{"id":"ab314298-7411-440a-ad10-aad30ea5481b","arxiv_id":"2507.13330","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"For thin blood vessels in tissue, solutions of the full 3D Darcy-Stokes system converge to reduced 3D-1D and 1D models at rate epsilon^{1/6}|log epsilon|.","lead":"This paper proves that a detailed 3D model of blood flow in and around a thin vessel converges to simpler 1D and 3D-1D models as the vessel radius shrinks, with an explicit error rate. It is the first rigorous link among three levels of perfusion models commonly used to study blood delivery in tissue.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2.2 constructs v1 via the multi-valued angle θ, so v1 is not 2π-periodic and cannot lie in H¹(Vε); the inf-sup condition and Theorem 1.2 are therefore not proven by the given argument.","rationale":"The reader's weakest assumption focuses on the companion-paper estimates from [42]; that is a legitimate concern about the reliance on unproved-in-this-paper results. However, I found a more fundamental internal issue: the construction in Lemma 2.2, essential for the well-posedness of the 3D-3D system, uses the multi-valued angle θ in the vector field v1. This field is not periodic in θ and therefore cannot be in H¹(Vε). The divergence identity behind the lemma is only valid away from the branch cut, so the proof of the inf-sup condition and of Theorem 1.2 is incomplete. Because the main convergence result assumes the 3D-3D solution exists, this gap is load-bearing: even if every imported Part I estimate is correct, the paper's own proof does not establish the existence of the object whose error is claimed. This does not prove the theorem false; a corrected periodic construction may exist, and the rest of the error analysis appears carefully done. I therefore keep the reader's CONDITIONAL verdict, but the condition should now include a corrected Lemma 2.2 in addition to the Part I verifiability.","tokens_in":46505,"tokens_out":19809,"duration_ms":218423,"concrete_test":"For fixed s and r∈(0,εa(s)), compute the jump of v1 defined in (2.17) across the branch cut θ=0/2π. Since eθ is continuous and the non-θ terms are periodic, v1(2π)−v1(0)=2π c r(1−rκ)^{-1}eθ≠0 for c≠0, so v1 is not H¹. A decisive check is to produce a periodic vector field w with the same divergence property and the same ϵ-independent bound (2.19), or to show that no such construction exists; either outcome settles whether Lemma 2.2 can be repaired.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central convergence result Theorem 1.4 presupposes existence of the 3D-3D solution (uε,pε,qε), established in Theorem 1.2. The proof of Theorem 1.2 relies on the inf-sup condition (Corollary 2.3), which in turn rests on Lemma 2.2. In Lemma 2.2, the constructed vector field v1 has eθ-component proportional to c r θ (equation (2.17)). On the full tubular domain Vε, the angle θ is not single-valued: θ=0 and θ=2π represent the same point, so v1 has a jump discontinuity across the branch cut. Hence v1 is not in H¹(Vε), the identity ∇·v1=c fails in the weak sense (there is an additional surface distribution on the cut), and the surjectivity of the divergence operator, the bound (2.19), and the inf-sup condition are not justified. Since Theorem 1.2 supplies the solution whose approximation is the paper's main object, the argument does not currently close. This is an internal gap, independent of the imported Part I estimates, and it must be repaired before the main theorem can be accepted as proved.