Gradient Shrinking Ricci Solitons

• Gradient shrinking Ricci solitons are complete Riemannian manifolds defined by Ric + ∇²f = ½g, serving as canonical self-similar solutions to the Ricci flow.
• They underpin analytic techniques like weighted integration by parts and drift-Laplacian estimates, which are crucial for deriving curvature bounds and asymptotic behavior.
• Four-dimensional and Kähler classifications reveal rigidity phenomena, leading to standard models such as Gaussian, spherical, and cylindrical shrinkers.

A gradient shrinking Ricci soliton is a complete Riemannian manifold $(M,g)$ equipped with a smooth potential $f$ such that

$$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$

and, after rescaling, most of the modern literature adopts the normalization

$\operatorname{Ric}+\nabla^2 f=\frac12 g.$

These metrics are self-similar solutions to the Ricci flow and serve as canonical models for finite-time singularity formation. Their study lies at the intersection of elliptic geometric analysis, singularity theory for Ricci flow, and global rigidity/classification problems, with especially sharp results in dimension four and in Kähler settings (Li et al., 2016, Munteanu et al., 2010).

1. Defining equations and weighted analytic structure

Under the normalization

$\operatorname{Ric}+\nabla^2 f=\frac12 g,$

the standard shrinker identities include

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$

and in dimension four these become

$R+\Delta f=2,\qquad R+|\nabla f|^2=f.$

The associated drift Laplacian is

$\Delta_f:=\Delta-\nabla f\cdot \nabla,$

which is self-adjoint with respect to the weighted measure $e^{-f}\,dV$. Standard differential identities on a shrinker include

$$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$

and

$f$0

These formulas explain why weighted integration by parts and drift-elliptic estimates are structurally natural in the subject (Li et al., 2016, Munteanu et al., 2010).

For complete noncompact shrinkers, the potential has quadratic growth. A standard estimate due to Cao–Zhou gives

$f$1

together with Euclidean-type volume growth bounds such as

$f$2

Weighted curvature integrability is also available in great generality; for instance, one has

$f$3

for every $f$4 in the setting used by Munteanu–Şeșum and by later four-dimensional classification arguments. These estimates are the analytic backbone behind global maximum principles, weighted $f$5-arguments, and compactness theorems (Li et al., 2016, Colding et al., 2023).

2. Standard models and the rigidity paradigm

The basic examples are the Gaussian shrinker on $f$6, the round sphere, and cylindrical/product shrinkers such as $f$7 and $f$8 in four dimensions. In dimension four, the round $f$9 is the compact positively curved Einstein model, while $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$0 with the shrinking round metric on $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$1 and flat metric on $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$2 is the standard noncompact cylindrical model (Li et al., 2016).

A recurring conclusion in the literature is that a shrinker is rigid in the sense of splitting, up to finite quotient, as

$$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$3

where the Euclidean factor carries the Gaussian structure and $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$4 is Einstein. In one formulation used in later rigidity work, a gradient soliton is rigid iff it has constant scalar curvature and is radially flat. This framework underlies many classification theorems obtained from curvature conditions weaker than full local symmetry (Yang et al., 2017).

Several sharp model lists recur under different hypotheses. For four-dimensional shrinkers with half harmonic Weyl curvature, the possibilities are Einstein, or finite quotients of

$$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$5

For Bach-flat shrinkers, the four-dimensional conclusion is that the metric is either Einstein or locally conformally flat, hence a finite quotient of $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$6 or $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$7, while in dimensions $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$8 the non-Einstein cases are finite quotients of $$\operatorname{Ric}+\nabla^2 f=\lambda g,\qquad \lambda>0,$$9 or $\operatorname{Ric}+\nabla^2 f=\frac12 g.$0 with $\operatorname{Ric}+\nabla^2 f=\frac12 g.$1 Einstein of positive scalar curvature (Wu et al., 2014, Cao et al., 2011).

3. Four-dimensional curvature decomposition and classification theorems

Dimension four is exceptional because the bundle of $\operatorname{Ric}+\nabla^2 f=\frac12 g.$2-forms splits as

$\operatorname{Ric}+\nabla^2 f=\frac12 g.$3

so the curvature operator admits the block form

$\operatorname{Ric}+\nabla^2 f=\frac12 g.$4

If $\operatorname{Ric}+\nabla^2 f=\frac12 g.$5 and $\operatorname{Ric}+\nabla^2 f=\frac12 g.$6 are the eigenvalues of $\operatorname{Ric}+\nabla^2 f=\frac12 g.$7 and $\operatorname{Ric}+\nabla^2 f=\frac12 g.$8, then positive isotropic curvature is equivalent to

$\operatorname{Ric}+\nabla^2 f=\frac12 g.$9

Writing

$\operatorname{Ric}+\nabla^2 f=\frac12 g,$0

one obtains an efficient analytic framework for four-dimensional PIC shrinkers. A sharp theorem states that any four-dimensional complete gradient shrinking Ricci soliton with positive isotropic curvature is either a quotient of $\operatorname{Ric}+\nabla^2 f=\frac12 g,$1 or a quotient of $\operatorname{Ric}+\nabla^2 f=\frac12 g,$2. With nonnegative isotropic curvature, the universal cover must be one of

$\operatorname{Ric}+\nabla^2 f=\frac12 g,$3

The novelty of this result is that it removes the extra hypotheses previously used by Wallach–Ni, including nonnegative curvature operator and pointwise curvature growth assumptions (Li et al., 2016).

