{"id":"3e830cdf-7584-4597-8056-762fa6df241e","arxiv_id":"2506.22558","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A variational phase-field model with strength degradation produces cohesive cracks whose sharp-interface limit is an explicit cohesive fracture law for any convex strength domain.","lead":"This paper presents a phase-field fracture model in which damage reduces the material's strength limit instead of its stiffness, enabling arbitrary convex strength criteria. It shows that under compatible stress states the model produces cohesive cracks, and derives the corresponding sharp-interface cohesive laws for several classical strength criteria.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The sharp-interface cohesive energy (10) is postulated, not proved: no Gamma-convergence result links (24) to (9), and the paper's own stability analysis is only formal.","rationale":"The reader's verdict is already CONDITIONAL, and its weakest assumption matches exactly: the sharp-interface limit is postulated. In good faith, the paper's constructive core, namely the explicit solutions on the cube, the compatibility characterization via Mohr's circles, and the hierarchy of strength criteria, is plausible and carefully argued, and the renormalized cohesive law (10) has a natural derivation from the localized energy (54)-(56). I therefore do not see an internal contradiction requiring rejection. However, the load-bearing step is the passage l to 0. Without Gamma-convergence or compactness, the limit energy may depend on the chosen ansatz; global optimality of the localized solution is only shown among phase-fields in D(alpha_bar), and the homogeneous branch is destabilized only by a formal perturbation argument. The antiplane constant-stiffness Gamma-convergence question isolates this step in the simplest setting where the proposed model differs from the existing result in [38]. The paper itself says that complete mathematical results and conclusive numerical tests are needed before the approach can serve as a general theory, so the CONDITIONAL verdict is appropriate and no change is needed.","tokens_in":52829,"tokens_out":7826,"duration_ms":96319,"concrete_test":"Prove or disprove Gamma-convergence of the antiplane specialization of (24) with constant shear modulus mu0, strength set {|p| <= tau_c}, k(alpha)=1-alpha, w(alpha)=alpha^2, and k(1)=0, to E0(u)=int psi0(nabla u) + int_{Ju} phi([[u]]) with phi from (10), following the strategy of [38] but with mu independent of alpha. If this scalar constant-stiffness benchmark does not Gamma-converge, the general vectorial claim in Section 5 is invalid; if it does, the same conclusion should be tested numerically for a mode-II cube with a Drucker-Prager cone, comparing finite-l minimizers of (24) at l/L tending to 0 with the closed-form localized response in equations (42), (53), and (56).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that energy (8)/(24) yields a cohesive crack whose sharp-interface limit is the cohesive model (9)-(10) requires that E0 with surface density phi([[u]],alpha) = k_hat(alpha) H_K0(n circles [[u]]) + Gc alpha be the Gamma-limit of (24) as l goes to 0. The paper does not prove this. The abstract and Section 5.1 state that the sharp-interface limit is 'postulated' and 'for a large part formal'. The only cited convergence result ([38]) concerns a different antiplane model with alpha-dependent shear modulus mu(alpha)=(1-alpha)mu0 and without irreversibility; the paper itself notes that extending it to constant mu0 and k(1)=0 would require reworking the entire argument. Section 3.4's instability analysis is explicitly 'would require rigorous justification', and Proposition 3.2 establishes optimality only within the restricted class D(alpha_bar), with beta <= alpha_bar and beta = alpha_bar on Gamma, not global minimality. Hence the identification of the sharp-interface surface energy rests on one specially constructed cube solution, not on a variational convergence theorem. If the limit is not unique, for example if another recovery sequence produces different surface dissipation, or if the homogeneous branch competes outside D(alpha_bar), the claimed unification of strength, cohesive fracture, and Griffith energy is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a variational phase-field model of fracture in which the phase-field variable degrades a convex strength domain K0 homothetically (K(alpha)=k(alpha)K0) without affecting the linear stiffness A0, leading to the energy (24). The central object is a multiaxial model problem on a cube (Problem 3.1) with imposed average strain