Generic regularity of intermediate complex structure limits

Valentino Tosatti; Yang Li

REVIEW 2 major objections 3 minor 2 cited by

Generic regularity of intermediate complex structure limits

T0 review · 2 major / 3 minor · reviewed 2026-08-03 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that on the generic region of an intermediate complex structure limit of Calabi-Yau manifolds, the Ricci-flat metric converges to an explicit ansatz metric in C^0, and in C^∞ after stretching the base and torus-fiber coordi

desk verdict Genuinely new metric regularity for intermediate complex structure limits, but the transfer from the model case to the actual family leans on an asserted exponential error estimate that is only sketched. read the letter →

arxiv 2511.04651 v2 pith:AINF45LS submitted 2025-11-06 math.DG math.AP

Yang Li , Valentino Tosatti This is my paper

classification math.DGmath.AP MSC 53C2553C5532Q2514J3235J96

keywords Calabi-YaumetricsintermediatecomplexstructurelimitdegenerationcollapsingansatzmetricgenericregionMonge-Ampèreequationconvergence

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper studies what happens to the Ricci-flat metric on a Calabi-Yau manifold as the complex structure degenerates at an intermediate complex structure limit — a middle ground between the large complex structure limit and no collapse. Previous work showed that the Kähler potentials of these metrics converge in C^0 to a potential solving a non-Archimedean Monge-Ampère equation. This paper upgrades that to genuine metric convergence: on a generic region carrying almost all the volume, the Ricci-flat metric converges to an explicit ansatz metric in C^0, and after stretching the torus-fiber and base coordinates by |log t|^{1/2}, the convergence is smooth with quantitative bounds. This gives the first general metric description of collapsing Calabi-Yau metrics at intermediate limits, extending the known large-complex-structure-limit picture.

What carries the argument

The engine is a truncated version of the small-perturbation bootstrap from elliptic PDE theory. The model geometry is a fibration over a ball in R^m with fibers that are T^m-bundles over a compact Calabi-Yau (n−m)-fold Y; the base directions collapse at rate |log t|^{-1/2} while the torus fibers collapse at rate |log t|^{-1}. The paper proves a Harnack inequality and a blow-up (De Giorgi-type) quadratic approximation that are valid only down to the torus-fiber scale, using the fact that the fiberwise maximum and minimum of the potential are viscosity sub- and supersolutions of a real Monge-Ampère equation on the base. Once a quadratic approximation is obtained at the fiber scale, a classical

What would settle it

Solve the model complex Monge-Ampère equation (2.10) numerically for a T^m-bundle over an elliptic curve (m=1, so Z = T^2) on a base ball of radius r = 100 T^{-1/2}. The truncated Harnack inequality predicts that the oscillation of ψ over the half-radius ball is at most Θ times the oscillation over the full ball, with Θ < 1 independent of T. A computed sequence showing oscillation decay slower than any fixed Θ down to the scale T^{-1/2} would falsify Theorem 3.2, which is the core estimate underlying Theorem 1.1.

Watch

Extended reading notes

Core claim

The central claim is that the Ricci-flat Kähler metric ω_CY,t on a degenerating Calabi-Yau family at an intermediate complex structure limit converges, on the generic region U_t ⊂ X_t, to an explicit ansatz metric ω_t constructed from the non-Archimedean potential: ∥ω_CY,t − ω_t∥_{C^0(U_t,ω_t)} → 0 as t→0. The generic region excludes only a set of arbitrarily small normalized Calabi-Yau measure. The underlying estimate is stronger: in coordinates that stretch the torus-fiber directions and the base directions by |log t|^{1/2}, the difference satisfies T∥i∂∂ψ_T∥_{C^k} ≤ C(k)(ε + T^{-α/2}) for T = |log t|, so the convergence is actually smooth at that scale. The proof works by establishing the

Load-bearing premise

The proof of the main theorem reduces the actual degeneration to a model fibration by assuming that the holomorphic volume form of the family differs from the model's by an exponentially small amount in T = |log t|; if the deviation were merely polynomially small, the truncated Harnack and Hölder estimates would not close and the conclusion would not follow.

