DIAMETER BOUNDS FOR DEGENERATING CALABI-YAU

METRICS
arXiv:2006.13068v2 [math.DG] 1 Apr 2022

YANG LI AND VALENTINO TOSATTI

Abstract. We obtain sharp upper and lower bounds for the diameter of Ricci-
flat Kähler metrics on polarized Calabi-Yau degeneration families, as conjec-
tured by Kontsevich-Soibelman.

1. Introduction
The main objects of study in this note are Ricci-flat Kähler metrics on Calabi-Yau
manifolds whose complex structure degenerates. More precisely, we assume that
we have π : X → ∆∗ ⊂ C a projective holomorphic submersion with connected
fibers of relative dimension n with KX/∆∗ ∼ = OX and which is meromorphic at 0
(in the sense of [4]), meaning that it extends to a proper flat map π : X → ∆ with X
normal. We also fix a relative polarization L → X, and we will refer to this data as
a polarized Calabi-Yau degeneration family, often without mentioning L explicitly.
The fibers Xt for t ∈ ∆∗ are thus polarized Calabi-Yau n-folds.
A choice of X as above will be called a model of X. Models are highly non-
unique, and in particular up to passing to a finite base change, we may P assume by
[12] that X admits a semistable model, where X is smooth and X0 = i∈I Ei is
reduced and has simple normal crossings. One can then apply a relative MMP to a
semistable model and obtain a relatively minimal dlt model [2, 10, 17, 19], which is
unique up to applying sequences of flops on the central fiber [3, 11]. Taking the dual
intersection complex of the central fiber of any such minimal dlt model, one obtains
a simplicial complex Sk(X), the essential skeleton of X, which was introduced in
this context by Kontsevich-Soibelman [13] with a different but equivalent definition
(cf. [18]), whose homeomorphism type is well-defined independent of any choice of
models [19].
In this note we will not make direct use of the skeleton Sk(X) itself, but only of
its real dimension which will be denoted by m, and which appears naturally [4, 13]
as the power of logarithmic blowup of the fiberwise integrals of the Calabi-Yau
volume forms, as we will recall in Section 2 below. As shown in [19] we always
have m 6 n, and the case when m = 0 happens if and only if (after possibly a
finite base change) X admits a semistable model with central fiber X0 which is a
Calabi-Yau variety with klt singularities. Furthermore, the case m = n is equivalent
to the monodromy transformation around 0 acting on H n (Xt , C) having a Jordan
block of size n + 1. This is the familiar notion of a “large complex structure limit”
from mirror symmetry, see e.g. [8], where these polarized Calabi-Yau degeneration
families play a crucial role.

Date: April 4, 2022.

1
2 YANG LI AND VALENTINO TOSATTI

Our main interest is in the behavior as t → 0 of the Ricci-flat Kähler metrics ωt