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a hierarchy of three blood-perfusion models and proves convergence between the 3D-3D Darcy--Stokes model and the 3D-1D Darcy--Poiseuille model, as well as to a 1D Green's-function model. The main results are well-posedness for the 3D-3D system (Theorem 1.2), well-posedness for the 3D-1D system (Theorem 1.3), a 3D-3D to 1D convergence estimate (Theorem 1.6), and, by chaining Theorem 1.6 with the companion result Theorem 1.5 from Part I [42], a 3D-3D to 3D-1D error estimate at rate O(epsilon^{1/6}|log epsilon|) (Theorem 1.4). The proof uses a slender-body velocity ansatz built from the 1D pressure, a cutoff isolating the degenerate tip, and a series of epsilon-dependent Korn, pressure-operator, and trace estimates proved in the appendices.","tokens_in":46831,"tokens_out":12618,"duration_ms":140766,"significance":"If the results are correct and the dependency on Part I is resolved, the paper would provide a rare quantitative link among three widely used modeling levels of tissue perfusion, with an explicit convergence rate that is valid up to the degenerate vessel tip. The construction of the velocity ansatz, the optimized epsilon^{4/3} tip cutoff, and the careful epsilon tracking in Sections 5 and Appendix A are genuine strengths, as is the explicit treatment of the free-tip geometry. However, the main theorems are conditional on unproved estimates from the companion paper [42], and one load-bearing existence proof in Section 2 contains a gap that must be repaired.","major_comments":[{"comment":"The surjectivity proof for the divergence operator on V_sigma^perp is not valid as written. The constructed field v_1 has e_theta-component proportional to c r theta (Eq. (2.17)); on the full tubular domain V_epsilon, the angle theta is a periodic coordinate and is not single-valued, so any representative of v_1 has a jump discontinuity across a branch cut. Hence v_1 is not in H^1(V_epsilon), the identity div v_1 = c fails in the weak sense (a surface distribution on the cut appears), and the decomposition in (2.18) does not produce the claimed zero-mean divergence. Consequently Corollary 2.3, the inf-sup condition, and the recovery of p_epsilon in Theorem 1.2 are not established by the given argument. This gap is local and likely repairable by a different construction, but the proof must be rewritten.","section":"§2.2, Eqs. (2.17)–(2.19)"},{"comment":"The central results are conditional on the companion paper [42]. Theorem 1.4 is obtained by chaining Theorem 1.6 with Theorem 1.5 [42], and even the proof of Theorem 1.6 invokes, without proof in this manuscript, Lemma 5.1 (weighted L^2 and L^infinity bounds (5.2)–(5.3)), Lemma 5.2 (near-theta-independence (5.4)), and Lemma 5.3 (residual estimates (5.7)–(5.8)). These imported bounds enter at load-bearing points, for example in (5.21), (5.38), and the E_9 estimate (5.58). If Part I is not independently available in refereed form or included as an appendix, the manuscript does not by itself prove Theorem 1.4. The authors should either supply the needed proofs or explicitly state that Theorems 1.4 and 1.6 are conditional on [42].","section":"§1.4, §5.1, §5.3"}],"minor_comments":[{"comment":"The phrase 'the 3D-3D to 1D convergence result of Theorem 1.5 in section 5' should refer to Theorem 1.6, not Theorem 1.5.","section":"§1.4, final paragraph"},{"comment":"The inequality 'a^{-1}(s) >= C epsilon^{-2/3}' should read 'a^{-1}(s) <= C epsilon^{-2/3}' (equivalently, a(s) >= C epsilon^{2/3}); the displayed direction is reversed and, as written, does not yield the stated epsilon^{2/3} bound.","section":"§5.3, estimate (5.58)"},{"comment":"There is a typographical error in the norm on the last line: the expression 'V x D^{1,2}(Omega_epsilon ||(w,w)||...' is missing a closing parenthesis and should read '||(u,q)||_{V x D^{1,2}(Omega_epsilon)} ||(w,w)||_{V x D^{1,2}(Omega_epsilon)}'.","section":"§2.2, Eq. (2.22)"},{"comment":"In the displayed coercivity computation, '+(zeta_epsilon nabla q, nabla q)' contains a doubled plus sign; it should be a single plus.","section":"§2.2, after Eq. (2.23)"},{"comment":"The sentence 'The proof of Theorem 1.2 appears in section 2, including justification for the choice of epsilon-scaling of the coefficients zeta epsilon^4 