in a fixed stress direction s. The authors construct two explicit solution families: a homogeneous nonlinear-elastic/damaging solution and a localized solution with a displacement jump and damage concentrated on a cross-section, valid when the normal to the strength surface is of the form d ⊙ n. They analyze stability of these solutions, characterize jump compatibility for isotropic strength criteria using Mohr's circles and an intrinsic domain K0, and postulate a sharp-interface cohesive model with surface energy (10): phi([[u]], alpha) = k_hat(alpha) H_{K0}(n ⊙ [[u]]) + Gc alpha. Several examples (von Mises, Drucker-Prager, Tresca with tension cut-off) illustrate the hierarchy of strength criteria for which cracks can nucleate. The paper explicitly acknowledges that the sharp-interface limit is postulated and that the stability analysis is partly formal.","tokens_in":53132,"tokens_out":5815,"duration_ms":64868,"significance":"If the formal sharp-interface identification can be made rigorous, the paper would unify strength criteria, cohesive fracture, plasticity, and Griffith-type surface energy in a single variational framework, which is a significant conceptual contribution to phase-field fracture modeling. The paper's strengths include a parameter-free derivation of the cohesive surface energy from the underlying convex strength domain (no data fitting), explicit analytical solutions of the cube model problem, and a clean geometric characterization (Proposition 4.3) of jump-compatible stress states via the intrinsic domain, with concrete consequences for classical criteria such as von Mises versus Tresca. The paper is honest about its limitations: the sharp-interface energy (9)-(10) is postulated rather than proved as a Gamma-limit of (24), and the stability results are either restricted to a subclass D(alpha) or labeled as requiring rigorous justification. The value of the paper currently rests on a well-motivated conjecture supported by explicit solutions, rather than on a proven convergence theorem.","major_comments":[{"comment":"The sharp-interface energy (9)-(10) is introduced as a postulate rather than derived as the Gamma-limit of the phase-field energy (24); the abstract and Section 5.1 state this explicitly ('we postulate' and 'for a large part formal'). The only rigorous convergence result cited, [38], concerns a different antiplane model with an alpha-dependent shear modulus and no irreversibility, and the paper itself notes that transferring it to constant mu0 and k(1)=0 'would require reworking the entire argument'. This is load-bearing because the central claim that the phase-field model regularizes the cohesive law (10) rests on the single constructed cube solution of Section 3.3 and the formal stability argument of Section 3.4. Please either provide a Gamma-convergence result, or a rigorous matched-asymptotic derivation of (10) from (24) in the cube geometry, or explicitly state the result as a conjecture whose evidence is the constructed solution and the heuristic analysis.","section":"Abstract and Section 5.1, Eqs. (9)-(10)"},{"comment":"The stability conclusions are not established in the full admissible class. Proposition 3.2 proves minimality only among phase-fields in D(alpha_bar), which imposes beta(x) ≤ alpha_bar everywhere and beta = alpha_bar on Gamma; the proof's lower bound (57) exploits k(beta) ≥ k(alpha_bar) and therefore does not control perturbations with beta > alpha_bar away from Gamma. The subsequent instability analysis of the homogeneous response is explicitly qualified by the authors as 'would require rigorous justification' and relies on heuristic concentration estimates for the terms T1-T4 without a specified norm or compactness framework (see also Remark 7, which leaves the notion of 'neighborhood' in the stability condition undefined). Since the paper's conclusions in Section 3.5 that 'the localized solutions are stable' and 'the homogeneous solutions are unstable' depend on these assertions, the statements should be softened to reflect that they are proven only within D(alpha_bar) and under a formal perturbation argument, or the paper should supply a rigorous local-minimality proof in a well-defined function space.","section":"Section 3.4, Proposition 3.2 and the perturbation argument"},{"comment":"The variational formulation is not fully specified: the admissible spaces C, D, and P are said to be 'suitable' without concrete definitions, and Remark 5 explicitly neglects Cantor-type terms in the strain decomposition. This matters beyond technical completeness because the stability condition (29a) and the construction of