity plus a variational identification. The construction is genuinely different from the Krylov–Röckner or Meyer-Brandis–Proske lines, and the idea of using Gross's H-C regularity is worth taking seriously.\n\nThe soft spot is not minor. In §3, the computation of the variational cost K contains a false algebraic identity. The paper concludes K(ξ)=E∫_0^1 |\\dot ξ_{1,s}+b(s, ξ_{1,s}+B_1(s))|² ds, which requires ||ξ||²_{H2} = ∫_0^1 |\\dot ξ_{1,s}|² ds. That is simply not true: the H2 norm is the integral over both time variables of |∂_t ∂_s ξ|², whereas \\dot ξ_{1,s} is only a boundary slice. Even for b=0, take the deterministic element k(t,s)=(t-1)² t s; then k(1,·)=0 and ∂_t k(1,·)=0, so the right side is zero while K=1/2 ||k||²_{H2} > 0. The claimed SDE for the minimizer is obtained by this step, so the derivation collapses. This is not a missing regularity condition; it is an elementary algebraic error.\n\nThere is a second gap, which the reader flagged: the invertibility of V^τ is asserted from the H-C^1 property of v^τ, but the change-of-variables formula in [21] requires more than H-C^1, and the multiplicity N(·,V^τ)=1 is not justified by (2.6) alone. These may be repairable, but they are not minor.\n\nWhat is good: the upper-floor construction is original, the approximation via Ornstein–Uhlenbeck regularization (3.9) is a plausible route, and the adaptation argument in Theorem 3 is creative. The paper cites related work, including the author's own, extensively; that is not a flaw so long as the cited results are genuinely used.\n\nBottom line: as written, this is a reject. But the underlying idea is not obviously dead. I would send it to a strong referee rather than desk-reject, so the author can learn exactly where the variational identity fails and whether the invertibility step can be repaired. If the identity cannot be fixed, then the theorem, as formulated, is unsupported.","headline":"Novel upper-floor construction, but the main variational identity is false; the paper as written does not prove the claimed strong existence.","tokens_in":7681,"tokens_out":4546,"would_cite":false,"duration_ms":50945,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H07","60H10","60H30","37A35","57C70","94A17"],"pacs":[],"model":"deepseek-v4-flash","headline":"A merely measurable Markovian drift forces a strong solution of $dX_t=b(t,X_t)dt+dW_t$ whenever the Girsanov exponential has a finite $L^{1+\\epsilon}$ moment, and the solution map is H-C regular.","keywords":["entropy","Girsanov theorem","strong solutions","stochastic differential equations","almost sure invertibility","cylindrical Brownian motion","abstract Wiener space","measurable drift"],"falsifier":"Check whether estimates (2.4)-(2.5) hold for a drift $b$ satisfying (1.1)-(1.2): if for some such $b$ the gradient $\\dot v_t(B_t)=-\\nabla\\log Q_{1-t}e^{-f}(B_t)$ is unbounded on Cameron-Martin balls of finite radius on a set of positive $P$-measure, then $V^{\\tau}$ is not H-C$^1$ and the claimed invertibility is unsupported. A direct computation of the multiplicity $N(\\omega,V^{\\tau})$ for an admissible rough drift that exceeds one on a positive-measure set would likewise contradict (2.6) and refute Theorem 1.","tokens_in":6631,"feed_emoji":"🎲","tokens_out":12932,"duration_ms":137349,"temperature":0.7,"pith_summary":"This paper proves that the stochastic differential equation $dX_t=b(t,X_t)dt+dW_t$ has a strong, pathwise unique solution when the drift $b$ is only assumed measurable and Markovian, provided the Girsanov density $\\rho(-\\tilde\\delta b)=e^{-f}$ satisfies $\\mathbb{E}[\\rho(-\\tilde\\delta b)]=1$ and $\\mathbb{E}[\\rho(-\\tilde\\delta b)^{1+c_b}]<\\infty$ for some fixed $c_b>0$. No continuity, boundedness, or spatial integrability of $b$ is needed; the entire hypothesis is carried by the exponential martingale that $b$ generates. The proof moves the problem to an abstract Wiener space over $\\Omega=C([0,1],W)$ with Cameron-Martin space $H_2=H\\otimes_2 H$, where an adapted perturbation of identity $V^{\\tau}$ is shown to be almost surely invertible, and its inverse solves the SDE. A variational identification then presents the solution as the unique minimizer of an entropic cost functional, and an approximation argument proves the solution is adapted to the driving Brownian motion, hence strong.","feed_headline":"One moment condition forces strong solutions for rough SDEs","feed_subtitle":"Merely measurable Markovian drifts solve the rough SDE if their Girsanov exponential has finite moment; solution