and kappa epsilon^3 appearing in (1.11c)' should refer to (1.10c), not (1.11c), since (1.11c) belongs to the 3D-1D system.","section":"§1.3, after Theorem 1.2"}],"recommendation":"major_revision","confidential_remarks":"The largest risk is the companion-paper dependency: Theorem 1.4 as advertised is only proved relative to [42]. If Part I is accepted or the authors provide the required estimates in an appendix, the paper becomes substantially stronger. The Lemma 2.2 gap is local but must be fixed before the existence theory for the 3D-3D system can be taken as established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, this is a genuinely new result: the first quantitative link across the 3D-3D, 3D-1D and 1D models, with an explicit epsilon^{1/6}|log epsilon| rate that holds up to the degenerate tip. The angle-averaged Neumann condition in the 3D-1D model is also new, and the formal asymptotics in Section 3 are clearly presented. Second, the paper as written has a load-bearing gap in its own proof, separate from the dependency on Part I.\n\nThe gap is in Lemma 2.2. The vector field v1 in (2.17) contains the term c r theta e_theta. On the full vessel V_epsilon, the angle theta is not single-valued; theta and theta+2pi label the same point. So v1 has a jump across a branch cut and is not in H^1(V_epsilon). The identity div v1 = c fails in the weak sense because the derivatives produce an additional surface distribution on the cut. That means the surjectivity of the divergence operator, the bound (2.19), the inf-sup condition (Corollary 2.3), and ultimately Theorem 1.2 are not justified by the given argument. Theorem 1.2 supplies the 3D-3D solution whose approximation is the paper's main object, so the error estimates in Theorems 1.4 and 1.6 currently rest on an unproven existence theorem.\n\nWhat the paper does well: the proof of Theorem 1.6, conditional on the 1D estimates, is detailed and uses standard machinery (Korn, Bogovskii-type operators, trace estimates) with careful epsilon tracking. The well-posedness of the 3D-1D model (Theorem 1.3) looks clean and self-contained. The citation pattern is honest about the relationship to Laurino-Zunino and Heltai-Zunino; no overclaiming there.\n\nThe heavy reliance on Part I (Theorem 1.5 and Lemmas 5.1-5.3) is a separate concern. Those results are not proved here and are not machine-checked; any referee should require them to be available and verifiable. But even with Part I granted, Lemma 2.2 has to be fixed.\n\nWho this is for: people working on mixed-dimensional models of perfusion will get a lot from the formal hierarchy and the 3D-1D analysis. The convergence claim itself needs repair. I would send it to peer review because the contribution is potentially important, but the referee should focus on the divergence isomorphisms and on the Part I dependency.","headline":"The 3D-3D-to-1D convergence program is novel and the asymptotic analysis is serious, but the proof of the 3D-3D existence theorem rests on a non-H^1 vector field and the main results currently do not close.","tokens_in":47342,"tokens_out":4891,"would_cite":false,"duration_ms":53102,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B25","76D07","76S05","76Z05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The pressure fields of the full 3D-3D perfusion model converge to the reduced 3D-1D model at rate C ε^{1/6}|log ε| |p0| as ε → 0, all the way to the tapering free tip.","keywords":["blood perfusion modelling","Darcy-Stokes flow","Darcy-Poiseuille reduction","1D Green's function model","slender body approximation","degenerate free tip","convergence rate","porous medium flow"],"falsifier":"Solve the 1D integrodifferential equation (1.21) numerically for a straight vessel with a spheroidal tip and check the weighted $L^\\infty$ bounds of Lemma 5.1 directly: if $\\|p_{\\mathrm{SB}}\\|_{L^\\infty}$ or $\\|a\\,p_{\\mathrm{SB},s}\\|_{L^\\infty}$ grows faster than $\\epsilon^{-1/2}|p_0|$ as $\\epsilon \\to 0$, the tip-cutoff estimates of Section 5 fail and Theorem 1.4 cannot hold. A complementary check is to solve the 3D-3D Darcy–Stokes and 3D-1D Darcy–Poiseuille systems in the same geometry and confirm that $(1/|V_\\epsilon|^{1/2})\\|p^\\epsilon-p\\|_{L^2(V_\\epsilon)} + \\|q^\\epsilon-q\\|_{D^{1,2}(\\Omega_\\epsilon)}$ decays like $\\epsilon^{1/6}|\\log\\epsilon|$ rather than saturating at a slower rate.","tokens_in":46292,"feed_emoji":"🩸","tokens_out":27811,"duration_ms":221012,"temperature":0.7,"pith_summary":"This paper is the second half of an argument that three standard descriptions of blood perfusion through tissue around a thin vessel — a full three-dimensional Darcy–Stokes system, a reduced model with a one-dimensional vessel coupled to a three-dimensional porous tissue, and a one-dimensional Green's-function model — describe the same physics in a quantitatively controlled way. Its central result is a convergence theorem: for a vessel whose radius decays to zero at the free tip, the pressures computed by the most detailed 3D-3D model differ from those of the reduced 3D-1D model by at most $C \\epsilon^{1/6}|\\log\\epsilon|\\,|p_0|$, where $\\epsilon$ is the maximum vessel radius and $p_0$ is the incoming pressure. The proof is indirect: it constructs an explicit approximate velocity field from the 1D model's pressure, proves the 3D-3D system converges directly to that 1D solution at the $\\epsilon^{1/6}|\\log\\epsilon|$ rate, and then borrows the 3D-1D-to-1D convergence of Part I, together with the 1D pressure bounds proved there, to close the remaining gap. If correct, this gives the much cheaper 1D and 3D-1D models used in microvascular simulation a rigorous error guarantee that extends up to the degenerate endpoint where the vessel becomes indistinguishable from a capillary.","feed_headline":"Proven: 3D blood-flow pressures match cheaper 1D models","feed_subtitle":"Error shrinks as ε^{1/6}|log ε| with vessel radius ε, even at the tapering tip—licensing cheap 1D models.","key_machinery":"The load-bearing object is the 1D 'slender-body' pressure $p_{\\mathrm{SB}}(s)$, the solution of the integrodifferential equation (1.21), in which the exterior pressure is expressed through the Neumann Green's function of the half-space as $q_{\\mathrm{SB}} = \\tfrac{\\pi}{8\\zeta\\mu} S_N\\big[\\tfrac{d}{ds}(a^4 \\tfrac{dp_{\\mathrm{SB}}}{ds})\\big]$. From the companion Part I the paper imports three ingredients: weighted $L^2$ and $L^\\infty$ bounds for $p_{\\mathrm{SB}}$ whose endpoint weights gain one or two powers of the radius $a(s)$ near the tip (Lemma 5.1), a near-$\\theta$-independence estimate for $q_{\\mathrm{SB}}$ on the vessel surface (Lemma 5.2), and residual bounds showing that $(p_{\\mathrm{SB}}, q_{\\mathrm{SB}})$ almost satisfies the 3D-1D weak form (Lemma 5.3). The comparison object is the velocity ansatz $U$ of (5.17), built from the variable-radius Poiseuille law and multiplied by a cutoff $\\phi_\\epsilon(s)$ that vanishes on the final tip segment of length $\\epsilon^{4/3}$; the cutoff length is the optimized balance between the ansatz failing near the curved tip and the geometric smallness of the tapering end, and that balance produces the $\\epsilon^{1/6}|\\log\\epsilon|$ rate. Supporting this core, Appendix A supplies $\\epsilon$-scaled Korn, pressure-operator, and trace inequalities for the slender vessel with a degenerate free end.","core_discovery":"The central claim, Theorem 1.4, is that for sufficiently small $\\epsilon$ the pressure difference between the 3D-3D Darcy–Stokes solution $(u^\\epsilon, p^\\epsilon, q^\\epsilon)$ and the 3D-1D Darcy–Poiseuille solution $(p,q)$ satisfies $(1/|V_\\epsilon|^{1/2})\\|p^\\epsilon-p\\|_{L^2(V_\\epsilon)} + \\|q^\\epsilon-q\\|_{D^{1,2}(\\Omega_\\epsilon)} \\le C \\epsilon^{1/6}|\\log\\epsilon|\\,|p_0|$, with $C$ independent