localized solutions with p as a measure on the jump set (Eq. (26)) require a precise functional framework to decide whether the formal constructions satisfy the stability notion. Please specify the function spaces (e.g., BD for u, measures for p, H^1 for alpha), state the exact meaning of local stability in (29a), and justify the neglect of Cantor terms for the solutions analyzed in Section 3.","section":"Section 2.3 and Remark 5"}],"minor_comments":[{"comment":"The inequality '0 ≠ sI - sIII/2τc > sI/σc' is ambiguous; it should read (sI - sIII)/(2τc) > sI/σc > 0 to match the intended comparison of the two ratios.","section":"Section 5.3.4, case (A)"},{"comment":"The entry for 'νs' contains a self-referential definition: 'Normal to the yield surface ∂K0 in the direction s and its s-projection νs := νs · s' defines νs twice; use separate symbols for the normal and its projection.","section":"Table 1"},{"comment":"The definition of ϕ*(δ) in (82) uses a dimensionless argument δ, and (83) sets δ = F0∥d∥/Gc; this is consistent, but the text would benefit from stating explicitly that F(∥d∥) is the derivative with respect to ∥d∥, not with respect to the argument δ.","section":"Section 5.2.1, Eq. (83)"},{"comment":"The sentence 'The authors report that they were unable to extend the Γ-convergence result...' is ambiguous about whether 'the authors' refers to the present authors or to the authors of [38]; rephrase to attribute the statement clearly, e.g., 'Dal Maso et al. report that...'.","section":"Section 5.3.2"},{"comment":"The name 'Hencky' is misspelled as 'Henky' in Section 5.1 ('standard Henky-like law') and as 'a laHencky' in Section 2.2.1; unify the spelling.","section":"Sections 2.2.1 and 5.1"}],"recommendation":"major_revision","confidential_remarks":"The authors are transparent about the formal nature of the sharp-interface limit and the restricted stability results, which is commendable. The main gap is the lack of a Gamma-convergence or rigorous asymptotic verification connecting (24) to (9)-(10); without it, the paper's central unification claim is a conjecture, albeit a well-supported one. The paper fits the journal's scope and the explicit solutions and Mohr-circle compatibility characterization are novel contributions. I recommend major revision rather than rejection because the core derivation is internally consistent and the limitations could be addressed by either a proof of the limit (perhaps under additional hypotheses) or by a clear reframing of the main result as a conjecture with a roadmap. A minor strengthening of the paper would be to add a small numerical example illustrating the convergence of the phase-field solution to the predicted cohesive law, which would increase confidence in the formal limit."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: this paper does real work on a long-standing problem—how to get a phase-field model that respects a prescribed multiaxial strength criterion while still producing cohesive cracks with finite toughness. The key idea is to degrade the strength domain instead of the stiffness, and the authors carry it through with explicit constructions and a clean geometric criterion.\n\nWhat is actually new: the Mohr-circle compatibility condition (Proposition 4.3) and the resulting hierarchy of strength criteria (never/sometimes/always crack) are a genuine step forward. The explicit cube solution for a cohesive crack, with the cohesive law derived from the phase-field ingredients, is elegant and useful. The paper also gives explicit support functions for several classical criteria, which will be handy for anyone implementing this class of models. The writing is honest: the authors repeatedly state that the sharp-interface limit is postulated and that parts of the stability analysis are formal.\n\nThe soft spots are exactly where the authors point. The sharp-interface energy (9)–(10) is not proven to be the Gamma-limit of (24). The only existing Gamma-convergence result in this family, by Dal Maso et al., handles the antiplane case with stiffness degradation and without irreversibility; extending it to the present setting would require substantial new work. The stability of the localized solution is proven only within a restricted class of phase fields, and the instability argument for the homogeneous solution is heuristic. These are real limitations for a paper that ultimately proposes a unified theory. But they are not hidden, and they do not undermine the constructive results—the cube solution and the cohesive law derivation stand on their own.