map is H-C regular.","key_machinery":"The central object is the adapted perturbation of identity $V^{\\tau}=I_{H_2}+v^{\\tau}$ on the upper abstract Wiener space. Its derivative in Cameron-Martin directions is controlled by the bounds $\\sup_{t\\le\\tau}\\sup_{\\|K\\|_2\\le M}|\\dot v_t(B_t(\\omega)+K_t)|_H\\le c_{1,M}(\\tau,\\omega)$ and the analogous bound for $\\nabla\\dot v_t$, which together give the H-C$^1$ (Hilbert-Cameron $C^1$) property. Because the perturbation is adapted, the modified Carleman-Fredholm determinant of $I_{H_2}+\\nabla v^{\\tau}$ equals one, so the abstract Wiener-space change-of-variables formula reduces to $\\mathbb{E}_P[G\\circ V^{\\tau}]=\\mathbb{E}_P[G N(\\cdot,V^{\\tau})]$, and the equality of expectations for all $G\\in C_b(\\Omega)$ forces $N(\\cdot,V^{\\tau})=1$ almost surely. The H-analyticity of $w\\mapsto Q_{1-t}e^{-f}(w)$, obtained through the shifted heat-kernel identity, supplies the regularity that makes the whole construction run, and the variational functional $K(\\xi)=\\mathbb{E}_P\\int_0^1|\\dot\\xi_{1,s}+b(s,\\xi_{1,s}+B_{1,s})|^2ds$ links the inverse of $V^{\\tau}$ back to the original SDE.","core_discovery":"The central claim is that under assumptions (1.1)-(1.2) the SDE $dX_t=b(t,X_t)dt+dW_t$ has a strong solution whose solution map is H-C regular, i.e., Hilbert-Cameron regular in the abstract Wiener space sense, on $(\nOmega,H_2,P)$ with $\\Omega=C([0,1],W)$ and $H_2=H\\otimes_2 H$. On this upper floor the adapted perturbation of identity $V^{\\tau}_t(B)=B_t(B)+\\int_0^{\\tau\\wedge t}\\dot v_s(B_s(B))ds$, with $\\dot v_s=-\\nabla\\log Q_{1-s}e^{-f}$, is proved $P$-a.s. invertible: the H-C$^1$ estimates (2.4)-(2.5) give $v^{\\tau}$ the needed regularity, the adaptedness makes the Carleman-Fredholm determinant equal to one, and the change-of-variables formula (2.6) forces the multiplicity $N(\\cdot,V^{\\tau})$ to equal one almost surely. The inverse $U$ satisfies $dU_t=-\\dot v(t,U_t)dt+dB_t$, and the variational argument identifies $X_t=\\xi_{1,t}+B_{1,t}$, where $\\xi$ is the unique minimizer of the entropic functional, as the solution of the original SDE. Theorem 3 completes the proof by showing $X$ is adapted to the filtration of $B_1$, so the weak solution constructed at the upper floor is in fact strong; the same construction yields real H-analyticity of the solution for $t<1$.","pith_inferences":["Extension the paper leaves implicit: the same upper-floor inversion scheme might prove strong existence for SDEs driven by other Gaussian processes, wherever the heat semigroup $Q_t$ and the H-analyticity of $Q_{1-t}e^{-f}$ survive.","A natural test is the critical case $c_b=0$: the proof needs the extra $L^{1+c_b}$ room to obtain the bounds (2.4)-(2.5), so it remains open whether plain $L^1$ integrability of the Girsanov density already forces a strong solution.","Because the argument is variational, one could attempt to compute the minimizer explicitly for structured drifts, turning the existence proof into a numerical or analytical recipe for the solution map."],"forward_implications":["Any measurable Markovian drift whose Girsanov density has mean one and a fixed $L^{1+c_b}$ moment produces a pathwise unique strong solution of the SDE, without any continuity or growth condition on the drift itself.","The solution map is H-C regular on the abstract Wiener space, so the solution varies smoothly with Cameron-Martin shifts rather than only in $L^p$.","The variational characterization identifies the strong solution as the unique minimizer of an explicit entropic cost, giving a variational calculus for singular drifts.","Strong existence is obtained through almost-sure invertibility of an adapted perturbation of identity, so uniqueness and regularity come from the geometry of the Gaussian space rather than from pathwise estimates on $b$."],"supporting_citations":[{"why":"establishes the abstract Wiener space setting and the H-C regularity notion on which the upper-floor construction rests","marker":"[5]"},{"why":"provides the Radonification of the cylindrical Gaussian measure on H2 and the cylindrical Brownian motion/Brownian sheet representation used at the upper floor","marker":"[9]"},{"why":"supplies the change-of-variables formula on abstract Wiener space used in (2.6) to force multiplicity one and almost-sure invertibility","marker":"[21]"},{"why":"gives the variational problem (3.8) whose unique minimizer is equivalent to the strong solution","marker":"[19]"},{"why":"extends the entropic variational