on Xt in the scaled class | log1 |t|| c1 (L)|Xt , whose existence is guaranteed by Yau’s
Theorem [28]. In [13, Conjecture 1] Kontsevich-Soibelman conjectured that if X →
∆∗ is a large complex structure limit of Calabi-Yau manifolds, then the diameter
of (Xt , ωt ) is bounded away from zero and infinity (note that there is a typo in the
statement of their conjecture), and furthermore they, and independently also Gross-
Wilson [9] and Todorov, conjectured that the collapsed Gromov-Hausdorff limit of
(Xt , ωt ) is a half-dimensional affine manifold with singularities in codimension 2,
which is homeomorphic to Sk(X), and which is expected to be the base of the
Strominger-Yau-Zaslow fibration of Xt [22], see e.g. [1, §7] and [26] for surveys of
these and related topics.
The main theorem of this note is to prove the conjectured sharp diameter bound
in [13, Conjecture 1], for all polarized Calabi-Yau degeneration families with m > 0,
thus also settling [26, Conjecture 4.7]:
Theorem 1.1. Let π : X → ∆∗ be a polarized Calabi-Yau degeneration family,
suppose that the dimension m of the essential skeleton Sk(X) is positive, and let ωt
be the Ricci-flat Kähler metric on Xt in the class | log1 |t|| c1 (L)|Xt , for t ∈ ∆∗ . Then
there is C > 0 such that
C −1 6 diam(Xt , ωt ) 6 C,
for all t ∈ ∆∗ with |t| sufficiently small.
The assumption that m > 0 is necessary, since when m = 0 we can find a
semistable model X with central fiber X0 a Calabi-Yau variety with klt singularities,
as mentioned above, and then it is known by work of Rong-Zhang [20] that we have
1
instead diam(Xt , ωt ) ∼ | log |t||− 2 .
Despite several recent works addressing diameter bounds for Kähler-Einstein
metrics under various assumptions, see e.g. [7, 14, 21], the only previously known
general results in the direction of our main theorem are the following. First, tracing
through the arguments given in [20, Theorem 2.1] (see also [23, Proposition 4.2])
gives the upper bound
diam(Xt , | log |t||ωt ) 6 C| log |t||m ,
which is worse than the one provided by Theorem 1.1. And secondly, it fol-
lows from the earlier works [27, 25, 23] that when m > 0 we necessarily have
diam(Xt , | log |t||ωt ) → ∞ (see also the exposition in [29]), but the arguments there
do not provide any explicit lower bound.
The rough idea of our proof is the following: as we will recall in Section 2, a well-
known computation in polar coordinates (cf. [4]) reveals that most of the mass of
the Calabi-Yau volume forms ωtn on Xt is carried by “very small” regions which are
near certain intersections of m + 1 irreducible components of the central fiber. In
Section 3 we then construct a Kähler metric ωt′ cohomologous to ωt which behaves
like a toric metric in this good region (in the directions z1 , . . . , zm normal to these
components). Using ωt′ we then obtain a uniform L1 bound for |dρt |ωt where ρt
looks like a paraboloid in the logarithmic coordinates x1 , . . . , xm in our good region,
and using Cheeger-Colding’s segment inequality [5] we deduce the diameter lower
bound. Lastly, in Section 4 we again use ωt′ to produce a unit-size ωt -geodesic ball
in Xt whose volume is a definite fraction of the total, from which the diameter
upper bound follows from an argument of Yau, as in [24, 20].
 DIAMETER BOUNDS FOR DEGENERATING CALABI-YAU METRICS 3

Acknowledgments. The first-named author is a 2020 Clay Research Fellow, and

was based at the Institute for Advanced Study, supported by the Zurich Insurance
Company Membership, when this article was written. He thanks Song Sun for
earlier discussions. The second-named author would like to thank S.Takayama and
Y.Zhang for earlier discussions on these topics, and N. McCleerey and T.-D. Tô for
email exchanges. He was partially supported by NSF grant DMS-1903147, and this
article was written during his visit at the Department of Mathematics and at the
Center for Mathematical Sciences and Applications at Harvard University, which
he would like to thank for the hospitality. We are also grateful to the referee for
useful suggestions.

2. Volume form asymptotics

In this section we recall some background on the asymptotic behavior of the
relative Calabi-Yau volume forms on a polarized Calabi-Yau degeneration family,
and set up some notation for later use.
As in the introduction, we assume we have a polarized Calabi-Yau family π :
X → ∆∗ with relative polarization L, and we fix a trivializing section Ω of KX
and define trivializations Ωt of KXt by Ω = dt ∧ Ωt along Xt . Up to passing to a
finite base change,Pwe may assume that X admits a semistable model, where X is
smooth and X0 = i∈I Ei is reduced and has simple normal crossings. In this case
we have
X
KX/∆ = ai Ei , ai ∈ Z,
i∈I

and letting κ = mini∈I ai , up to replacing Ωt by t−κ Ωt , we may assume without

loss that κ = 0.
Recall that ωt denotes the Calabi-Yau metric on Xt in the class | log1 |t|| c1 (L)|Xt ,
and that the dimension m of the essential skeleton of X is assumed to be stricty
positive (and necessarily m 6 n). Denote by µt the Calabi-Yau volume form on Xt
normalized to be a probability measure, i.e.
2
ωtn i n Ωt ∧ Ωt
(2.1) µt = R n = .
in2 Ωt ∧ Ωt
R
Xt ωt X t

2
Let us now recall the asymptotic behavior of the integrals Xt in Ωt ∧ Ωt , largely
R
T
following [4]. For any J ⊂ I we denote by EJ = j∈J Ej . As in [15, 16], we fix a
Kähler metric on X and for ε > 0 small and J ⊂ I with EJ 6= ∅ we define

EJ0 = {x ∈ Xt | d(x, EJ ) < ε}\{x ∈ Xt | d(x, EJ ′ ) < ε for some J ′ ) J}.