\n\nWho should read this: anyone working on variational fracture, phase-field models of cracking, or limit analysis who wants a principled way to incorporate strength criteria into a fracture model. It would be a good reading-group paper because the geometric characterization is thought-provoking and the formal steps invite discussion.\n\nRecommendation: send it to peer review. A serious referee can verify the convex analysis and the cube construction, and the formal sharp-interface limit can be flagged as an open problem rather than treated as a fatal flaw. I would suggest the authors state even more clearly which statements are proven and which are conjectured, and ideally add a numerical example to show the model behaves as advertised at finite regularization length.","headline":"A genuinely useful extension of phase-field fracture to arbitrary convex strength criteria, with the sharp-interface limit honestly labeled formal rather than proven.","tokens_in":53649,"tokens_out":1579,"would_cite":true,"duration_ms":20337,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["74R10","74C05","74G65"],"pacs":["46.50.+a"],"model":"deepseek-v4-flash","headline":"This paper builds any closed convex strength criterion into a variational phase-field fracture model, and derives the cohesive crack law that emerges as the sharp-interface limit.","keywords":["phase-field fracture","convex strength criterion","cohesive crack","Mohr's circles","intrinsic domain","variational fracture","limit analysis","sharp-interface limit"],"falsifier":"Compute the $\\Gamma$-limit of the phase-field energy (24) in three-dimensional vector elasticity: if minimizers converge to a limiting energy different from (9) — for instance, if the surface term depends on the full phase-field profile rather than only on the rescaled pair $(\\hat k,\\hat\\alpha)$, or if the Cantor part of the strain contributes — the central claim fails. A cheaper test: run the phase-field model on the cube with a von Mises strength domain under uniaxial tension; the paper predicts that the homogeneous solution remains stable and no crack nucleates, so observing a nucleated transverse crack in that simulation would falsify the compatibility criterion.","tokens_in":52568,"feed_emoji":"💥","tokens_out":17896,"duration_ms":156777,"temperature":0.7,"pith_summary":"The paper sets out to build a single variational theory of fracture that can encode any closed convex strength criterion — von Mises, Tresca, Drucker-Prager, Mohr-Coulomb, with or without tension cut-offs — while still producing a Griffith-type surface energy for fully developed cracks. In the proposed phase-field model, the damage variable does not degrade the elastic stiffness; instead it shrinks the admissible stress domain toward zero, so that the elastic energy grows only linearly outside the strength domain. The central result is that, in a fundamental cube problem, a transverse cohesive crack nucleates and grows exactly when the loading direction is compatible with displacement jumps, a condition that, for isotropic criteria, is read off the Mohr representation of the stress. The paper then postulates the sharp-interface limit and derives an explicit cohesive law whose parameters come from the strength domain itself. If the construction is correct, crack nucleation (strength), crack propagation (toughness), damage, plasticity, and limit analysis are unified in one variational framework.","feed_headline":"Phase-field model folds any convex strength criterion into fracture","feed_subtitle":"Cracks nucleate only along jump-compatible stress directions, and the cohesive law is derived, not assumed.","key_machinery":"The load-bearing object is the closed convex strength domain $K_0$ and its normal cone. A stress state $\\sigma\\in\\partial K_0$ admits a displacement jump exactly when some outer normal is a symmetrized tensor product $\\nu=d\\odot n$ of a jump direction $d$ and a crack normal $n$, with $d\\cdot n\\ge 0$ encoding non-interpenetration. For isotropic criteria the paper recasts this condition in Mohr's representation: the set of admissible traction vectors on any plane is a surface of revolution, the intrinsic domain, which is the envelope of all Mohr's circles, and jumps are possible exactly at stress states whose stress vector lies on the boundary of that domain. The paper shows that the support function of this intrinsic domain, evaluated on the Mohr components of a displacement jump, equals $H_{K_0}(n\\odot\\llbracket u\\rrbracket)$, the stress-space support function applied to the symmetrized jump; this quantity is what converts the phase-field energy into the cohesive law of the sharp-interface limit.","core_discovery":"The