Other four-dimensional classifications replace isotropic-curvature assumptions by half-Weyl or divergence conditions. If $\operatorname{Ric}+\nabla^2 f=\frac12 g,$4, then a four-dimensional gradient shrinking Ricci soliton is either Einstein or a finite quotient of

$\operatorname{Ric}+\nabla^2 f=\frac12 g,$5

The proof combines the four-dimensional decomposition of $\operatorname{Ric}+\nabla^2 f=\frac12 g,$6, an explicit algebraic description of $\operatorname{Ric}+\nabla^2 f=\frac12 g,$7 in terms of trace-free Ricci eigenvalues, a half-Weyl Weitzenböck formula, and a weighted maximum principle applied to $\operatorname{Ric}+\nabla^2 f=\frac12 g,$8 (Wu et al., 2014).

A parallel rigidity mechanism arises from higher-order divergence conditions. If

$\operatorname{Ric}+\nabla^2 f=\frac12 g,$9

or

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$0

then a complete gradient shrinking Ricci soliton is rigid; in dimension four this again yields the model list “Einstein, or a finite quotient of $R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$1, $R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$2, or $R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$3.” The same four-dimensional conclusion holds under the Weyl analogues

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$4

These results show that surprisingly high-order scalar divergence conditions can collapse the geometry to the standard rigid models (Yang et al., 2017).

Constant scalar curvature yields another strong four-dimensional rigidity channel. If a complete noncompact four-dimensional shrinker satisfies

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$5

then it is isometric to a finite quotient of

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$6

Two recent proofs emphasize different mechanisms: one uses asymptotic geometry and the classification of three-dimensional ancient $R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$7-solutions, while another uses a maximum principle for the sum of the two smallest Ricci eigenvalues. In both approaches, the key output is the rank-two Ricci pattern

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$8

which forces global splitting (Wang et al., 28 Apr 2026, Ou et al., 2024).

4. Curvature estimates, asymptotics, and compactness theory

A central analytic theme is that mild control on Ricci curvature propagates to stronger control on the full curvature tensor. If a complete gradient shrinking Ricci soliton has bounded Ricci curvature, then the full Riemann tensor satisfies a polynomial growth bound

$R+\Delta f=\frac n2,\qquad R+|\nabla f|^2=f,$9

for some $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$0. The proof combines weighted $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$1-estimates for $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$2, the drift-Laplacian inequalities for curvature, and local Moser iteration. This leads to a gap theorem: if

$R+\Delta f=2,\qquad R+|\nabla f|^2=f.$3

everywhere, then the shrinker is the Gaussian soliton $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$4 (Munteanu et al., 2010).

In four dimensions, bounded scalar curvature has particularly strong consequences. If a four-dimensional shrinker has bounded scalar curvature, then

$R+\Delta f=2,\qquad R+|\nabla f|^2=f.$5

the curvature operator is asymptotically nonnegative with lower bound

$R+\Delta f=2,\qquad R+|\nabla f|^2=f.$6

and, if $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$7 at infinity, the manifold is $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$8-asymptotic to a cone for all $R+\Delta f=2,\qquad R+|\nabla f|^2=f.$9. These results depend on exploiting the geometry of level sets of the potential $\Delta_f:=\Delta-\nabla f\cdot \nabla,$0, whose dimension is three, so the intrinsic curvature of the level set is determined by its intrinsic Ricci tensor (Munteanu et al., 2014).

Compactness theory for families of four-dimensional shrinkers also uses the weighted structure in an essential way. Under strong $\Delta_f:=\Delta-\nabla f\cdot \nabla,$1-noncollapsing and an embedding hypothesis into a closed four-manifold with $\Delta_f:=\Delta-\nabla f\cdot \nabla,$2, a sequence of noncompact four-dimensional normalized shrinkers has a smoothly convergent subsequence whose limit is again a smooth nonflat shrinker. The key point is that one obtains this without any a priori curvature bound: curvature concentration is excluded by a blow-up analysis producing Ricci-flat ALE limits, which are then ruled out by four-dimensional topology (Zhang, 2017).