Editorial extensions

If this is right

On the generic region U_t, the Calabi-Yau metric and the ansatz metric are mutually uniformly equivalent as t→0, and their C^0 distance goes to zero.
After stretching both the torus-fiber and base coordinates by |log t|^{1/2}, the metrics converge in C^k for every k, with bounds that are polynomial in T = |log t|.
The measure of the complement of the generic region can be made arbitrarily small, so the description covers almost all of the volume of X_t.
The proof gives the intermediate-limit analogue of the large-complex-structure-limit smooth convergence, completing that program for all 0 < m ≤ n.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The truncated-Harnack strategy is likely to extend to degenerations whose essential skeleton is any simplicial complex, not just a simplex, as long as the model fibers are homogeneous torus bundles over a Calabi-Yau base; the exponential error estimate on the volume form is probably generic for normal-crossing degenerations.
A quantitative rate is implicit in the estimates: the C^0 distance scales like |log t|^{-α/2} (up to the ε-term), so one can read off a rate of Gromov-Hausdorff convergence of the rescaled metrics to the Euclidean cone over the dual complex.
One could test the ansatz directly in explicit examples (e.g., toric hypersurfaces with m=1) by computing the Monge-Ampère energy of ω_t; the theorem predicts this energy approaches the Calabi-Yau energy with error O(|log t|^{-α/2}).
The reliance on exponential smallness of the model error suggests a robustness principle: any degeneration whose volume form deviates from the model by a sufficiently high power of t (rather than exponentially) may still admit the same estimates with a polynomial loss; the threshold is exactly the scale at which the truncated estimates are used.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague—

You should know two things about arXiv:2511.04651. First, it proves a genuinely new regularity result: for intermediate complex structure limits (0

What is new and good. The model-case analysis in Sections 3–4 is the core contribution: a truncated Harnack inequality and a De Giorgi blow-up argument that work only up to the torus scale T^{-1/2}, because the fiber is a T^m-bundle over a still-collapsing Calabi-Yau Y. That two-scale difficulty is real, and the techniques (fiberwise sub/supersolutions, unwrapping the torus but not Y, extracting harmonic limits for the fiberwise extrema) are well-chosen. The final Savin bootstrap in Corollary 5.1 is natural once the C^0-small quadratic approximation is obtained. The paper also clearly separates the model case from the geometric setting and does not pretend to prove more than it does. The input from [10] is used legitimately—the C^0 potential convergence is the starting point, not the conclusion.

Soft spots. The load-bearing point is Remark 2.1: the actual family is modeled on the normal bundle of Y up to an O(e^{-cT}) error in the complex Monge-Ampère equation, coming from the deviation of the holomorphic volume form. This exponential error is what makes the extra term in (3.31) and (4.52) negligible. If the error were merely polynomial, the truncated Harnack/Hölder estimates would not close at the required scales, and the blow-up limits need not be harmonic. The paper asserts this estimate and says it will comment briefly; Remarks 3.3 and 4.2 are sketches, not derivations. In a paper of this technical depth, that is a real gap, though likely fixable: near Y, after writing F_0...F_m = -tF and rescaling logs, the deviation should be O(|t|) = O(e^{-T}). A referee should require a precise statement and proof of this exponential decay, and a verification that the sub/supersolution comparison in Lemma 3.1 survives the O(e^{-cT}) perturbation of u.

The citation pattern is clean and the result is not circular. The model-case PDE work is detailed enough to be checked. This is a paper for specialists in collapsing Calabi-Yau metrics and the SYZ program; it deserves a serious referee. My recommendation: send it to review, with instructions to focus the referee on Remark 2.1 and the model-to-geometry transfer. If that point gets quantified, the theorem will stand.

