• 検索結果がありません。

Exact WKB Analysis and Cluster Algebras

N/A
N/A
Protected

Academic year: 2022

シェア "Exact WKB Analysis and Cluster Algebras"

Copied!
56
0
0

読み込み中.... (全文を見る)

全文

(1)

Exact WKB Analysis and

Cluster Algebras

Kohei Iwaki (RIMS, Kyoto University)

(joint work with Tomoki Nakanishi)

Winter School on Representation Theory January 21, 2015

1 / 22

(2)

Exact WKB analysis

Schr ¨odinger equation:

(d2

dz2 −η2Q(z) )

ψ(z, η)=0

wherezis an complex variable,η=~1 >0is alarge parameter.

WKB (Wentzel-Kramers-Brillouin) solutions: ψ±(z, η) = e±η

z

Q(z0)dz0 n=0

ηn12ψ±,n(z)

In general, WKB solutions aredivergent(i.e., formal solutions).

Exact WKB analysis=WKB method+Borel resummation. S[ψ±](z, η)∼ψ±(z, η) asη→+∞

Monodromy/connection matrices of (Borel resummed) WKB solutions are described by “Voros symbols”.

[Voros 83], [Sato-Aoki-Kawai-Takei 91], [Delabaere-Dillinger-Pham 93], ...

(3)

Exact WKB analysis

Schr ¨odinger equation:

(d2

dz2 −η2Q(z) )

ψ(z, η)=0

wherezis an complex variable,η=~1 >0is alarge parameter.

WKB (Wentzel-Kramers-Brillouin) solutions:

ψ±(z, η) = e±η

z

Q(z0)dz0 n=0

ηn12ψ±,n(z)

In general, WKB solutions aredivergent(i.e., formal solutions).

Exact WKB analysis=WKB method+Borel resummation. S[ψ±](z, η)∼ψ±(z, η) asη→+∞

Monodromy/connection matrices of (Borel resummed) WKB solutions are described by “Voros symbols”.

[Voros 83], [Sato-Aoki-Kawai-Takei 91], [Delabaere-Dillinger-Pham 93], ...

2 / 22

(4)

Exact WKB analysis

Schr ¨odinger equation:

(d2

dz2 −η2Q(z) )

ψ(z, η)=0

wherezis an complex variable,η=~1 >0is alarge parameter.

WKB (Wentzel-Kramers-Brillouin) solutions:

ψ±(z, η) = e±η

z

Q(z0)dz0 n=0

ηn12ψ±,n(z)

In general, WKB solutions aredivergent(i.e., formal solutions).

Exact WKB analysis=WKB method+Borel resummation.

S[ψ±](z, η)∼ψ±(z, η) asη→+∞

Monodromy/connection matrices of (Borel resummed) WKB solutions are described by “Voros symbols”.

(5)

Cluster algebras (of rank

n≥ 1)

Acluster algebra[Fomin-Zelevinsky 02] is defined in terms ofseeds.

• A seed is a triplet(B,x,y)where

∗ skew-symmetric integer matrixB=(bi j)ni,j=1

clusterx-variables x=(xi)ni=1

clustery-variables y=(yi)ni=1 These two variables satisfyyi=rin

j=1(xj)bji (ri : “coefficient”).

A“signed” mutationatk∈ {1, . . . ,n}with signε∈ {±}: µ(kε): (B,x,y)7→(B0,x0,y0)defined by

b0i j=

−bi j i=kor j=k

bi j+[bik]+bk j+bik[bk j]+ otherwise.

x0i =



xk1





n

j=1

xj[−εbjk]+



(1+ykε) i=k

xi i,k.

y0i =

yk1 i=k

yiyk[εbki]+

(1+ykε)bki i,k.

Here[a]+=max(a,0). (The coefficientsrialso mutate.)

3 / 22

(6)

Cluster algebras (of rank

n≥ 1)

Acluster algebra[Fomin-Zelevinsky 02] is defined in terms ofseeds.

• A seed is a triplet(B,x,y)where

∗ skew-symmetric integer matrixB=(bi j)ni,j=1

clusterx-variables x=(xi)ni=1

clustery-variables y=(yi)ni=1 These two variables satisfyyi=rin

j=1(xj)bji (ri : “coefficient”).