For any given x ∈ EJ ⊂ X let p = |J| − 1 and pick local coordinates z0 , . . . , zn on

x ∈ V ⊂ X, defined in the unit polydisc, such that z0 , . . . , zp are defining equations
for Ej , j ∈ J, so that in these coordinates we have t = z0 · · · zp . We shall call these
adapted coordinates. We can then write
p
Y n
Y
Ω = fJ ziai dzi ∧ dzj ,
i=0 j=p+1
4 YANG LI AND VALENTINO TOSATTI

where fJ is a local non-vanishing holomorphic function. Since dt ∧ Ωt = Ω along

Xt , on EJ0 we get
p n
Y dzj Y
Ωt = fJ z0a0 · · · zpap ∧ dzk ,
z
j=1 j k=p+1
p n
2 Y dzj dzj Y
in Ωt ∧ Ωt = |fJ |2 |z0 |2a0 · · · |zp |2ap i ∧ ∧ idzk ∧ dzk ,
j=1
zj zj
k=p+1

from which using polar coordinates zj = exp(xj log |t| + iθj ), j ∈ J, one can easily
see as in [4] that Z
2
in Ωt ∧ Ωt ∼ | log |t||mJ ,
EJ0
where
mJ = |{j ∈ J | aj = 0}| − 1,
while Z
2
in Ωt ∧ Ωt ∼ | log |t||m ,
Xt
where
(2.2) m = max{|J| − 1 | EJ 6= ∅, aj = 0 for all j ∈ J} = dim Sk(X).
log |z |
The local logarithmic variables xj = log |t|j vary in the standard simplex
 
 X p 
∆J = 0 6 xj 6 1 | xj = 1 ,
 
j=0
S
and in this way one obtains a map Logt : V \ j Ej → ∆J , see [4].
For each i ∈ I we fix now a defining section σi ∈ H 0 (X, O(Ei )) and a Hermitian
metric hi on O(Ei ), so that ri := |σi |2hi is a smooth nonnegative function of X which
vanishes precisely along Ei and is uniformly comparable to |zi |2 in any adapted
coordinate chart as above where zi is the local defining equation of Ei . In particular,
log ri
x̃i := ,
log |t|
S
is now defined on the whole X\ j Ej , and in the adapted coordinates as above it is
equal to 2xi up to very small errors (as t approaches 0). It follows that on Xt (for
|t| sufficiently small) in
an adapted coordinate chart near a point of EJ0 as above,
the point 21 x̃0 , . . . , 12 x̃p will lie very close to ∆J .

3. Diameter lower bound

In this section we prove the diameter lower bound in Theorem 1.1.
The setting is the same as in the previous section, so X → ∆∗ is a polarized
Calabi-Yau degeneration
P family with m = P dim Sk(X) > 0, with a semistable model
X → ∆ with X0 = i∈I Ei and KX/∆ = i∈I ai Ei . We fix also an embedding of
the family X ֒→ PN0 × ∆ and denote by L → X the restriction of the hyperplane
bundle.
We choose a nonempty EJ which realizes the maximum in (2.2), with m = |J|−1,
and relabel so that J = {0, . . . , m}. We also denote by U an open neighborhood of
 DIAMETER BOUNDS FOR DEGENERATING CALABI-YAU METRICS 5

EJ in X which can be covered by finitely many adapted coordinate charts as above.

In particular, in these charts we have
m n
2 Y dzj dzj Y
(3.1) in Ωt ∧ Ωt = |fJ |2 i ∧ ∧ idzk ∧ dzk .
j=1
zj zj
k=m+1

We need the following construction:

Proposition 3.1. We can find a metric ωt′ on Xt in the class | log1 |t|| c1 (L)|Xt , a
Lipschitz function ρt on Xt , and an open neighborhood U of EJ in X as above, with
the following properties:
(a) The function ρt is supported on the closure of Bt = {ρt < 0} ⊂ U ∩ Xt . On
Bt the function ρt is comparable to a quadratic function in the logarithmic
log |zj |
variables xj = | log |t|| , for j = 1, 2, . . . m, in adapted coordinate charts, with
min ρt = −1 and max ρt = 0.
(b) On Bt in adapted coordinate charts we have
m
i X dzj dzj
(3.2) ωt′ >C −1
∧ ,
| log |t||2 j=1 zj zj
and
(3.3) |dρt |2ωt′ 6 C,
for a fixed constant C independent of t.
For ease of notation, in the rest of the paper we will denote by C > 0 a generic
uniform constant, independent of t, which may vary from line to line.
Proof. Given any point x ∈ EJ we can find k ≫ 1 and sections s0 , . . . , sN ∈
H 0 (X, Lk ) (for some N = N (k)) so that in some adapted coordinate chart Vx
near x we have that none of the sections s0 , sm+1 , . . . , sN vanishes, while sj = 0
is a defining equation for Ej , 1 6 j 6 m, and so sj /s0 is comparable to zj for
1 6 j 6 m.
′
We construct a Kähler metric ωx,t on Xt by pulling back a suitable toric metric
N0
on P , which on the complement of the zeros of all the si ’s is given by
 
′ 1 ¯ log |s1 /s0 | log |sN /s0 |
ωx,t = i∂ ∂u ,..., ,
k | log |t|| | log |t||
where u(x1 , . . . , xN ) is a smooth convex function in RN which is asymptotic to
v(x1 , . . . , xN ) = max(0, x1 , . . . , xN ) at infinity, and with D2 u > C −1 Id on a ball of
radius comparable to 1 containing the image of Vx ∩ Xt in the logarithmic coor-
dinates. For example, an explicit such u can be produced as the convolution of v
with a smooth mollifier. By construction, ωx,t ′
lies in the class | log1 |t|| c1 (L)|Xt , and
it satisfies (3.2) on Vx ∩ Xt .
We then choose finitely many x(1) , . . . , x(M) ∈ EJ such that the corresponding
Vx(1) , . . . , Vx(M ) cover EJ , let U be their union, and define
M
1 X ′
ωt′ = ωx(k) ,t .
M
k=1
1
This is our desired Kähler metric on Xt in | log |t|| c1 (L)|Xt which satisfies (3.2) in
adapted coordinate charts on U ∩ Xt .
6 YANG LI AND VALENTINO TOSATTI

Next, we consider the function

m  2 m  2
X x̃j X log rj
ρ̂ = A − bj − 1 = A − bj − 1,
j=1
2 j=1
2 log |t|
S
on U \ i∈I Ei . Choosing the constants bj in the strict interior of ∆J we can ensure
the minimum of ρ̂ on U ∩ Xt equals −1, and choosing A suitably large independent
of t, we can ensure {ρ̂ 6 0} is compactly contained in U . Now we define
(
min(ρ̂, 0)|Xt on U ∩ Xt
ρt = ,
0 on Xt \U
which satisfies the requirements in (a). We let Bt = {ρt < 0} ⊂ Xt . Lastly, (3.3)
follows immediately from part (a) and (3.2). 
We can now give the proof of the diameter lower bound in Theorem 1.1:
Proof of the diameter lower bound in Theorem 1.1. Thanks to Proposition 3.1, on
Bt ⊂ Xt we have
|dρt |2ωt′ 6 C,
for some constant C independent of t. We then use this together with the elemen-
tary inequality |dρt |2ωt 6 |dρt |2ω′ trωt ωt′ to get
t
Z 2 Z 2 Z Z
2
|dρt |ωt dµt = |dρt |ωt dµt 6 |dρt |ωt dµt 6 C trωt ωt′ dµt ,
Xt Bt Bt Xt

while from (2.1) we get

n Xt ωt′ ∧ ωtn−1 n Xt c1 (L) · c1 (L)n−1
Z R R
′
trωt ωt dµt = = 6 C,
ωn
R R
Xt Xt t
c (L)n
Xt 1

and so
Z
(3.4) |dρt |ωt dµt 6 C.
Xt

Define two subsets of Xt by A1 = {ρt < − 23 } and A2 = {− 31 6 ρt 6 0}. Given

two points x ∈ A1 , y ∈ A2 which are connected by a unique minimal geodesic γx,y
(w.r.t. ωt ), we can bound
Z
(3.5) ρt (y) − ρt (x) 6 |dρt |ωt ds,
γx,y

where γx,y is parametrized with respect to ωt -arclength.