Referee Report

2 major / 3 minor

Summary. The paper studies polarized degenerations of Calabi-Yau hypersurfaces X_t defined by tF + F_0...F_m = 0 with 0 < m < n, near an intermediate complex structure limit. Building on the C^0 potential convergence obtained in the first author's earlier work [10], it constructs ansatz metrics ω_t and proves that, on a 'generic region' U_t of almost full Calabi-Yau measure, the Ricci-flat metrics ω_CY,t converge in C^0 to ω_t as t→0. The proof is by adapting Savin's small-perturbation theorem to a model problem on a T^m-bundle over a Calabi-Yau (n−m)-fold: a truncated Harnack inequality (Thm 3.2), a De Giorgi-type blow-up argument (Thm 4.1), and a bootstrap to C^k estimates in stretched coordinates (Cor 5.1). The geometric application is then obtained by asserting that the actual family differs from the model by an exponentially small error in the complex Monge-Ampère equation. The main results are stated as Theorem 1.1 and Remark 1.2/Corollary 5.1.

Significance. If the proof is completed, this is a substantial and timely contribution: it extends the smooth convergence results known for large complex structure limits to the intermediate case 0 < m < n, where the model involves a collapsing T^m-bundle over a collapsing Calabi-Yau manifold. The model-case PDE analysis is detailed and internally coherent, with explicit scales and quantitative estimates; the paper does not assume the desired metric convergence, and the upgrade from the C^0 potential bound of [10] is genuine. The main uncertainty, acknowledged by the authors, is the transfer from the model to the geometric family, which is sketched rather than fully proved. If that gap is filled, the result would be a clear advance in the understanding of collapsing Ricci-flat metrics.

major comments (2)

[§2.5, Remark 2.1; also Remarks 3.3, 4.2, and §5] The transfer from the model problem to the geometric family rests on the assertion that the actual complex Monge-Ampère equation differs from the model equation (2.10) by an O(e^{-cT}) error in the volume form. This is load-bearing: it produces the extra term in (3.31), the version (4.52), and the conclusion that the blow-up limit w_∞ is harmonic. The paper explicitly says this is only sketched. If the error were merely polynomial in T, the e^{-cT} term in (4.52) would not be negligible under (4.41), the limiting w_∞ could fail to be harmonic, and Corollary 4.3 / Theorem 1.1 would not follow. Please state and prove a precise quantitative comparison of the normalized Calabi-Yau volume form on U_t with the model form of §2.5, including the normal-bundle identification, with constants independent of t and of the base point y.
[§4, proof of Theorem 4.1] The proof begins with 'we first use Cauchy-Kovalevskaya' to find a power series eQ_r with prescribed leading Taylor polynomial Q_r solving the Monge-Ampère equation, and then asserts eQ_r−Q_r = O(r^3). Cauchy-Kovalevskaya requires real-analyticity of the background potential u in the chosen chart; the paper only states that u is smooth on each Δ_k (§2.3). Please justify real-analyticity (for instance from (2.6) via standard elliptic regularity for uniformly elliptic equations with real-analytic coefficients) or replace this step by an approximation argument. This is needed before the Harnack inequality is applied to w.

minor comments (3)

[Title/header] The header contains a typo: 'INTERMEDIA TE' should be 'INTERMEDIATE'.
[§5, final paragraph] The passage from local estimates on eB_{T^{-1/2}}(y) to a global C^0 estimate on U_t is written informally ('up to slightly shrinking', 'for any y∈K_k'). To control the supremum over y, please spell out the finite cover of the compact sets K_k by such base balls and the uniformity of the constants, or state explicitly that the estimate is uniform in y by compactness of Y and the fixed r_0.
[Theorem 1.1] The generic region U_t depends on the choice of compact subsets K_k inside Δ_k, and the theorem allows U_t to be slightly shrunken. The statement should make explicit that the convergence holds for any fixed such choice, or that U_t can be chosen with complement of arbitrarily small normalized Calabi-Yau measure.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: heavy but legitimate dependence on [10]; the sketched O(e^{-cT}) transfer is a rigor gap, not a circular step.