A“signed” mutationatk∈ {1, . . . ,n}with signε∈ {±}: µ(kε): (B,x,y)7→(B0,x0,y0)defined by

b0i j=

bi j i=kor j=k

bi j+[bik]+bk j+bik[bk j]+ otherwise.

x0i =



xk1





n

j=1

xj[−εbjk]+



(1+ykε) i=k

xi i,k.

y0i =

yk1 i=k

yiyk[εbki]+

(1+ykε)bki i,k.

= ,

(7)

Results and Application [I-Nakanishi 14]

• Cluster algebraic structure appears in many contexts:

I representation of quivers

I Teichm ¨uller theory

I hyperbolic geometry

I discrete integrable systems

I Donaldson-Thomas invariants and their wall-crossing

I supersymmetric gauge theory

I · · ·

Main result: We addExact WKB analysisin the above list: skew-symmetric matrixB ↔ Stokes graph

cluster variables ↔ Voros symbols

cluster mutation ↔ Stokes phenomenon (forη→ ∞)

Application:Identities of Stokes automorphsimsin the exact WKB analysis (c.f., [Delabaere-Dillinger-Pham 93]) follow from periodicityof corresponding cluster algebras.

For example: Sγ1Sγ2=Sγ2Sγ21Sγ1

Generalized cluster algebras([Chekhov-Shapiro 11]) also appear when Schr ¨odinger equation has a certain type of regular singularity.

4 / 22

(8)

Results and Application [I-Nakanishi 14]

• Cluster algebraic structure appears in many contexts:

I representation of quivers

I Teichm ¨uller theory

I hyperbolic geometry

I discrete integrable systems

I Donaldson-Thomas invariants and their wall-crossing

I supersymmetric gauge theory

I · · ·

Main result: We addExact WKB analysisin the above list:

skew-symmetric matrixB ↔ Stokes graph cluster variables ↔ Voros symbols

cluster mutation ↔ Stokes phenomenon (forη→ ∞)

Application:Identities of Stokes automorphsimsin the exact WKB analysis (c.f., [Delabaere-Dillinger-Pham 93]) follow from periodicityof corresponding cluster algebras.

For example: Sγ1Sγ2=Sγ2Sγ21Sγ1

Generalized cluster algebras([Chekhov-Shapiro 11]) also appear when Schr ¨odinger equation has a certain type of regular singularity.

(9)

Results and Application [I-Nakanishi 14]

• Cluster algebraic structure appears in many contexts:

I representation of quivers

I Teichm ¨uller theory

I hyperbolic geometry

I discrete integrable systems

I Donaldson-Thomas invariants and their wall-crossing

I supersymmetric gauge theory

I · · ·

Main result: We addExact WKB analysisin the above list:

skew-symmetric matrixB ↔ Stokes graph cluster variables ↔ Voros symbols

cluster mutation ↔ Stokes phenomenon (forη→ ∞)

Application:Identities of Stokes automorphsimsin the exact WKB analysis (c.f., [Delabaere-Dillinger-Pham 93]) follow from periodicityof corresponding cluster algebras.

For example: Sγ1Sγ2=Sγ2Sγ21Sγ1

Generalized cluster algebras([Chekhov-Shapiro 11]) also appear when Schr ¨odinger equation has a certain type of regular singularity.

4 / 22

(10)

Results and Application [I-Nakanishi 14]

• Cluster algebraic structure appears in many contexts:

I representation of quivers

I Teichm ¨uller theory

I hyperbolic geometry

I discrete integrable systems

I Donaldson-Thomas invariants and their wall-crossing

I supersymmetric gauge theory

I · · ·

Main result: We addExact WKB analysisin the above list:

skew-symmetric matrixB ↔ Stokes graph cluster variables ↔ Voros symbols

cluster mutation ↔ Stokes phenomenon (forη→ ∞)

Application:Identities of Stokes automorphsimsin the exact WKB analysis (c.f., [Delabaere-Dillinger-Pham 93]) follow from periodicityof corresponding cluster algebras.