Combining (3.5) with Cheeger-Colding’s segment inequality [5, Theorem 2.11]
applied to the function |dρt |ωt we obtain
(3.6)
Z Z Z !
Dt (µt (A1 ) + µt (A2 )) |dρt |ωt dµt > C −1 |dρt |ωt ds dµx dµy
Xt A1 ×A2 γx,y
Z
> C −1 (ρt (y) − ρt (x))dµx dµy
A1 ×A2
C −1
> µt (A1 )µt (A2 ),
3
 DIAMETER BOUNDS FOR DEGENERATING CALABI-YAU METRICS 7

R
where Dt = diam(Xt , ωt ), and in the A1 ×A2 we are actually only integrating over
the subset of pairs (x, y) which are joined by a unique ωt -minimal geodesic, which
has full measure (cf. [5]).
Combining (3.4) and (3.6) gives
µt (A1 )µt (A2 ) 6 CDt (µt (A1 ) + µt (A2 )) 6 CDt .
Lastly, from the definition of ρt and from (3.1), a direct computation in polar
coordinates (analogous to the one in [4]) gives
µt (A1 ) > C −1 , µt (A2 ) > C −1 ,
for a fixed constant C, and so Dt > C −1 , as desired.


4. Diameter upper bound

Here we prove the diameter upper bound in Theorem 1.1.

Proof of the diameter upper bound in Theorem 1.1. Let ωt′ be the Kähler metric
defined in Proposition 3.1, and let ωFS,t = | log1 |t|| ωFS |Xt , where ωFS is a suitable
Fubini-Study metric on PN0 scaled so that ωFS |Xt ∈ c1 (L)|Xt . Then the metrics ωt′
and ωFS,t are cohomologous, and on Bt ⊂ Xt in any adapted coordinate chart we
have
m n
i X dzj dzj i X
(4.1) ωt′ + ωFS,t > C −1 ∧ + C −1
dzj ∧ dzj .
| log |t||2 j=1 zj zj | log |t|| j=m+1

Let us then fix a point x ∈ EJ0 , an adapted coordinate chart near x, and in these
coordinates define a local Kähler metric ω̃t on Xt by the RHS of (4.1). Inside this
coordinate chart intersected Xt , define also B̃t to be a Euclidean rectangle which is
contained inside Bt so that in the metric ω̃t , in the first m complex directions B̃t has
length ∼ 1 in the radial directions and length ∼ | log |t||−1 in the logarithmic angular
1
directions, and in the other n − m complex directions B̃t has length ∼ | log |t||− 2 .
Therefore, given any x, y ∈ B̃t , if we denote by γx,y the Euclidean straight line in
B̃t joining them, parametrized linearly by 0 6 s 6 1, then |γ̇x,y (s)|ω̃t 6 C for a
uniform constant C independent of t and s.
On B̃t we have
C −1 C
n
µt 6 ω̃tn 6 µt ,
| log |t|| | log |t||n
and again by direct computation in polar coordinates (as in [4]), thanks to the
definition of B̃t and to (3.1) we have that
Z
(4.2) C −1 6 dµt 6 1,
B̃t

and so
C −1 C
Z
6 ω̃tn 6 .
| log |t||n B̃t | log |t||n
If we then define
ω̃tn
µ̃t = R ,
ω̃ n
B̃t t
8 YANG LI AND VALENTINO TOSATTI

then µ̃t is uniformly comparable to µt on B̃t and

(4.3) µt (B̃t ) > C −1 µt (Bt ) > C −1 .
Thanks to (4.1) we have
n B̃t ωt ∧ ω̃tn−1
Z R Z
n
trω̃t ωt dµ̃t = R n 6 C| log |t|| ωt ∧ (ωt′ + ωFS,t )n−1
B̃t ω̃ t X
(4.4) Z B̃t
t

6C c1 (L) · c1 (L)n−1 6 C.
Xt

We wish to use this to prove that

Z
(4.5) distωt (x, y)dµ̃t (x)dµ̃t (y) 6 C.
B̃t ×B̃t

Indeed, since |γ̇x,y (s)|ω̃t 6 C, we can estimate

Z 1
1
distωt (x, y) 6 C (trω̃t ωt (sy + (1 − s)x)) 2 ds,
0

and
Z Z Z 1
1
distωt (x, y)dµ̃t (x)dµ̃t (y) 6 C (trω̃t ωt (sy+(1−s)x)) 2 ds dµ̃t (x)dµ̃t (y).
B̃t ×B̃t B̃t ×B̃t 0