full rationale

The paper's derivation chain is not circular. The target theorem (Theorem 1.1 / Corollary 5.1) controls the rescaled Hessian T i∂∂ψ_t in C^k on the generic region. This control is not assumed anywhere: the cited C^0 input from [10, Prop. 6.13] is only ∥ψ_t∥_{L∞} → 0 (equation (2.5)), which is strictly weaker than metric convergence. The ansatz metrics ω_t are constructed independently from the optimal-transport potential u on the skeleton, not fitted to the Calabi-Yau metrics. The model-case PDE (2.10) is solved exactly, and the truncated Harnack / De Giorgi / Savin arguments genuinely produce the higher-order estimates for its solution ψ. The only place where the geometric family is identified with the model is Remark 2.1's asserted O(e^{-cT}) closeness of the holomorphic volume form, used in Remarks 3.3 and 4.2. That exponential smallness is not derived in the paper, so the transfer is sketched rather than proved; but this is a rigor/completeness gap, not circularity, because the volume form is a fixed geometric datum rather than the target metric, and the estimates do not presuppose the desired convergence. The self-citations, notably [10], are prior independent results whose stated conclusions do not contain the target metric convergence, so they provide legitimate support rather than a circular premise.

Assumptions & free parameters 0 free parameters · 7 assumptions · 0 invented entities

No free parameters or invented entities: this is a pure mathematics proof. The central claim rests on standard background (Yau's theorem, Savin's theory) plus a chain of domain assumptions inherited from [10] (ansatz construction, C^0 potential convergence, generic region) and one asserted geometric reduction (O(e^{-cT}) closeness to the model case).

assumptions (7)

domain assumption The family X_t = {tF + ∏_{i=0}^m F_i = 0} ⊂ M with generic sections is a dlt degeneration with essential skeleton the m-simplex Δ (from [17, Prop 4.1], [10]).
Sets up the entire class of degenerations studied; taken from prior work and standard transversality.
standard math Yau's theorem provides Ricci-flat metrics ω_CY,t in c_1(L) with normalized volume form (2.4).
Classical background result invoked for existence of the collapsed metrics.
domain assumption The ansatz potential u solves the real Monge-Ampère equation (2.6) with constant c_0 and is smooth on each Δ_k (from [10, §5, Lemmas 4.7–4.8]).
The reference metric ω_t and its PDE are imported from Li [10]; the PDE is the structural backbone of the Harnack argument.
domain assumption C^0 potential convergence: ||ψ_t||_{L∞(X_t)} → 0 as t→0 (eq. (2.5), [10, Prop 6.13]).
The input being upgraded to metric convergence; it is a theorem of the first author in the to-appear paper [10].
standard math The operator F in (3.16) satisfies the structural hypotheses (H1)–(H3) of Savin [19].
Asserted in §3; required for the small-perturbation machinery to apply to the limiting real Monge-Ampère equation.
domain assumption The geometric setting deviates from the model case only by O(e^{-cT}) in the complex Monge-Ampère equation via identification of the normal bundle of Y (Remark 2.1).
Load-bearing for carrying model-case estimates to X_t; asserted and sketched, not fully derived in this paper.
domain assumption The generic region U_t covers an arbitrarily large fraction of the normalized Calabi-Yau measure (from [10, Cor 4.6]).
Justifies the 'generic region' conclusion and the final density argument.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Generic regularity of intermediate complex structure limits." pith.science (2026). https://pith.science/paper/AINF45LS

@misc{pith2026251104651,
  author       = {Pith},
  title        = {Pith review of: Generic regularity of intermediate complex structure limits},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AINF45LS}},
  note         = {Machine review of arXiv:2511.04651}
}

abstract

We study certain polarized degenerations of Calabi-Yau manifolds near an intermediate complex structure limit, and improve the potential $C^0$-convergence to a metric convergence result on the generic region for the corresponding collapsing Ricci-flat K\"ahler metrics.

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

What to do with a Ricci-flat Calabi--Yau metric?
hep-th 2026-05 conditional novelty 3.0 of 10

Numerical Ricci-flat Calabi–Yau metrics turn string compactifications from topological existence statements into computable geometries, unlocking normalized couplings, spectra, and metric-level tests of mirror symmetry.
What to do with a Ricci-flat Calabi--Yau metric?
hep-th 2026-05 unverdicted novelty 2.0 of 10

A roadmap paper describing potential applications of numerical Ricci-flat Calabi-Yau metrics to heterotic string phenomenology and mathematical questions in special geometry.