For example: Sγ1Sγ2=Sγ2Sγ21Sγ1

(11)

Contents

§1

Exact WKB analysis

§2

Main results

Refferences

A. Voros, “The return of the quartic oscillator. The complex WKB method”, Ann. Inst. Henri Poincar ´e

39

(1983), 211–338.

T. Kawai and Y. Takei, “Algebraic Analysis of Singular Perturbations”, AMS translation, 2005.

5 / 22

(12)

Contents

§1 Exact WKB analysis

§2

Main results

Refferences

A. Voros, “The return of the quartic oscillator. The complex WKB method”, Ann. Inst. Henri Poincar ´e

39

(1983), 211–338.

T. Kawai and Y. Takei, “Algebraic Analysis of Singular

Perturbations”, AMS translation, 2005.

(13)

Schr ¨odinger equation and WKB solutions

Schr ¨odinger equation: (d2

dz2 −η2Q(z) )

ψ(z, η)=0

∗ η=~1: large parameter

Q(z): rational function (“potential”)

∗ Assume that all zeros ofQ(z)are of order 1, and all poles ofQ(z)are of order≥2.

(We may generalizeQ=Q0(z)+η1Q1(z)+η2Q2(z)+· · ·: finite sum)

WKB solutions(formal solution ofη−1 with exponential factor): ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z)

• WKB solutions aredivergentin general: (|ψ±,n(z)| ∼CAnn!).

6 / 22

(14)

Schr ¨odinger equation and WKB solutions

Schr ¨odinger equation: (d2

dz2 −η2Q(z) )

ψ(z, η)=0

∗ η=~1: large parameter

Q(z): rational function (“potential”)

∗ Assume that all zeros ofQ(z)are of order 1, and all poles ofQ(z)are of order≥2.

(We may generalizeQ=Q0(z)+η1Q1(z)+η2Q2(z)+· · ·: finite sum)

WKB solutions(formal solution ofη−1with exponential factor):

ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z)

• WKB solutions aredivergentin general: (|ψ±,n(z)| ∼CAnn!).

(15)

Schr ¨odinger equation and WKB solutions

Schr ¨odinger equation: (d2

dz2 −η2Q(z) )

ψ(z, η)=0

∗ η=~1: large parameter

Q(z): rational function (“potential”)

∗ Assume that all zeros ofQ(z)are of order 1, and all poles ofQ(z)are of order≥2.

(We may generalizeQ=Q0(z)+η1Q1(z)+η2Q2(z)+· · ·: finite sum)

WKB solutions(formal solution ofη−1with exponential factor):

ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z)

• WKB solutions aredivergentin general: (|ψ±,n(z)| ∼CAnn!).

6 / 22

(16)

Borel resummation method

• Expansion of WKB solution:

ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z) (|ψ±,n(z)| ∼CAnn!).

TheBorel sumofψ±(as a formal series ofη1): S[ψ±]=

a(z)

eyηψ±,B(z,y)dy. Herea(z)=∫z

z0

Q(z0)dz0 and

ψ±,B(z,y)=∑

n=0

ψ±,n(z) Γ(n+12)

(y±a(z))n12 :Borel transformofψ±

• Borel transform=termwise inverse Laplace transform: (

c.f. η−α=

0

eyηyα−1

Γ(α)dy ifReα >0. )

• If the Borel sumsS[ψ±]are well-defined, they give analytic solutions of the Sch ¨odinger equation andS[ψ±]∼ψ±whenη→+∞.

(17)

Borel resummation method

• Expansion of WKB solution:

ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z) (|ψ±,n(z)| ∼CAnn!).

TheBorel sumofψ±(as a formal series ofη1):

S[ψ±]=

a(z)

eyηψ±,B(z,y)dy. Herea(z)=∫z

z0

Q(z0)dz0 and

ψ±,B(z,y)=∑

n=0

ψ±,n(z) Γ(n+12)

(y±a(z))n12 :Borel transformofψ±

• Borel transform=termwise inverse Laplace transform: (

c.f. η−α=

0

eyηyα−1

Γ(α)dy ifReα >0. )

• If the Borel sumsS[ψ±]are well-defined, they give analytic solutions of the Sch ¨odinger equation andS[ψ±]∼ψ±whenη→+∞.