of $\\epsilon$, and the estimate holds up to the free tip $s=1$ where the vessel radius decays spheroidally to zero and the reduced models become degenerate. Because the angle-averaged Robin condition in the 3D-1D model does not easily control the exterior pressure pointwise on the vessel surface, the proof detours through the 1D slender-body model: a velocity ansatz built from the variable-radius Poiseuille law is cut off in a tip region of length $\\epsilon^{4/3}$, and the 3D-3D system is shown to converge to this explicit 1D solution (Theorem 1.6) at the same rate; the 3D-3D-to-3D-1D result then follows from the Part I estimate of Theorem 1.5. The $\\epsilon^{1/6}$ rate is entirely a consequence of the free-end behavior; away from the degenerate endpoint the limiting factor would instead be the faster Part I rate $\\epsilon^{1/2}|\\log\\epsilon|$.","pith_inferences":["Because the paper's route goes 3D-3D to 1D and then re-uses the 3D-1D-to-1D result, the convergence of the full model hinges far more on the quality of the 1D pressure bounds (Lemma 5.1) than on the details of the 3D-1D coupling; this suggests the 3D-1D model's role in the hierarchy is largely motivational and structural.","The $\\epsilon^{1/6}$ exponent is probably not sharp: a boundary-layer ansatz that correctly describes the flow near the spheroidal tip should push the rate toward the faster $\\epsilon^{1/2}|\\log\\epsilon|$ of Part I, since away from the tip the present proof is limited only by that faster rate.","The half-space geometry is used purely for the explicit Neumann Green's function; for a bounded domain with a vessel reconnecting to the boundary at both ends, one would expect an analogous and faster convergence with no degenerate tip, a setting the paper leaves as future work.","A numerical check of Lemma 5.1 on the 1D equation (1.21) would independently test the assumed Part I input, since the weighted $L^\\infty$ bounds can be computed without ever solving the 3D systems."],"forward_implications":["The full 3D-3D Darcy–Stokes description of perfusion can be replaced, with a certified error $C\\epsilon^{1/6}|\\log\\epsilon|\\,|p_0|$, by the reduced 3D-1D Darcy–Poiseuille system for a thin vessel whose radius decays at the free tip.","Through Part I's Theorem 1.5, the same guarantee descends to the fully 1D Green's-function model, closing the three-level hierarchy at a single rate.","The estimate is valid all the way to the endpoint $s=1$, where the vessel radius vanishes and neither reduced model admits a classical boundary value, so the degenerate capillary-scale tip does not spoil the approximation.","The vessel-interior velocity field is also approximated by the explicit Poiseuille-type ansatz $U$ at the same $\\epsilon^{1/6}|\\log\\epsilon|$ rate, in the sense of Theorem 1.6.","The $\\epsilon^{1/6}$ rate is imposed entirely by the free end; refined asymptotics for the tip boundary layer would give more accurate but more complicated reduced models."],"supporting_citations":[{"why":"It supplies the 3D-1D-to-1D convergence of Theorem 1.5 and the weighted pressure bounds, near-θ-independence, and residual estimates of Lemmas 5.1–5.3 that Section 5 assumes; the ε^{1/6} argument cannot close without them.","marker":"[42]"},{"why":"It provides the free-end slender-body machinery—extension operator, non-uniform tip partition, trace inequalities—adapted in Appendix A for the Korn and endpoint-trace lemmas.","marker":"[37]"},{"why":"It is the source of the two-scale thin-domain estimates yielding the ε-scaled Korn and pressure-operator bounds used in the energy and error estimates.","marker":"[30]"},{"why":"It supplies the surjectivity of the divergence operator used in the inf-sup argument for 3D-3D well-posedness and in the construction of the pressure operator with its ε^{-1} scaling.","marker":"[18]"},{"why":"It provides the Hardy-type inequality used to prove the