5. Kähler shrinkers and related geometric structures

Within the Kähler category, the shrinker equation becomes

$\Delta_f:=\Delta-\nabla f\cdot \nabla,$3

in the normalization used for shrinking Kähler-Ricci solitons. In complex dimension two, a complete shrinking gradient Kähler-Ricci soliton with scalar curvature tending to zero at infinity is, up to pullback by an element of $\Delta_f:=\Delta-\nabla f\cdot \nabla,$4, either the flat Gaussian shrinker on $\Delta_f:=\Delta-\nabla f\cdot \nabla,$5 or the $\Delta_f:=\Delta-\nabla f\cdot \nabla,$6-invariant Feldman–Ilmanen–Knopf shrinker on the blowup of $\Delta_f:=\Delta-\nabla f\cdot \nabla,$7 at one point. More generally, with bounded Ricci curvature, the only complete shrinking gradient Kähler-Ricci soliton on $\Delta_f:=\Delta-\nabla f\cdot \nabla,$8 is the Gaussian shrinker, and on $\Delta_f:=\Delta-\nabla f\cdot \nabla,$9 for $e^{-f}\,dV$0 it is the unique $e^{-f}\,dV$1-invariant Feldman–Ilmanen–Knopf example (Conlon et al., 2019).

The classification in complex dimension two was later completed, under bounded scalar curvature, by the construction of a new noncompact shrinking gradient Kähler-Ricci soliton on

$e^{-f}\,dV$2

the blowup of $e^{-f}\,dV$3 at a torus fixed point. This produces the final missing case in the classification of complete two-dimensional shrinking gradient Kähler-Ricci solitons with bounded scalar curvature: the possibilities are the compact Fano examples, the flat Gaussian on $e^{-f}\,dV$4, the Feldman–Ilmanen–Knopf shrinker on $e^{-f}\,dV$5, the cylinder $e^{-f}\,dV$6, and the new blowup example on $e^{-f}\,dV$7 (Bamler et al., 2022).

Several rigidity mechanisms tie Kähler and conformal geometry to shrinkers in dimension four. Under a sharp pinching involving $e^{-f}\,dV$8 and the scalar curvature,

$e^{-f}\,dV$9

a complete four-dimensional shrinker is locally a Kähler-Ricci soliton. Under lower bounds on the modified sectional curvature

$$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$0

the compact four-dimensional theory yields classification results by $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$1 and $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$2, weighted gap estimates for $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$3, and a Hitchin–Thorpe type inequality for compact shrinkers (Cao et al., 25 Sep 2025).

The generalized Ricci-flow setting provides a contrasting rigidity statement. On a compact manifold, every gradient shrinking generalized soliton is actually an ordinary Ricci soliton: the extra closed $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$4-form $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$5 must vanish identically. Likewise, a compact gradient shrinking pluriclosed soliton is Kähler, and under a broad cohomological assumption a compact pluriclosed soliton becomes a gradient Kähler-Ricci soliton. This shows that, at least in the compact shrinking gradient case, the generalized framework does not enlarge the classical Ricci-soliton class (Li et al., 2024).

6. Recent directions: symmetry propagation, higher-dimensional rigidity, and Sasaki analogues

One recent analytic development is the “shrinker principle” for approximate symmetries. On a noncompact shrinker with bounded curvature, if a vector field is approximately Killing on one sufficiently large scale, then one can construct a global eigenfield $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$6 of

$$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$7

with small eigenvalue, and the defect $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$8 remains quantitatively controlled on larger level sets of $$\nabla_i R=2R_{ij}f_j,\qquad \nabla_kR_{jk}=R_{jk}f_k,\qquad \nabla_iR_{ijkl}=R_{ijkl}f_i,$$9. This provides a one-scale-to-larger-scale rigidity mechanism designed for asymptotically cylindrical or approximately symmetric shrinkers (Colding et al., 2023).

Constant-scalar-curvature rigidity has also begun to extend beyond dimension four. For a complete noncompact shrinker with

$f$00

if each level set of $f$01 has vanishing Weyl curvature, then the manifold is isometric to a finite quotient of

$f$02

A related theorem shows that if

$f$03

the Ricci curvature is nonnegative, and sectional curvature is bounded above by

$f$04

then the shrinker is a finite quotient of $f$05. This can be viewed as a higher-dimensional extension, under extra hypotheses, of the four-dimensional $f$06 rigidity phenomenon (Wang et al., 27 Apr 2026).

The Sasaki setting supplies an odd-dimensional analogue of several classical Ricci-shrinker results. For complete gradient shrinking Sasaki-Ricci solitons, one-end theorems, compactness under positive sectional curvature, and sphere-quotient rigidity under nonnegative curvature operator all have been established. These results parallel the Riemannian and Kähler shrinking theories of Perelman, Naber, and Munteanu–Wang, but in the transversely Kähler geometry determined by the Reeb foliation (Chang et al., 19 Aug 2025).

An important open direction remains the search for high-dimensional analogues of the cleanest four-dimensional classification theorems. In particular, the four-dimensional PIC theorem strongly suggests that isotropic curvature is the correct rigidity condition in that dimension, and the question whether an analogous classification holds in higher dimensions remains explicitly open (Li et al., 2016).