7 / 22

(18)

Borel resummation method

• Expansion of WKB solution:

ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z) (|ψ±,n(z)| ∼CAnn!).

TheBorel sumofψ±(as a formal series ofη1):

S[ψ±]=

a(z)

eyηψ±,B(z,y)dy. Herea(z)=∫z

z0

Q(z0)dz0 and

ψ±,B(z,y)=∑

n=0

ψ±,n(z) Γ(n+12)

(y±a(z))n12 :Borel transformofψ±

• Borel transform=termwise inverse Laplace transform:

(

c.f. η−α=

0

eyηyα−1

Γ(α)dy ifReα >0. )

• If the Borel sumsS[ψ±]are well-defined, they give analytic solutions of the Sch ¨odinger equation andS[ψ±]∼ψ±whenη→+∞.

(19)

Borel resummation method

• Expansion of WKB solution:

ψ±(z, η)=e±η

z z0

Q(z0)dz0 n=0

ηn12ψ±,n(z) (|ψ±,n(z)| ∼CAnn!).

TheBorel sumofψ±(as a formal series ofη1):

S[ψ±]=

a(z)

eyηψ±,B(z,y)dy. Herea(z)=∫z

z0

Q(z0)dz0 and

ψ±,B(z,y)=∑

n=0

ψ±,n(z) Γ(n+12)

(y±a(z))n12 :Borel transformofψ±

• Borel transform=termwise inverse Laplace transform:

(

c.f. η−α=

0

eyηyα−1

Γ(α)dy ifReα >0. )

• If the Borel sumsS[ψ±]are well-defined, they give analytic solutions of the Sch ¨odinger equation andS[ψ±]∼ψ±whenη→+∞.

7 / 22

(20)

Stokes graph and Stokes segent

Stokes graph:

∗ Vertices:turning points(i.e., zeros ofQ(z)) and singular points.

∗ Edges:Stokes curvesemanating from turning points.

(real one-dimensional curves defined byIm∫z

Q(z0)dz0=const.) Stokes curves aretrajectoriesof the quadratic differentialQ(z)dz2.

Q(z)=z. Q(z)=z(z+1)(z+i). Q(z)=(z2)(z3) z2(z1)2 .

Q(z)=1z2.

Stokes segmentis a Stokes curve connecting turning points (= saddle trajectory ofQ(z)dz⊗2).

• Stokes graph is said to besaddle-freeif it doesn’t contain Stokes segments.

(21)

Stokes graph and Stokes segent

Stokes graph:

∗ Vertices:turning points(i.e., zeros ofQ(z)) and singular points.

∗ Edges:Stokes curvesemanating from turning points.

(real one-dimensional curves defined byIm∫z

Q(z0)dz0=const.) Stokes curves aretrajectoriesof the quadratic differentialQ(z)dz2.

Q(z)=z. Q(z)=z(z+1)(z+i). Q(z)=(z2)(z3) z2(z1)2 .

Q(z)=1z2.

Stokes segmentis a Stokes curve connecting turning points (= saddle trajectory ofQ(z)dz⊗2).

• Stokes graph is said to besaddle-freeif it doesn’t contain Stokes segments.

8 / 22

(22)

Stokes graph and Stokes segent

Stokes graph:

∗ Vertices:turning points(i.e., zeros ofQ(z)) and singular points.

∗ Edges:Stokes curvesemanating from turning points.

(real one-dimensional curves defined byIm∫z

Q(z0)dz0=const.) Stokes curves aretrajectoriesof the quadratic differentialQ(z)dz2.

Q(z)=z. Q(z)=z(z+1)(z+i). Q(z)=(z2)(z3) z2(z1)2 .

Stokes segmentis a Stokes curve connecting turning points (= saddle trajectory ofQ(z)dz⊗2).

• Stokes graph is said to besaddle-freeif it doesn’t

(23)

Stokes graph and Borel summability

.

Theorem (Koike-Sch ¨afke)

.

.

.

.. .

. .

Suppose that the Stokes graph issaddle-free. Then,

• ψ±(z, η)areBorel summable(as a formal series ofη1) on each Stokes region(= a face of the Stokes graph).

