Post on 21-Oct-2020
Hypertoric varieties and hyperplane arrangements
Takahiro Nagaoka
Kyoto Univ.
June 16, 2018
Motivation - Study of the geometry of symplectic variety
Symplectic variety (Y0, ω) · · ·Very special but interestingeven dim algebraic variety
Topic
Classification and finding new examples
⇝ The case of dim ≥ 4 is open.
Universal (Poisson) deformation and its construction
⇝ Its discriminant locus DC is hyperplane arr. with W -action.
“Good”(= crepant) resolution π : (Y , ω) → (Y0, ω)⇝ construction and counting all crepant resolutions of Y0.⇝ This is related to the structure of DC.
We will study these problems for hypertoric variety.
Takahiro Nagaoka
Contents
Intro - symplectic variety and Poisson deformation
Hypertoric variety - definition and its universalPoisson deformation
Application 1 - classification of affine hypertoricvarieties by matroids
Application 2 - counting good resolutions of affinehypertoric variety by hyperplane arr.
Takahiro Nagaoka
Intro - symplectic variety and Poisson deformation
Takahiro Nagaoka
Symplectic variety
(Y0, ω) : symplectic varietydef⇔ normal alg. var. with holomo. symplectic form ω on (Y0)reg .
(Y0, ω) : conical symplectic varietydef⇔ affine symplectic variety with “good” C∗-action.
Example
V⊕V ∗/Γ : Symplectic quotient singularity (Γ ⊂ Sp(V ⊕V ∗))
⇝ eg. ADE -type surface singularity (An : zn+1 − xy = 0).
O : Nilpotent orbit closure in g
M0(Q, v,w) : (affine) quiver variety
Y (A, 0) : (affine) hypertoric variety
Takahiro Nagaoka
Symplectic resolution = crepant resolution
For symplectic variety (Y0, ω),
π : (Y , ω) → (Y0, ω) : symplectic resolutiondef⇔ π∗ω extends to a symplectic form ω on whole Y .
Remark
For resolution π : Y → Y0 of sympleccitc variety Y0,
π : symplectic resolution ⇔ π : crepant, i.e., π∗KY0 = KY
Takahiro Nagaoka
Poisson variety and Poisson deformation
(Y , ω) has natural Poisson str. (Y , {−,−}0)(Poisson str. · · · bracket on OY satisfying Leipnitz rule & Jacobi id)
Poisson deformation
(Y , {−,−}0) (Y, {−,−}) : Poisson deform. of (Y , {−,−}0)
0 S
⊂
flat
∈
Definition (UPD (Universal Poisson deformation))
(Yuniv , {−,−}) → S is the universal Poisson deformation of Y .def⇔ ∀ (infinitesimal) Poisson deformation (X , {−,−}′) → SpecA,
X ∼= Yuniv ×S SpecA Yuniv
SpecA S∃!f
Takahiro Nagaoka
UPD of conical symplectic variety
Theorem (Namikawa)
π : Y → Y0 : proj. sympl. resolution of conical sympl. variety Y0⇒ There exists UPDs Yuniv , Yuniv0 of Y and Y0. Moreover,
Y Y0
Yuniv Yuniv0
0 0
H2(Y ,C) H2(Y ,C)/W
⊃
π
⊃Π
µ
∋ ∋ψ
µW
, where W is Namikawa-Weyl group W ⊂ GL(H2(Y ,C)).
Discriminant locusDC:={h∈ H2(Y ,C) |fiber of Yuniv0 ×H2/WH2 →H2 at h is singular}
⇝ Hyperplane Arrangement !! i.e., ∃AC s.t. DC =∪
H∈AH.
Takahiro Nagaoka
Example of discriminant arrangement ACIn general, W ↷AC⊂H2(Y ,C) as reflection w.r.t. some H ∈ AC.