Reference graph

Works this paper leans on

26 extracted references · 4 linked inside Pith · cited by 1 Pith paper

[10]

Li,Intermediate complex structure limit for Calabi-Yau metrics, to appear in Invent

Y. Li,Intermediate complex structure limit for Calabi-Yau metrics, to appear in Invent. Math
[1]

Andreasson, J

R. Andreasson, J. Hultgren,Solvability of Monge-Amp` ere equations and tropical affine structures on reflexive polytopes, to appear in Amer. J. Math
[2]

Boucksom, M

S. Boucksom, M. Jonsson,Tropical and non-Archimedean limits of degenerating fam- ilies of volume forms, J. ´Ec. polytech. Math.4(2017), 87–139

2017
[3]

Collins, Y

T.C. Collins, Y. Li,Complete Calabi-Yau metrics in the complement of two divisors, Duke Math. J.173(2024), no. 18, 3559–3604

2024
[4]

Collins, F

T.C. Collins, F. Tong, S.-T. Yau,A free-boundary Monge-Amp` ere equation and ap- plications to complete Calabi-Yau metrics, preprint, arXiv:2402.10111

arXiv
[5]

Donaldson, S

S.K. Donaldson, S. Sun,Gromov-Hausdorff limits of K¨ ahler manifolds and algebraic geometry, Acta Math.213(2014), no. 1, 63–106

2014
[6]

Eyssidieux, V

P. Eyssidieux, V. Guedj, A. Zeriahi,Singular K¨ ahler-Einstein metrics, J. Amer. Math. Soc.22(2009), no. 3, 607–639

2009
[7]

H.-J. Hein, V. Tosatti,Smooth asymptotics for collapsing Calabi-Yau metrics, Comm. Pure Appl. Math.78(2025), no. 2, 382–499

2025

Show all 26 references

[8]

Hultgren, M

J. Hultgren, M. Jonsson, E. Mazzon, N. McCleerey,Tropical and non-Archimedean Monge–Amp` ere equations for a class of Calabi–Yau hypersurfaces, Adv. Math.439 (2024), Paper No. 109494

2024
[9]

Li,Strominger-Yau-Zaslow conjecture for Calabi-Yau hypersurfaces in the Fermat family, Acta Math.229(2022), no

Y. Li,Strominger-Yau-Zaslow conjecture for Calabi-Yau hypersurfaces in the Fermat family, Acta Math.229(2022), no. 1, 1–53. 20 YANG LI AND V ALENTINO TOSATTI

2022
[11]

Li,Metric SYZ conjecture and non-Archimedean geometry, Duke Math

Y. Li,Metric SYZ conjecture and non-Archimedean geometry, Duke Math. J.172 (2023), no. 17, 3227–3255

2023
[12]

Li,Metric SYZ conjecture for certain toric Fano hypersurfaces, Camb

Y. Li,Metric SYZ conjecture for certain toric Fano hypersurfaces, Camb. J. Math. 12(2024), no. 1, 223–252

2024
[13]

Li,Degeneration of Calabi-Yau metrics and canonical basis, preprint, arXiv:2505.11087

Y. Li,Degeneration of Calabi-Yau metrics and canonical basis, preprint, arXiv:2505.11087

arXiv
[14]

Y. Li, V. Tosatti,Diameter bounds for degenerating Calabi-Yau metrics, J. Differen- tial Geom.127(2024), no. 2, 603–614

2024
[15]

Mustat ¸˘ a, J

M. Mustat ¸˘ a, J. Nicaise,Weight functions on non-Archimedean analytic spaces and the Kontsevich-Soibelman skeleton, Algebr. Geom.2(2015), no. 3, 365–404

2015
[16]

Nicaise, C

J. Nicaise, C. Xu,The essential skeleton of a degeneration of algebraic varieties, Amer. J. Math.138(2016), no. 6, 1645–1667

2016
[17]

Pille-Schneider,Hybrid toric varieties and the non-Archimedean SYZ fibration on Calabi-Yau hypersurfaces, preprint, arXiv:2210.05578

L. Pille-Schneider,Hybrid toric varieties and the non-Archimedean SYZ fibration on Calabi-Yau hypersurfaces, preprint, arXiv:2210.05578

arXiv
[18]

