• Nie Znaleziono Wyników

Geodesics in the Spaces of K¨ ahler Metrics and Volume Forms

N/A
N/A
Protected

Academic year: 2021

Share "Geodesics in the Spaces of K¨ ahler Metrics and Volume Forms"

Copied!
12
0
0

Pełen tekst

(1)

Geodesics in the Spaces of K¨ ahler Metrics and Volume Forms

Zbigniew B locki

Uniwersytet Jagiello´ nski, Krak´ ow, Poland http://gamma.im.uj.edu.pl/ eblocki

Complex Geometry and Cauchy-Riemann Equation August 23, 2016

Center for Advanced Study, Oslo

(2)

(M, ω) compact K¨ ahler manifold We can write {ω} ' H/ ∼, where

H = {ϕ ∈ C (M) : ω ϕ := ω + dd c ϕ > 0}

is the space of K¨ ahler potentials, and

ϕ 1 ∼ ϕ 2 ⇔ ϕ 1 − ϕ 2 = const.

Riemannian structure on H (Mabuchi, 1987 / Donaldson, 1999) hhψ, ηii := 1

V Z

M

ψ η ω ϕ n , ψ, η ∈ T ϕ H ' C (M), where V = R

M ω n .

Levi-Civita connection: if ϕ ∈ C ([0, 1], H) ⊂ C (M × [0, 1]) and ψ is a vector field along ϕ (i.e. ψ ∈ C (M × [0, 1]), then

ϕ ˙ ψ = ˙ ψ − h∇ψ, ∇ ˙ ϕi, so that d

dt hhψ, ηii = hh∇ ϕ ˙ ψ, ηii + hhψ, ∇ ϕ ˙ ηii.

(3)

Normalization Aubin-Yau functional I : H → R is uniquely defined by

I (0) = 0, d dt

t=0

I (ϕ + tψ) = 1 V

Z

M

ψω ϕ n , ϕ ∈ H, ψ ∈ C (M).

One can show that I (ϕ) = 1

n + 1

n

X

p=0

1 V

Z

M

ϕ ω ϕ p ∧ ω n−p .

Then H 0 = I −1 (0) ' {ω} defines a natural Riemannian structure

on {ω} which is independent of the choice of ω.

(4)

Geodesics A curve ϕ : [0, 1] → H is a geodesic if ∇ ϕ ˙ ϕ = 0, that is ˙

¨

ϕ − |∇ ˙ ϕ| 2 = 0.

Locally write u = g + ϕ, where ω = dd c g . Then it is equivalent to u tt − u i ¯ j u ti u t ¯ j = 0,

which is equivalent to

det

u 1t (u j ¯ k ) .. .

u nt

u 1 . . . u t ¯ n u tt

= 0.

This means that

(ω + dd c ϕ) n+1 = 0,

where t = log |z n+1 | (Semmes, 1992 / Donaldson, 1999).

(5)

To find a geodesic connecting ϕ 0 , ϕ 1 ∈ H one has to solve HCMA (ω + dd c ϕ) n+1 = 0

in M × {0 ≤ log |z n+1 | ≤ 1} with boundary condition.

Donaldson Conjecture, 1999: Every ϕ 0 , ϕ 1 ∈ H can be joined by a smooth geodesic.

Consequence: uniqueness of constant scalar curvature (csc) metrics up to holomorphic automorphisms

X.X. Chen, 2000: There exists unique, weak (ω + dd c ϕ ≥ 0), almost C 1,1 (∆ϕ ∈ L ) geodesic.

Lempert-Vivas, 2013: A geodesic need not be C 3 . Darvas-Lempert, 2012: A geodesic need not be C 2 . Remaining question: Are geodesics fully C 1,1 ?

B., 2012: If bisec(M) ≥ 0 then geodesics are C 1,1 .

(6)

Theorem Assume that (M, ω) is a compact K¨ ahler manifold with boundary (possibly empty). Let ϕ ∈ C 4 (M) be such that ω ϕ > 0 and ω ϕ n = f ω n . Then

|∇ 2 ϕ| ≤ C ,

where C depends only on upper bounds for n, |R|, |∇R|, |ϕ|,

|∇ϕ|, ∆ϕ, sup ∂M |∇ 2 ϕ|, ||f 1/n || C

1,1

(M) , |∇(f 1/2n )| and a lower positive bound for f . If M has nonnegative bisectional curvature then the estimate is independent of the latter.

Sketch of proof α := |∇ 2 ϕ| + |∇ϕ| 2 − Aϕ, where

|∇ 2 ϕ| = max

X 6=0

h∇ X ∇ϕ, X i

|X | 2

and A  0. α attains max for some x 0 ∈ M and X ∈ T x

0

M.

α = e h∇ X ∇ϕ, X i

|X | 2 + |∇ϕ| 2 − Aϕ also attains max at x 0 but is smooth! Then

2

∂z p ∂ ¯ z p

h∇ X ∇ϕ, X i g j ¯ k X j X ¯ k

!

= · · · + X j X ¯ k R j ¯ kp ¯ p D X 2 ϕ.