Example of ACC2/G : ADE -type surface singularity (G ⊂ SL2(C))
⇝{
W = WG (usual) Weyl groupAC = Weyl arrangement
Symn+1(C2/G ) : symplectic quotient singularity
⇝{
W = Z/2Z×WGAC = cone over extended Catalan arr. Cat
[−n,n]ΦG
(Cat[−n,n]ΦG
:= {Hλ,k : ⟨λ,−⟩ = k | λ ∈ ΦG ,−n ≤ k ≤ n} ⊂ h)
Moreover, DC =∪
H∈AC H has connection to birational geometryof Y0 (see later).
Goal
Describe the diagram of UPD for hypertoric variety.Takahiro Nagaoka
Hypertoric variety - definition and its universal Poissondeformation
Takahiro Nagaoka
Q. What is hypertoric variety?
A. algebraic variety with “combinatorial” flavor (like toric variety).
Combinatorics Geometry
{hyperplane arrangement HαB} {hypertoric variety Y (A, α)}
,where matrices A and B satisfy 0 Zn−d Zn Zd 0B A
Philosophy
Read off the geometric properties of Y (A, α) from thecombinatorics of associated hyperplane arrangement HαB .
Takahiro Nagaoka
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
[1:0:0]
[0:1:0] [0:0:1]
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
What is hypertoric variety?
Example
Let A = (1 1 1), BT =(1 0 −10 1 −1
)= ( b1 b2 b3 ).
For α̃ = (1, 1, 1), set HαB := {Hi : ⟨bi,−⟩ = α̃i}.
Takahiro Nagaoka
0
Singular pt
0
Translation
Y(A,α)=
Y(A, 0)
Hypertoric variety
A ∈ Matm×n(Z) s.t. unimodular matrix, i.e., ∀d × d-minors=0,±1
Take B as 0 Zn−d Zn N ∼= Zd 0B A exact.
BT =
(b1 b2 · · · bn
)A =
(a1 a2 · · · an
)
TdC ↷ (C2n = Cn⊕Cn, ωC :=n∑
j=1
dzj ∧ dwj), “Hamiltonian action”
t · (zi ,wi ) := (taizi , t−aiwi )
⇝ ∃ moment map µ : C2n → Cd : (z,w) 7→∑n
j=1 zjwjaj
Define hypertoric variety as the “quotient µ−1(0)/TdC”.
Takahiro Nagaoka
Hypertoric variety
Definition (Hypertoric variety Y (A, α))
For each parameter α := Aα̃ (α̃ ∈ Zn), the “quotient” space
Y (A, α) := µ−1(0)α−ss//TdC
is hypertoric variety. (µ−1(0)α−ss ⊆µ−1(0) is α-semistable set)
Inclusion µ−1(0)α−ss ⊆ µ−1(0)0−ss = µ−1(0)
⇝ projective morphism π : Y (A, α) → Y (A, 0)
Fact
For generic α,
1 Y (A, α) is 2(n − d) = 2 rankBT dim smooth symplecticvariety.
2 π : Y (A, α) → Y (A, 0) is (conical) symplectic resolution.
Takahiro Nagaoka
Example: hypertoric variety associated to graph
Example (Toric quiver variety)
G : directed graph, Z|E | AG−−→→ N ⊂ Z|V | : eij 7→ ei − ej,
1•
0• 2• 4•
3•
G
⇝ AG =
1 1 1 0 0 0−1 0 0 1 0 00 −1 0 0 1 00 0 −1 0 0 10 0 0 −1 −1 −1
⇝ Y (AG , α) : toric quiver variety⇝ dimY (AG , α) = 2(circuit rank of G )
Takahiro Nagaoka
Example
Example (A2-type surface singularity)
A = AG =
(1 0 −10 1 −1
), G =
•• •
, BT =(1 1 1
)
Y (A, α) S̃A2 : minimal resolution of SA2
Y (A, 0) SA2 : {u3 − xy = 0} : A2-type singularity
∼
π
∼
Remark
In general, for G = edge graph of ℓ+ 1-gon,Y (AG , 0) = SAℓ : Aℓ-type surface singularity.
Takahiro Nagaoka
Description of UPD for hypertoric variety
We want to determine the UPDs for π : Y (A, α) → Y (A, 0).