paper claims that fracture with an arbitrary closed convex strength criterion can be captured by a variational phase-field energy in which the phase-field variable leaves the elastic stiffness untouched and instead contracts the admissible stress set: $E_\\ell(u,p,\\alpha)=\\int_\\Omega\\big[\\tfrac12 A_0(\\varepsilon(u)-p)\\cdot(\\varepsilon(u)-p)+k(\\alpha)H_{K_0}(p)\\big]\\,dV+\\frac{G_c}{4c_w}\\int_\\Omega\\big[\\tfrac{w(\\alpha)}{\\ell}+\\ell|\\nabla\\alpha|^2\\big]\\,dV$. In a model problem on a cube loaded by a prescribed average strain along a stress direction $s$, whenever the outer normal $\\nu_s$ to $\\partial K_0$ at the yield point $\\sigma_c^s s$ is of the form $\\nu_s=d\\odot n$ — a direction compatible with a displacement jump — the paper constructs an explicit solution in which the plastic strain concentrates as a measure on a cross-section of the cube, the damage attains its maximum there, and the stress decreases as the opening grows: the nucleation and progressive evolution of a cohesive crack. For isotropic criteria the same condition becomes the statement that the stress vector reaches the boundary of the intrinsic domain, the envelope of Mohr's circles, which yields a hierarchy in which fracture is never, sometimes, or always possible depending on the criterion. Postulating the sharp-interface limit, the paper derives the cohesive surface energy $\\phi(\\llbracket u\\rrbracket,\\hat\\alpha)=\\hat k(\\hat\\alpha)H_{K_0}(n\\odot\\llbracket u\\rrbracket)+G_c\\hat\\alpha$ on the jump set, with a bulk energy $\\psi_0(\\varepsilon)$ of linear growth, so that strength, toughness, and damage are fixed by a single variational construction.","pith_inferences":["Looking beyond the paper, the compatibility criterion ranks classical strength criteria by a purely geometric property — how much of their yield surface is fracture-ready — so the framework could be used prospectively to select or design yield surfaces for applications where crack nucleation is either wanted or unwanted.","If the postulated sharp-interface limit is later established by a rigorous $\\Gamma$-convergence proof in full three-dimensional vector elasticity, the model would provide existence of quasistatic evolutions for cohesive fracture with arbitrary convex strength criteria, generalizing the antiplane von Mises result that is currently the only rigorous case.","One test the paper does not spell out: under uniaxial tension a von Mises material should never nucleate a crack, whereas under pure shear it should; a biaxial experiment comparing those two loading modes could separate this theory from nucleation criteria that depend only on a scalar stress threshold.","The model implies criterion-dependent residual strength after full damage — a von Mises crack can still carry hydrostatic stress, while a Tresca crack with tension cut-off can carry only compression — which is a measurable prediction for post-peak testing of confined specimens."],"forward_implications":["Any closed convex strength domain can be inserted into a variational phase-field fracture computation, and the cohesive law that governs the resulting crack is derived from that domain rather than chosen independently.","For isotropic criteria, crack nucleation is decided by a single geometric condition in Mohr's representation — the stress vector must reach the boundary of the intrinsic domain — which splits criteria into classes where fracture is never possible, sometimes possible, or always possible.","For loading directions that are not compatible with a displacement jump, the homogeneous response remains stable up to a critical strain that diverges as the regularization length goes to zero, recovering the classical Griffith conclusion that nucleation is impossible.","Bulk strength and cohesive surface energy are no longer independent inputs: the support function of the strength domain appears directly in the cohesive law, linking quantities that classical phase-field models keep separate.","The model contains perfect plasticity and the classical Griffith model as limiting cases, with the damage variable tracking the fraction of the critical energy-release rate $G_c$ that has been dissipated."],"supporting_citations":[{"why":"Gives the only existing Gamma-convergence result for this class of models, in antiplane shear with a von Mises criterion; the result the paper conjectures to generalize to full elasticity.","marker":"[38]"},{"why":"Supplies the functional setting (spaces of bounded deformation, strain measures) and the quasistatic evolution principles for elasto-perfectly-plastic materials on which the