(7)

Weak Solutions to CMA

Ko lodziej, 1998 Let (M, ω) be the compact K¨ ahler manifold. If f ∈ L p (M) for some p > 1 is such that f ≥ 0 and R

M f ω n = R

M ω n then there exists unique (up to an additive constant) ϕ ∈ C (M) such that ω ϕ ≥ 0 and

ω ϕ n = f ω n . Yau, 1978: f > 0, f ∈ C ⇒ ϕ ∈ C

B., 2002: f ≥ 0, f 1/(n−1) ∈ C 1,1 ⇒ ∆ϕ ∈ L

B., 2009: f ≥ 0, f 1/n ∈ C 0,1 ⇒ ϕ ∈ C 0,1

bisec(M) ≥ 0, f ≥ 0, f 1/n ∈ C 1,1 ⇒ ϕ ∈ C 1,1

(8)

Space of volume forms (Donaldson, 2010) (M, g ) compact Riemannian manifold

dV 0 = pdet(g ij ) Riemannian volume form on M , V 0 = R

M dV 0 V := {dV volume form on M with

Z

M

dV = V 0 } Then every element of V can be written in the form dV = (∆ϕ + 1)dV 0 , and V = H/ ∼, where

H = {ϕ ∈ C (M) : ∆ϕ + 1 > 0.}

and ϕ 1 ∼ ϕ 2 ⇔ ϕ 1 − ϕ 2 = const.

Riemannian structure on H hhψ, ηii = 1

V 0 Z

M

ψ η (1+∆ϕ)dV 0 , ϕ ∈ H, ψ, η ∈ T ϕ H ' C (M).

Levi-Civita connection: if ϕ ∈ C ([0, 1], H) ⊂ C (M × [0, 1]) and ψ is a vector field along ϕ (i.e. ψ ∈ C (M × [0, 1]), then

ϕ ˙ ψ = ˙ ψ − h∇ψ, ∇ ˙ ϕi

∆ϕ + 1 .

(9)

Geodesics ϕ : [0, 1] → H is a geodesic if (∆ϕ + 1)ϕ tt − |∇ϕ t | 2 = 0.

Chen-He, 2011: Given ϕ 0 , ϕ 1 ∈ H, there exists unique, weak, almost C 1,1 geodesic connecting them.

By Darvas-Lempert we cannot expect better regularity than C 2 . B.-Gu If M has nonnegative sectional curvature then geodesics are C 1,1 .

Sketch of proof Define

α = |∇ 2 ϕ| + |∇ϕ| 2 + A(−ϕ + t 2 /2).

Then α attains max for some (x 0 , t 0 ) ∈ M × (0, 1) and X ∈ T x

0

M.

We may assume X = e 1 , then

α = ∇ e 11 ϕ + |∇ϕ| 2 + A(−ϕ + t 2 /2).

One can show that

11ii ϕ−∇ ii ∇ 11 ϕ = −2R 11ii (∇ 11 ϕ−∇ ii ϕ)−∇ i R 1i 1 m ϕ m −∇ 1 R 1ii m ϕ m ≤ C .

(10)

Relation to Nahm’s equations T 1 , T 2 , T 3 : (0, 2) → U(n) dT 1

dt = [T 2 , T 3 ], dT 2

dt = [T 3 , T 1 ], dT 3

dt = [T 1 , T 2 ].

Fixing B ∈ GL(n, C), Donaldson (1984) showed that they are equivalent to a 2nd order ODE for h(t) valued in the space of positive Hermitian matrices H ' GL(n, C)/U(n) which is Euler-Lagrange for the Lagrangian

E (h) =

Z

dh dt

2 H

+ V B (h)

! dt,

where V B (h) = Tr (hBh −1 B ). He proved that given h 0 , h 1 ∈ H one can find unique h(t) joining them. (h(t) is a path of a particle moving under the influence of a potential −V B .)

If M is a Riemann surface then the space of K¨ ahler potentials H

behaves similarly as G c /G, where G is the group of area preserving

diffeomorphisms (although G c does not really exist!).

(11)

Recent developments (Sz´ ekelyhidi, Tossatti-Weinkove, Chu-Tossatti-Weinkove, Sz´ ekelyhidi-Tossatti-Weinkove, . . . ) Various C 2 -estimates

Lemma Let ϕ be a C 4 function defined near x 0 ∈ R n . Assume that D 2 ϕ is diagonal at x 0 and ϕ 11 > ϕ ii , i > 1, there. Near x 0 define λ := λ max (D 2 ϕ). Then at x 0 we have λ = ϕ 11 , λ p = ϕ 11p and

λ pp = ϕ 11pp + 2 X

i >1

ϕ 2 1ip

ϕ 11 − ϕ ii .

(12)

Thank you!

Cytaty

Powiązane dokumenty

We present a stability theorem of Ulam–Hyers type for K-convex set-valued functions, and prove that a set-valued function is K-convex if and only if it is K-midconvex

It is rather unusual in the theory of nonlinear elliptic equations of second order that the second derivative estimate can be obtained directly from the uniform esti- mate,

This paper completes the results of paper [11] in which regu- larity of solutions to equations with elliptic corner operators (i.e. n-dimensional Fuchsian operators) is studied in

Indeed, a rather useful theorem of Kechris, Louveau and Woodin [6] states that a coanalytic σ-ideal of compact sets is either a true coanalytic set or a G δ set.. For each

This corollary implies a theorem on existence of local analytic solutions of nonlinear systems of partial differential equations, of the type of the Cartan–K¨ ahler theorem, as

In this paper we consider para-K¨ ahler manifolds which satisfy curvature conditions of pseudosymmetric type. In Section 2 we give precise defini- tions. [14]) spaces,

THEOREM 2.1 Let K be a compact metric space. We should notice that Theorem 2.1 may be deduced from Proposition 3.6 of [2]. The Alspach result, in fact, does not concern operators,

By means of a connected sum on the pair: (X, the Z m -manifold), along two points of ψ −1 (0), we can change the manifold so that the monodromy along a connected component of ψ −1