X. Rong, Y. Zhang,Continuity of extremal transitions and flops for Calabi-Yau man- ifolds,Appendix B by Mark Gross, J. Differential Geom. 89 (2011), no. 2, 233–269

2011
[19]

Savin,Small perturbation solutions for elliptic equations, Comm

O. Savin,Small perturbation solutions for elliptic equations, Comm. Partial Differen- tial Equations32(2007), no. 4-6, 557–578

2007
[20]

Strominger, S.-T

A. Strominger, S.-T. Yau, E. Zaslow,Mirror symmetry isT-duality, Nuclear Phys. B 479(1996), no. 1-2, 243–259

1996
[21]

S. Sun, R. Zhang,Complex structure degenerations and collapsing of Calabi-Yau met- rics, preprint, arXiv:1906.03368

1906 arXiv
[22]

Takayama,On moderate degenerations of polarized Ricci-flat K¨ ahler manifolds, J

S. Takayama,On moderate degenerations of polarized Ricci-flat K¨ ahler manifolds, J. Math. Sci. Univ. Tokyo22(2015), no. 1, 469–489

2015
[23]

Tosatti,Families of Calabi-Yau manifolds and canonical singularities, Int

V. Tosatti,Families of Calabi-Yau manifolds and canonical singularities, Int. Math. Res. Not. IMRN 2015, no. 20, 10586–10594

2015
[24]

Tosatti,Collapsing Calabi-Yau manifolds, Surveys in Differential Geometry23 (2018), 305–337, International Press, 2020

V. Tosatti,Collapsing Calabi-Yau manifolds, Surveys in Differential Geometry23 (2018), 305–337, International Press, 2020

2018
[25]

Tosatti, Y

V. Tosatti, Y. Zhang,Infinite time singularities of the K¨ ahler-Ricci flow, Geom. Topol.19(2015), no. 5, 2925–2948

2015
[26]

Yau,On the Ricci curvature of a compact K¨ ahler manifold and the complex Monge-Amp` ere equation, I, Comm

S.-T. Yau,On the Ricci curvature of a compact K¨ ahler manifold and the complex Monge-Amp` ere equation, I, Comm. Pure Appl. Math.31(1978), 339–411. Department of Pure Mathematics and Mathematical Statistics, University of Cambridge, UK Email address:yl454@cam.ac.uk Courant In...

1978

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [10]

Li,Intermediate complex structure limit for Calabi-Yau metrics, to appear in Invent

Y. Li,Intermediate complex structure limit for Calabi-Yau metrics, to appear in Invent. Math

[2] [1]

Andreasson, J

R. Andreasson, J. Hultgren,Solvability of Monge-Amp` ere equations and tropical affine structures on reflexive polytopes, to appear in Amer. J. Math

[3] [2]

Boucksom, M

S. Boucksom, M. Jonsson,Tropical and non-Archimedean limits of degenerating fam- ilies of volume forms, J. ´Ec. polytech. Math.4(2017), 87–139

2017

[4] [3]

Collins, Y

T.C. Collins, Y. Li,Complete Calabi-Yau metrics in the complement of two divisors, Duke Math. J.173(2024), no. 18, 3559–3604

2024

[5] [4]

Collins, F

T.C. Collins, F. Tong, S.-T. Yau,A free-boundary Monge-Amp` ere equation and ap- plications to complete Calabi-Yau metrics, preprint, arXiv:2402.10111

arXiv

[6] [5]

Donaldson, S

S.K. Donaldson, S. Sun,Gromov-Hausdorff limits of K¨ ahler manifolds and algebraic geometry, Acta Math.213(2014), no. 1, 63–106

2014

[7] [6]

Eyssidieux, V

P. Eyssidieux, V. Guedj, A. Zeriahi,Singular K¨ ahler-Einstein metrics, J. Amer. Math. Soc.22(2009), no. 3, 607–639

2009

[8] [7]

H.-J. Hein, V. Tosatti,Smooth asymptotics for collapsing Calabi-Yau metrics, Comm. Pure Appl. Math.78(2025), no. 2, 382–499

2025