We can then argue as in [6, Lemma 1.3], using Fubini

Z Z 1
1
(trω̃t ωt (sy + (1 − s)x)) 2 ds dµ̃t (x)dµ̃t (y)
B̃t ×B̃t 0
Z 1Z
1
= (trω̃t ωt (sy + (1 − s)x)) 2 dµ̃t (x)dµ̃t (y)ds
0 B̃t ×B̃t
Z 1 Z
2 1
= (trω̃t ωt (sy + (1 − s)x)) 2 dµ̃t (x)dµ̃t (y)ds
0 B̃t ×B̃t
Z 1 Z
1
+ (trω̃t ωt (sy + (1 − s)x)) 2 dµ̃t (y)dµ̃t (x)ds
1
2 B̃t ×B̃t

and in the two innermost integrals we change variable from x (resp. y) to z =

sy + (1 − s)x with 0 6 s 6 12 (resp. 21 6 s 6 1), noting that µ̃t (x) 6 C µ̃t (z) (resp.
µ̃t (y) 6 C µ̃t (z)). Thus both of these innermost integrals can be bounded above by
Z Z  12
1
C (trω̃t ωt (z)) 2 dµ̃t (z) 6 C trω̃t ωt (z)dµ̃t (z) 6 C,
B̃t B̃t

by (4.4), and (4.5) follows.

But (4.5) is equivalent to
Z
distωt (x, x′ )dµt (x)dµt (x′ ) 6 C,
B̃t ×B̃t

hence for some x′ ∈ B̃t we have

Z
distωt (x, x′ )dµt (x) 6 C,
B̃t
 DIAMETER BOUNDS FOR DEGENERATING CALABI-YAU METRICS 9

which gives a weak L1 -estimate

C
µt {x ∈ B̃t | distωt (x, x′ ) > r} 6 .
r
Taking r ≫ 1 independent of t, we can ensure that
µt (Bωt (x′ , r)) > µt (B̃t ∩ Bωt (x′ , r))
= µt (B̃t ) − µt {x ∈ B̃t | distωt (x, x′ ) > r}
C
> C −1 −
r
> C −1 = C −1 µt (Xt ),
using here (4.3). The Bishop-Gromov volume comparison theorem then gives us
that µt (Bωt (x′ , 1)) > C −1 µt (Xt ), or equivalently
ωn
R
R Xt t n 6 C.
ω
Bω (x′ ,1) t
t

This bound can then be inserted into a well-known result of Yau (see e.g. [24,
Lemma 3.2]): given a complete d-dimensional Riemannian manifold (M, g) with
nonnegative Ricci curvature, for any x ∈ M and 1 < R < diam(M, g) we have
R−1 Vol(Bg (x, 2R + 2))
6 .
2d Vol(Bg (x, 1))
Indeed, it suffices to choose R = diam(Xt , ωt ) − 1 to obtain the desired uniform
diameter upper bound for (Xt , ωt ). 