• The Borel sumsS[ψ±](z, η)giveanalytic(in bothzandη) solutions of the Schr ¨odinger equation on each Stokes region satisfying

S[ψ±](z, η)∼ψ±(z, η) as η→+∞.

9 / 22

(24)

Stokes graph and Borel summability

.

Theorem (Koike-Sch ¨afke)

.

.

.

. .

Suppose that the Stokes graph issaddle-free. Then,

• ψ±(z, η)areBorel summable(as a formal series ofη1) on each Stokes region(= a face of the Stokes graph).

• The Borel sumsS[ψ±](z, η)giveanalytic(in bothzandη) solutions of the Schr ¨odinger equation on each Stokes region satisfying

S[ψ±](z, η)∼ψ±(z, η) as η→+∞.

9 / 22

(25)

Stokes graph and Borel summability

.

Theorem (Koike-Sch ¨afke)

.

.

.

.. .

. .

Suppose that the Stokes graph issaddle-free. Then,

• ψ±(z, η)areBorel summable(as a formal series ofη1) on each Stokes region(= a face of the Stokes graph).

• The Borel sumsS[ψ±](z, η)giveanalytic(in bothzandη) solutions of the Schr ¨odinger equation on each Stokes region satisfying

S[ψ±](z, η)∼ψ±(z, η) as η→+∞.

9 / 22

(26)

Voros symbols

Again suppose that the Stokes graph issaddle-free. Then,

• An explicit connection formula for (Borel resummed) WKB solutions on Stokes curves emanating from a turning point of order 1 ([Voros 83], [Aoki-Kawai-Takei 91]).

• Connection formulas and monodromy matrices of WKB solutions are written by (the Borel sum of)Voros symbolseWβ(η)andeVγ(η), where

Wβ(η)=

β

(Sodd(z, η)−η√ Q(z))

dz, Vγ(η)= I

γSodd(z, η)dz. (c.f., [Kawai-Takei 05,§3]). Here

I S±(z, η)= d

dzlogψ±(z, η)=±η√

Q(z)+· · ·, and

Sodd(z, η)=1

2(S+(z, η)−S(z, η))=η√

Q(z)+· · ·

I β∈H1(R,P;Z)(“path”), γ∈H1(R;Z)(“cycle”). R=Riemann surface of √

Q(z), P=the set of poles ofQ(z).

• Voros symbolseWβ(η)andeVγ(η)(for any pathβand any cycleγ) are Borel summableif the Stokes graph is saddle-free.

(27)

Voros symbols

Again suppose that the Stokes graph issaddle-free. Then,

• An explicit connection formula for (Borel resummed) WKB solutions on Stokes curves emanating from a turning point of order 1 ([Voros 83], [Aoki-Kawai-Takei 91]).

• Connection formulas and monodromy matrices of WKB solutions are written by (the Borel sum of)Voros symbolseWβ(η)andeVγ(η), where

Wβ(η)=

β

(Sodd(z, η)−η√ Q(z))

dz, Vγ(η)= I

γSodd(z, η)dz. (c.f., [Kawai-Takei 05,§3]). Here

I S±(z, η)= d

dzlogψ±(z, η)=±η√

Q(z)+· · ·, and

Sodd(z, η)=1

2(S+(z, η)−S(z, η))=η√

Q(z)+· · ·

I β∈H1(R,P;Z)(“path”), γ∈H1(R;Z)(“cycle”). R=Riemann surface of √

Q(z), P=the set of poles ofQ(z).

• Voros symbolseWβ(η)andeVγ(η)(for any pathβand any cycleγ) are Borel summableif the Stokes graph is saddle-free.

10 / 22

(28)

Voros symbols

Again suppose that the Stokes graph issaddle-free. Then,

• An explicit connection formula for (Borel resummed) WKB solutions on Stokes curves emanating from a turning point of order 1 ([Voros 83], [Aoki-Kawai-Takei 91]).