Theorem (Braden-Licata-Proudfoot-Webster, N-.)
Y (A, α) Y (A, 0)
X (A, α) X (A, 0)/WB
0 0
Cd Cd/WB
⊃
π
⊃ΠWB
µ
ψ∋ ∋
µWB.
Assume BT=
ℓ1︷ ︸︸ ︷ ℓs︷ ︸︸ ︷(c1 · · · c1 · · · cs · · · cs ), where ck1 ̸= ±ck2 if k1 ̸= k2.
Description of WB -action and discriminant locus DCWB := Sℓ1 × · · · ×Sℓs ↷ Cd = SpanC(a1, . . . . . . , an).AC={H⊂Cd | codimH = 1 and H is generated by some ai’s}.
Takahiro Nagaoka
Application 1 - classification of affine hypertoricvarieties Y (A, 0) by matroids
Takahiro Nagaoka
Classification of Y (A, 0)
What operation to A preserves the isomorphism calss of Y (A, 0)?
Definition
A ∼ A′ def⇔ A′ is obtained from A by a seq. of the followings:· elementary row operations over Z,· interchanging column vectors, i.e., ai ↔ aj,· multiplying column vector by −1, i.e., ai 7→ −ai.
A ∼ A′ ⇒Y (A, 0)∼=Y (A′, 0) : Tn−dC -eq. iso as conical sympl. var.
Theorem (Arbo-Proudfoot, N-.)
Y (A, 0) ∼= Y (A′, 0) as conical symplectic variety ⇔ A ∼ A′.
Remark
M(A) : regular matroid associated to A whose dual is M(BT ).
A ∼ A′ ⇔ M(A) ∼= M(A′) ⇔ M(BT ) ∼= M(B ′T ).
Takahiro Nagaoka
Corollary of Theorem
By classification theorem, in particular, for toric quiver varieties,
Y (AG , 0) ∼= Y (AG ′ , 0) ⇔ M(AG ) ∼= M(AG ′).
Theorem (Whitney’s 2-isomorphism theorem)
Let G and G ′ be graphs without isolated vertices. Then,M(AG ) ∼= M(AG ′)⇔ G ′ can be transformed into G by a sequence of the followingoperations;
(i) Vertex identification and vertex cleaving
(ii) Whitney twist
• ••
• •
(i)∼• •
• •• •
.• • •
• • •(ii)∼
• • •
• • •
Takahiro Nagaoka
4-dimensional classification
We only have to classify M(BT ) of rank = n − d = 12 dimY (A, 0).
Theorem (N-. Classification of 4-dimensional Y (A, 0))
Every 4-dimensional Y (A, 0) is isomorphic to one of the followings;
(i) SAℓ1−1 × SAℓ2−1 .
(ii) Omin(ℓ1, ℓ2, ℓ3):=(u1 x1 x3y1 u2 x2y3 y2 u3) ∈ sl3∣∣∣∣∣∣ All 2×2-minors of
uℓ11 x1 x3y1 uℓ22 x2y3 y2 u
ℓ33
= 0.
(i) BT =
ℓ1︷ ︸︸ ︷ ℓ2︷ ︸︸ ︷(1 · · · 1 0 · · · 0 )0 · · · 0 1 · · · 1
⇝ G =
(ii) BT =
ℓ1︷ ︸︸ ︷ ℓ2︷ ︸︸ ︷ ℓ3︷ ︸︸ ︷(1 · · · 1 0 · · · 0 1 · · · 1 )0 · · · 0 1 · · · 1 1 · · · 1
⇝ G =
Takahiro Nagaoka
6-dimensional classification
(1) G =
(2) G =
(3) G =
(4) G =
(5) G =
Takahiro Nagaoka
Application 2 - counting good resolutions of affinehypertoric variety by hyperplane arr.
Takahiro Nagaoka
Counting crepant resolutions of Y (A, 0)
It is important to study “good”(=crepant) resolutions of singularity.