References
[1] P.S. Aspinwall, T. Bridgeland, A. Craw, M.R. Douglas, M. Gross, A. Kapustin, G.W. Moore,
G. Segal, B. Szendrői, P.M.H. Wilson, Dirichlet branes and mirror symmetry, Clay Mathe-
matics Monographs, 4. American Mathematical Society, Providence, RI; Clay Mathematics
Institute, Cambridge, MA, 2009.
[2] C. Birkar, Existence of log canonical flips and a special LMMP, Publ. Math. Inst. Hautes
Études Sci. 115 (2012), 325–368.
[3] S. Boucksom, Remarks on minimal models of degenerations, preprint, 2014.
[4] S. Boucksom, M. Jonsson, Tropical and non-Archimedean limits of degenerating families of
volume forms, J. Éc. polytech. Math. 4 (2017), 87–139.
[5] J. Cheeger, T.H. Colding, Lower bounds on Ricci curvature and the almost rigidity of warped
products, Ann. of Math. (2) 144 (1996), no. 1, 189–237.
[6] J.-P. Demailly, T. Peternell, M. Schneider, Kähler manifolds with numerically effective Ricci
class, Compositio Math. 89 (1993), no. 2, 217–240.
[7] X. Fu, B. Guo, J. Song, Geometric estimates for complex Monge-Ampère equations, J. Reine
Angew. Math. 765 (2020), 69–99.
[8] M. Gross, Calabi-Yau manifolds and mirror symmetry, in Calabi-Yau manifolds and related
geometries (Nordfjordeid, 2001), 69–159, Universitext, Springer, Berlin, 2003.
[9] M. Gross, P.M.H. Wilson, Large complex structure limits of K3 surfaces, J. Differential
Geom. 55 (2000), no. 3, 475–546.
[10] C.D. Hacon, C. Xu, Existence of log canonical closures, Invent. Math. 192 (2013), no. 1,
161–195.
[11] Y. Kawamata, Flops connect minimal models, Publ. Res. Inst. Math. Sci. 44 (2008), no. 2,
419–423.
[12] G. Kempf, F. Knudsen, D. Mumford, B. Saint-Donat, Toroidal embeddings. I. Lecture Notes
in Mathematics, Vol. 339. Springer-Verlag, Berlin-New York, 1973. viii+209 pp.
[13] M. Kontsevich, Y. Soibelman, Affine structures and Non-Archimedean analytic spaces, in
The Unity of Mathematics, Progress in Mathematics Volume 244, Springer, (2006), 321–385.
10 YANG LI AND VALENTINO TOSATTI

[14] Y. Li, SYZ conjecture for Calabi-Yau hypersurfaces in the Fermat family, to appear in Acta
Math.
[15] Y. Li, Uniform Skoda integrability and Calabi-Yau degeneration, preprint, arXiv:2006.16961.
[16] Y. Li, Metric SYZ conjecture and non-Archimedean geometry, preprint, arXiv:2007.01384.
[17] J. Kollár, J. Nicaise, C. Xu, Semi-stable extensions over 1-dimensional bases, Acta Math.
Sin. (Engl. Ser.) 34 (2018), no. 1, 103–113.
[18] M. Mustaţă, J. Nicaise, Weight functions on non-Archimedean analytic spaces and the
Kontsevich-Soibelman skeleton, Algebr. Geom. 2 (2015), no. 3, 365–404.
[19] J. Nicaise, C. Xu, The essential skeleton of a degeneration of algebraic varieties, Amer. J.
Math. 138 (2016), no. 6, 1645–1667.
[20] X. Rong, Y. Zhang, Continuity of extremal transitions and flops for Calabi-Yau manifolds,
Appendix B by Mark Gross, J. Differential Geom. 89 (2011), no. 2, 233–269.
[21] J. Song, Riemannian geometry of Kähler-Einstein currents, preprint, arXiv:1404.0445.
[22] A. Strominger, S.-T. Yau, E. Zaslow, Mirror symmetry is T -duality, Nuclear Phys. B 479
(1996), no. 1-2, 243–259.
[23] S. Takayama, On moderate degenerations of polarized Ricci-flat Kähler manifolds, J. Math.
Sci. Univ. Tokyo 22 (2015), no. 1, 469–489.
[24] V. Tosatti, Limits of Calabi-Yau metrics when the Kähler class degenerates, J. Eur. Math.
Soc. (JEMS) 11 (2009), no. 4, 755–776.
[25] V. Tosatti, Families of Calabi-Yau manifolds and canonical singularities, Int. Math. Res.
Not. IMRN 2015, no. 20, 10586–10594.
[26] V. Tosatti, Collapsing Calabi-Yau manifolds, Surveys in Differential Geometry 23 (2018),
305–337, International Press, 2020.
[27] C.-L. Wang, On the incompleteness of the Weil-Petersson metric along degenerations of
Calabi-Yau manifolds, Math. Res. Lett. 4 (1997), no. 1, 157–171.
[28] S.-T. Yau, On the Ricci curvature of a compact Kähler manifold and the complex Monge-
Ampère equation, I, Comm. Pure Appl. Math. 31 (1978), 339–411.
[29] Y. Zhang, Note on equivalences for degenerations of Calabi-Yau manifolds, in Surveys in
Geometric Analysis 2017, 186–202, Science Press, Beijing, 2018.

Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA

02139
Email address: yangmit@mit.edu

Department of Mathematics and Statistics, McGill University, Montréal, Québec

H3A 0B9, Canada
Email address: valentino.tosatti@mcgill.ca