• Connection formulas and monodromy matrices of WKB solutions are written by (the Borel sum of)Voros symbolseWβ(η)andeVγ(η), where

Wβ(η)=

β

(Sodd(z, η)−η√ Q(z))

dz, Vγ(η)= I

γSodd(z, η)dz. (c.f., [Kawai-Takei 05,§3]). Here

I S±(z, η)= d

dzlogψ±(z, η)=±η√

Q(z)+· · ·, and

Sodd(z, η)=1

2(S+(z, η)−S(z, η))=η√

Q(z)+· · ·

I β∈H1(R,P;Z)(“path”), γ∈H1(R;Z)(“cycle”).

R=Riemann surface of √

Q(z), P=the set of poles ofQ(z).

• Voros symbolseWβ(η)andeVγ(η)(for any pathβand any cycleγ) are Borel summableif the Stokes graph is saddle-free.

(29)

Voros symbols

Again suppose that the Stokes graph issaddle-free. Then,

• An explicit connection formula for (Borel resummed) WKB solutions on Stokes curves emanating from a turning point of order 1 ([Voros 83], [Aoki-Kawai-Takei 91]).

• Connection formulas and monodromy matrices of WKB solutions are written by (the Borel sum of)Voros symbolseWβ(η)andeVγ(η), where

Wβ(η)=

β

(Sodd(z, η)−η√ Q(z))

dz, Vγ(η)= I

γSodd(z, η)dz. (c.f., [Kawai-Takei 05,§3]). Here

I S±(z, η)= d

dzlogψ±(z, η)=±η√

Q(z)+· · ·, and

Sodd(z, η)=1

2(S+(z, η)−S(z, η))=η√

Q(z)+· · ·

I β∈H1(R,P;Z)(“path”), γ∈H1(R;Z)(“cycle”).

R=Riemann surface of √

Q(z), P=the set of poles ofQ(z).

• Voros symbolseWβ(η)andeVγ(η)(for any pathβand any cycleγ) are Borel summableif the Stokes graph is saddle-free.

10 / 22

(30)

Mutation of Stokes graphs

G

G0

G−δ

(The figure describes a part of Stokes graph.)

• Suppose that the Stokes graphG0has aStokes segment.

• Consider theS1-family of the potential: Q(θ)(z)=e2iθQ(z) (θ∈R). Gθ: Stokes graph forQ(θ)(z).

• For any sufficiently smallδ >0,G±δaresaddle-freesince the existence of the Stokes segment implies

along Stokes segment

Q(z)dz∈R,0

S1-action causes a “mutation of Stokes graphs” (= a discontinuous change of topology of Stokes graphs caused by a Stokes segment).

(31)

Mutation of Stokes graphs

G

G0

G−δ

(The figure describes a part of Stokes graph.)

• Suppose that the Stokes graphG0has aStokes segment.

• Consider theS1-family of the potential: Q(θ)(z)=e2iθQ(z) (θ∈R). Gθ: Stokes graph forQ(θ)(z).

• For any sufficiently smallδ >0,G±δaresaddle-freesince the existence of the Stokes segment implies

along Stokes segment

Q(z)dz∈R,0

S1-action causes a “mutation of Stokes graphs” (= a discontinuous change of topology of Stokes graphs caused by a Stokes segment).

11 / 22

(32)

Mutation of Stokes graphs

G G0 G−δ

(The figure describes a part of Stokes graph.)

• Suppose that the Stokes graphG0has aStokes segment.

• Consider theS1-family of the potential: Q(θ)(z)=e2iθQ(z) (θ∈R). Gθ: Stokes graph forQ(θ)(z).

• For any sufficiently smallδ >0,G±δaresaddle-freesince the existence of the Stokes segment implies

along Stokes segment

Q(z)dz∈R,0

S1-action causes a “mutation of Stokes graphs” (= a discontinuous

(33)

DDP’s jump formula of Voros symbols

G G0 G−δ

γ0

• Suppose thatG0has a Stokes segment connecting two distinct turning points, and no other Stokes segments.

LetS[eW

(θ)

β ],S[eVγ(θ)]be the Borel sum of Voros symbols for Q(θ)(z)and S±[eWβ] := lim

θ→±0S[eW

(θ)

β ], S±[eVγ] := lim

θ→±0S[eVγ(θ)].

.

Theorem (Delabaere-Dillinger-Pham 93)

.

.

.

.. .

.

.