representation of Laplace transforms to the setting used on the upper floor","marker":"[20]"},{"why":"yields the H-C regularity of the approximate densities e^{-f_n}=P_{1/n}E[e^{-f}|V_n], which drives the smooth approximation in Theorem 3","marker":"[6]"},{"why":"supplies the generalized Paley-Wiener integral used to prove the H-analyticity of w -> Q_{1-t}e^{-f}(w) in (2.3)","marker":"[10]"},{"why":"provides the invertibility criterion for adapted perturbations of identity that underlies the conclusion that V^tau is almost surely invertible","marker":"[17]"}],"fun_headline_variants":["Girsanov moment bound suffices for strong SDE solutions","Measurable drift gives strong solutions if Girsanov moment exists","H-C regular solution maps for rough SDEs under one moment condition","Strong SDE solutions from Girsanov integrability alone","One moment condition forces strong SDE solution maps"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the shift map $V^{\\tau}$ built from the drift is smooth enough in the Cameron-Martin sense that the abstract change-of-variables formula applies and forces its multiplicity to be one almost surely, a regularity step that may fail for Girsanov densities only known to lie in $L^{1+\\varepsilon}$.","fun_headline_variants_meta":{"raw":{"variants":["Girsanov moment bound suffices for strong SDE solutions","Measurable drift gives strong solutions if Girsanov moment exists","H-C regular solution maps for rough SDEs under one moment condition","Strong SDE solutions from Girsanov integrability alone","One moment condition forces strong SDE solution maps"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001071,"raw_usage":{"total_tokens":4492,"prompt_tokens":958,"completion_tokens":3534,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":574,"completion_tokens_details":{"reasoning_tokens":3450}},"tokens_in":574,"tokens_out":3534,"duration_ms":26047,"temperature":1.0,"reasoning_tokens":3450,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T12:38:28.816576+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Check whether estimates (2.4)-(2.5) hold for a drift $b$ satisfying (1.1)-(1.2): if for some such $b$ the gradient $\\dot v_t(B_t)=-\\nabla\\log Q_{1-t}e^{-f}(B_t)$ is unbounded on Cameron-Martin balls of finite radius on a set of positive $P$-measure, then $V^{\\tau}$ is not H-C$^1$ and the claimed invertibility is unsupported. A direct computation of the multiplicity $N(\\omega,V^{\\tau})$ for an admissible rough drift that exceeds one on a positive-measure set would likewise contradict (2.6) and refute Theorem 1.","supporting_citations":[{"cited_title":"Abstract Wiener spaces","cited_arxiv_id":null,"evidence_quote":"establishes the abstract Wiener space setting and the H-C regularity notion on which the upper-floor construction rests"},{"cited_title":"Kuo: Gaussian Measures on Banach Spaces","cited_arxiv_id":null,"evidence_quote":"provides the Radonification of the cylindrical Gaussian measure on H2 and the cylindrical Brownian motion/Brownian sheet representation used at the upper floor"},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the change-of-variables formula on abstract Wiener space used in (2.6) to force multiplicity one and almost-sure invertibility"},{"cited_title":"Entropy, invertibility and variational calculus of adapted shifts on Wiener space","cited_arxiv_id":null,"evidence_quote":"gives the variational problem (3.8) whose unique minimizer is equivalent to the strong solution"},{"cited_title":"Variational calculation of Laplace transforms via entropy on Wiener space and applications","cited_arxiv_id":null,"evidence_quote":"extends the entropic variational representation of Laplace transforms to the setting used on the upper floor"},{"cited_title":"Variational calculus for diffusions","cited_arxiv_id":"1607.05488","evidence_quote":"yields the H-C regularity of the approximate densities e^{-f_n}=P_{1/n}E[e^{-f}|V_n], which drives the smooth approximation in Theorem 3"},{"cited_title":"Precise asymptotics of certain Wiener functionals","cited_arxiv_id":null,"evidence_quote":"supplies the generalized Paley-Wiener integral used to prove the H-analyticity of w -> Q_{1-t}e^{-f}(w) in (2.3)"},{"cited_title":"A necessary and sufficient condition for invertibility of adapted perturbations of identity on Wiener space","cited_arxiv_id":null,"evidence_quote":"provides the invertibility criterion for adapted perturbations of identity that underlies the conclusion that V^tau is almost surely invertible"}],"review_version":1}