model is built.","marker":"[2]"},{"why":"Relaxation result showing that bulk strength and cohesive surface energy are linked; the paper inverts the construction by starting from the bulk strength and deriving the cohesive law.","marker":"[4]"},{"why":"Source of the limit-analysis machinery — support functions of strength domains, Mohr-Coulomb and Drucker-Prager criteria, intrinsic curves — used throughout Sections 4 and 5.","marker":"[15]"},{"why":"Introduces gradient damage coupled with plasticity and the nucleation of cohesive cracks; the model family this paper generalizes to arbitrary convex strength domains.","marker":"[36]"},{"why":"Derives intrinsic curves in the Mohr diagram for variational cohesive models, the class of surface energies to which the sharp-interface limit belongs.","marker":"[40]"},{"why":"Provides the analysis of uniqueness and stability of the homogeneous response in gradient damage models, the technique adapted here to the multiaxial cube problem.","marker":"[7]"},{"why":"Frames the variational approach to fracture — irreversibility, stability, energy balance — that the evolution problem of Section 2 follows.","marker":"[47]"},{"why":"Characterizes which normals of the von Mises yield surface are compatible with displacement jumps, the prototype of the paper's compatibility criterion.","marker":"[3]"}],"fun_headline_variants":["Phase-field fracture: strength domain shrinks, stiffness fixed","Cohesive law derived, not assumed, from strength criteria","Fracture only along jump-compatible stress directions","Unified variational theory of failure from one model"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that, as the regularization length goes to zero, the phase-field energy truly converges to the cohesive limit energy the paper writes down — a $\\Gamma$-convergence statement that is postulated, not proven, and that the paper itself describes as being 'for a large part formal.'","fun_headline_variants_meta":{"raw":{"variants":["Phase-field fracture: strength domain shrinks, stiffness fixed","Cohesive law derived, not assumed, from strength criteria","Fracture only along jump-compatible stress directions","Unified variational theory of failure from one model"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000409,"raw_usage":{"total_tokens":2226,"prompt_tokens":1152,"completion_tokens":1074,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":768,"completion_tokens_details":{"reasoning_tokens":1010}},"tokens_in":768,"tokens_out":1074,"duration_ms":11662,"temperature":1.0,"reasoning_tokens":1010,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:03:49.100323+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the $\\Gamma$-limit of the phase-field energy (24) in three-dimensional vector elasticity: if minimizers converge to a limiting energy different from (9) — for instance, if the surface term depends on the full phase-field profile rather than only on the rescaled pair $(\\hat k,\\hat\\alpha)$, or if the Cantor part of the strain contributes — the central claim fails. A cheaper test: run the phase-field model on the cube with a von Mises strength domain under uniaxial tension; the paper predicts that the homogeneous solution remains stable and no crack nucleates, so observing a nucleated transverse crack in that simulation would falsify the compatibility criterion.","supporting_citations":[{"cited_title":"Cours de calcul des structures an´ elastiques","cited_arxiv_id":null,"evidence_quote":"Source of the limit-analysis machinery — support functions of strength domains, Mohr-Coulomb and Drucker-Prager criteria, intrinsic curves — used throughout Sections 4 and 5."},{"cited_title":"Archive for Rational Mechanics and Analysis 214(2), 575–615 (2014) https: //doi.org/0.1007/s00205-014-0763-8","cited_arxiv_id":null,"evidence_quote":"Introduces gradient damage coupled with plasticity and the nucleation of cohesive cracks; the model family this paper generalizes to arbitrary convex strength domains."},{"cited_title":"European Journal of Mechanics A/Solids 25(4), 649–669 (2006) https://doi.org/10.1016/ j.euromechsol.2006.05.00","cited_arxiv_id":null,"evidence_quote":"Derives intrinsic curves in the Mohr diagram for variational cohesive models, the class of surface energies to which the sharp-interface limit belongs."},{"cited_title":"Interfaces and Free Boundaries 17(4), 497–516 (2015) https://doi.org/10.4171/IFB/ 351 46","cited_arxiv_id":null,"evidence_quote":"Characterizes which normals of the von Mises yield surface are compatible with displacement jumps, the prototype of the paper's compatibility criterion."}],"review_version":1}