S[eWβ]=S+[eWβ](1+S+[eVγ0])−hγ0, βi, S[eVγ]=S+[eVγ](1+S+[eVγ0])−hγ0,γi.

Hereh, iis the intersection form (normalized ashx-axis,y-axisi= +1), and γ0is the cycle around the Stokes segment oriented asH

γ0

Q(z)dz∈R<0.

• This formula describes theStokes phenomenonfor Voros symbols.

12 / 22

(34)

DDP’s jump formula of Voros symbols

G G0 G−δ

γ0

• Suppose thatG0has a Stokes segment connecting two distinct turning points, and no other Stokes segments.

LetS[eW

(θ)

β ],S[eVγ(θ)]be the Borel sum of Voros symbols forQ(θ)(z)and S±[eWβ] := lim

θ→±0S[eW

(θ)

β ], S±[eVγ] := lim

θ→±0S[eVγ(θ)].

.

Theorem (Delabaere-Dillinger-Pham 93)

.

.

.

.. .

.

.

S[eWβ]=S+[eWβ](1+S+[eVγ0])−hγ0, βi, S[eVγ]=S+[eVγ](1+S+[eVγ0])−hγ0,γi.

Hereh, iis the intersection form (normalized ashx-axis,y-axisi= +1), and γ0is the cycle around the Stokes segment oriented asH

γ0

Q(z)dz∈R<0.

• This formula describes theStokes phenomenonfor Voros symbols.

(35)

DDP’s jump formula of Voros symbols

G G0 G−δ

γ0

• Suppose thatG0has a Stokes segment connecting two distinct turning points, and no other Stokes segments.

LetS[eW

(θ)

β ],S[eVγ(θ)]be the Borel sum of Voros symbols forQ(θ)(z)and S±[eWβ] := lim

θ→±0S[eW

(θ)

β ], S±[eVγ] := lim

θ→±0S[eVγ(θ)].

.

Theorem (Delabaere-Dillinger-Pham 93)

.

.

.

.. .

.

.

S[eWβ]=S+[eWβ](1+S+[eVγ0])−hγ0, βi, S[eVγ]=S+[eVγ](1+S+[eVγ0])−hγ0,γi.

Hereh, iis the intersection form (normalized ashx-axis,y-axisi= +1), and γ0is the cycle around the Stokes segment oriented asH

γ0

Q(z)dz∈R<0.

• This formula describes theStokes phenomenonfor Voros symbols.

12 / 22

(36)

DDP’s jump formula of Voros symbols

G G0 G−δ

γ0

• Suppose thatG0has a Stokes segment connecting two distinct turning points, and no other Stokes segments.

LetS[eW

(θ)

β ],S[eVγ(θ)]be the Borel sum of Voros symbols forQ(θ)(z)and S±[eWβ] := lim

θ→±0S[eW

(θ)

β ], S±[eVγ] := lim

θ→±0S[eVγ(θ)].

.

Theorem (Delabaere-Dillinger-Pham 93)

.

.

.

.

.

S[eWβ]=S+[eWβ](1+S+[eVγ0])−hγ0, βi, S[eVγ]=S+[eVγ](1+S+[eVγ0])−hγ0,γi.

Hereh, iis the intersection form (normalized ashx-axis,y-axisi= +1), and γ0is the cycle around the Stokes segment oriented asH

γ0

Q(z)dz∈R<0.

• This formula describes theStokes phenomenonfor Voros symbols.

12 / 22

(37)

Contents

§1

Exact WKB analysis

§2 Main results

Refferences

K. I and T. Nakanishi, “Exact WKB analysis and cluster algebras”, J. Phys. A: Math. Theor. 47 (2014) 474009.

K. I and T. Nakanishi, “Exact WKB analysis and cluster algebras II: simple poles, orbifold points, and generalized cluster algebras”, arXiv:1401.7094.