weighted Poincaré inequality of Lemma 4.2 in the 3D-1D well-posedness argument.","marker":"[32]"},{"why":"It introduces the angle-averaged Neumann-to-Dirichlet map that motivates the geometrically constrained coupling defining the 3D-1D system compared here.","marker":"[41]"}],"fun_headline_variants":["Blood flow: 3D to 1D convergence proven rigorously","Cheaper 1D blood flow models validated by new proof","3D blood flow models shrink to 1D with proven error","Tapering vessels included: 3D model matches 1D","New proof: 1D blood flow models suffice for thin vessels"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The convergence proof stands on endpoint-weighted estimates for the one-dimensional model's pressure near the vessel tip that are proved in the companion Part I paper and assumed rather than reproved here; if those estimates fail, the $\\epsilon^{1/6}|\\log\\epsilon|$ rate does not close.","fun_headline_variants_meta":{"raw":{"variants":["Blood flow: 3D to 1D convergence proven rigorously","Cheaper 1D blood flow models validated by new proof","3D blood flow models shrink to 1D with proven error","Tapering vessels included: 3D model matches 1D","New proof: 1D blood flow models suffice for thin vessels"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001209,"raw_usage":{"total_tokens":5074,"prompt_tokens":1139,"completion_tokens":3935,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":755,"completion_tokens_details":{"reasoning_tokens":3844}},"tokens_in":755,"tokens_out":3935,"duration_ms":30891,"temperature":1.0,"reasoning_tokens":3844,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:26:29.249700+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Solve the 1D integrodifferential equation (1.21) numerically for a straight vessel with a spheroidal tip and check the weighted $L^\\infty$ bounds of Lemma 5.1 directly: if $\\|p_{\\mathrm{SB}}\\|_{L^\\infty}$ or $\\|a\\,p_{\\mathrm{SB},s}\\|_{L^\\infty}$ grows faster than $\\epsilon^{-1/2}|p_0|$ as $\\epsilon \\to 0$, the tip-cutoff estimates of Section 5 fail and Theorem 1.4 cannot hold. A complementary check is to solve the 3D-3D Darcy–Stokes and 3D-1D Darcy–Poiseuille systems in the same geometry and confirm that $(1/|V_\\epsilon|^{1/2})\\|p^\\epsilon-p\\|_{L^2(V_\\epsilon)} + \\|q^\\epsilon-q\\|_{D^{1,2}(\\Omega_\\epsilon)}$ decays like $\\epsilon^{1/6}|\\log\\epsilon|$ rather than saturating at a slower rate.","supporting_citations":[{"cited_title":"Ohm and S","cited_arxiv_id":null,"evidence_quote":"It supplies the 3D-1D-to-1D convergence of Theorem 1.5 and the weighted pressure bounds, near-θ-independence, and residual estimates of Lemmas 5.1–5.3 that Section 5 assumes; the ε^{1/6} argument cannot close without them."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides the free-end slender-body machinery—extension operator, non-uniform tip partition, trace inequalities—adapted in Appendix A for the Korn and endpoint-trace lemmas."},{"cited_title":"Maruˇ si´ c and E","cited_arxiv_id":null,"evidence_quote":"It is the source of the two-scale thin-domain estimates yielding the ε-scaled Korn and pressure-operator bounds used in the energy and error estimates."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the surjectivity of the divergence operator used in the inf-sup argument for 3D-3D well-posedness and in the construction of the pressure operator with its ε^{-1} scaling."},{"cited_title":"Mironescu","cited_arxiv_id":null,"evidence_quote":"It provides the Hardy-type inequality used to prove the weighted Poincaré inequality of Lemma 4.2 in the 3D-1D well-posedness argument."},{"cited_title":"On an angle-averaged Neumann-to-Dirichlet map for thin filaments","cited_arxiv_id":"2308.06592","evidence_quote":"It introduces the angle-averaged Neumann-to-Dirichlet map that motivates the geometrically constrained coupling defining the 3D-1D system compared here."}],"review_version":1}