Fact
For “generic”α ∈ Zd , πα : Y (A, α) → Y (A, 0) gives a crepantresolution.
Conversely, all crepant resolutions are obtained by this form.
Remark
A := {H ⊂ Rd | codimH = 1 and H is generated by some ai’s}.
α ∈ Zd : generic def⇔ α /∈ D :=∪H∈A
H.
Question Which α gives you different resolutions?
Takahiro Nagaoka
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
a2 a3
a4
a1
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Example
Example
AG =
1 0 −1 10 1 1 −1−1 −1 0 0
, G = • • • , BT = (1 −1 1 00 0 1 1
),
A = AG ⊂ NR. Recall for α, HαB := {Hi : ⟨bi,−⟩ = α̃i} ⊂ Rn−d .
Question: Which chambers give different resolutions ?⇝ Note WB := S2-action on A ⊂ NRAnswer: Chambers in a fundamental domain of WB -action.
Takahiro Nagaoka
0
0
Counting crepant resolutions of Y (A, 0)
Assume BT=
ℓ1︷ ︸︸ ︷ ℓs︷ ︸︸ ︷(c1 · · · c1 · · · cs · · · cs
).
Recall WB :=Sℓ1×· · ·×Sℓs ↷ Rd=SpanR({aj}) as permutationof aj’s.
Theorem (Namikawa, Braden-Licata-Proudfoot-Webster)
Each chamber in a fundamental domain of WB -action gives alldifferent crepant resolutions of Y (A, 0).
Corollary
The number of crepant resolutions of Y (A, 0) is
#{chamber of A}#WB
=|χA(−1)|ℓ1! · · · ℓs !
,
where χA(t) is the characteristic polynomial of A.
Takahiro Nagaoka
Counting crepant resolutions of Omin(ℓ1, ℓ2, ℓ3)
Lemma
For Omin(ℓ1, ℓ2, ℓ3), the corresponding A is (the essentializationof) Aℓ1,ℓ2,ℓ3 in Rℓ1+ℓ2+ℓ3 :
Hijk : xi + yj + zk = 0 (1 ≤ i ≤ ℓ1, 1 ≤ j ≤ ℓ2, 1 ≤ k ≤ ℓ3)Hxi1,i2 : xi1 − xi2 = 0 (1 ≤ i1 < i2 ≤ ℓ1)Hyj1,j2 : yj1 − yj2 = 0 (1 ≤ j1 < j2 ≤ ℓ2)Hzk1,k2 : zk1 − zk2 = 0 (1 ≤ k1 < k2 ≤ ℓ3)
Proposition (Edelman-Reiner)
χAℓ1,ℓ2,ℓ3 (t) =t2(t − 1)(t − 2) · · · (t − (ℓ1 + ℓ2 − 1)) (ℓ3 = 1)
t2(t − 1)ℓ1+ℓ2∏i=ℓ1+1
(t − i)ℓ1+ℓ2−1∏j=ℓ2+1
(t − j) (ℓ3 = 2)
Takahiro Nagaoka
Counting crepant resolutions of Omin(ℓ1, ℓ2, ℓ3)
Corollary
The number of crepant resolutions of Omin(ℓ1, ℓ2, ℓ3) is(ℓ1+ℓ2ℓ1
)(ℓ3 = 1)
(ℓ1+ℓ2+1ℓ1 )(ℓ1+ℓ2+1
ℓ2)
ℓ1+ℓ2+1(ℓ3 = 2)
This gives a geometric meaning of Aℓ1,ℓ2,ℓ3 considered by Edelmanand Reiner.
Remark
For ℓ1, ℓ2, ℓ3 ≥ 3, it is known Aℓ1,ℓ2,ℓ3 is NOT free arrangement.χA3,3,3(t) = t
2(t−1)(t−5)(t−7)(t4−23t3+200t2−784t+1188)
Question
Compute the number of chambers of Aℓ1,ℓ2,ℓ3 for ℓ1, ℓ2, ℓ3 ≥ 3.Takahiro Nagaoka
Thank you for listening!!
Takahiro Nagaoka