13 / 22

(38)

Dictionary

Exact WKB analysis Cluster algebras

saddle-free Stokes graph skew-symmetric matirxB mutation of Stokes graphs mutation ofB (Borel sum of) Voros symboleWβi clusterx-variablexi

(Borel sum of) Voros symboleVγi clustery-variableyi eη

H

γi

Q(z)dz

coefficientri

Stokes phenomenon for Voros symbols mutation of cluster variables

Wβ(η)=

β

(Sodd(z, η)−η√ Q(z))

dz, Vγ(η)= I

γSodd(z, η)dz.

b0i j=

bi j i=korj=k

bi j+[bik]+bk j+bik[bk j]+ otherwise.

x0i=



xk1





n

j=1

xj[−εbjk]+



(1+ykε) i=k

xi i,k.

y0i=

yk−1 i=k

yiyk[εbki]+(1+ykε)bki i,k.

(39)

Stokes graph

{

Skew-symmetric matrix

• A saddle-free Stokes graph

{Atriangulated surface: (Three Stokes curve emanate from an order 1 turning point.) [Gaiotto-Moore-Neitzke 09]

• A triangulated surface{Aquiver [Fomin-Shapiro-Thurston 08]:

∗ Put vertices on edges of triangulation.

∗ Draw arrows on each triangle in clockwise direction.

∗ Remove vertices on “boundary edges” together with attached arrows. (boundary / internal edge↔digon-type / rectangular Stokes region)

Stokes graph

{

Triangulated surface {

Quiver

• A quiver{Askew-symmetric matrixB=(bi j)ni,j=1by bi j=(]of arrowsi→ ◦j)−(]of arrowsj → ◦i) (Assign labelsi∈ {1, . . . ,n}to rectangular Stokes regions.)

15 / 22

(40)

Stokes graph

{

Skew-symmetric matrix

• A saddle-free Stokes graph {Atriangulated surface:

(Three Stokes curve emanate from an order 1 turning point.) [Gaiotto-Moore-Neitzke 09]

• A triangulated surface{Aquiver [Fomin-Shapiro-Thurston 08]:

∗ Put vertices on edges of triangulation.

∗ Draw arrows on each triangle in clockwise direction.

∗ Remove vertices on “boundary edges” together with attached arrows. (boundary / internal edge↔digon-type / rectangular Stokes region)

Stokes graph

{

Triangulated surface

{

Quiver

• A quiver{Askew-symmetric matrixB=(bi j)ni,j=1by bi j=(]of arrowsi→ ◦j)−(]of arrowsj → ◦i) (Assign labelsi∈ {1, . . . ,n}to rectangular Stokes regions.)

(41)

Stokes graph

{

Skew-symmetric matrix

• A saddle-free Stokes graph {Atriangulated surface:

(Three Stokes curve emanate from an order 1 turning point.) [Gaiotto-Moore-Neitzke 09]

• A triangulated surface{Aquiver [Fomin-Shapiro-Thurston 08]:

∗ Put vertices on edges of triangulation.

∗ Draw arrows on each triangle in clockwise direction.

∗ Remove vertices on “boundary edges” together with attached arrows.

(boundary / internal edge↔digon-type / rectangular Stokes region)

Stokes graph

{

Triangulated surface {

Quiver

• A quiver{Askew-symmetric matrixB=(bi j)ni,j=1by bi j=(]of arrowsi→ ◦j)−(]of arrowsj → ◦i) (Assign labelsi∈ {1, . . . ,n}to rectangular Stokes regions.)

15 / 22

(42)

Stokes graph

{

Skew-symmetric matrix

• A saddle-free Stokes graph {Atriangulated surface:

(Three Stokes curve emanate from an order 1 turning point.) [Gaiotto-Moore-Neitzke 09]

• A triangulated surface{Aquiver [Fomin-Shapiro-Thurston 08]:

∗ Put vertices on edges of triangulation.

∗ Draw arrows on each triangle in clockwise direction.

∗ Remove vertices on “boundary edges” together with attached arrows.

(boundary / internal edge↔digon-type / rectangular Stokes region)

Stokes graph

{

Triangulated surface {

Quiver

• A quiver{Askew-symmetric matrixB=(bi j)ni,j=1by bi j=(]of arrowsi→ ◦j)−(]of arrowsj → ◦i)

(43)

Muation of Stokes graph and quiver mutation

S1-family of potentials:Q(θ)(z)=e2iθQ(z).

• Mutation of Stokes graph{Quiver mutationatk-th vertex:

(k= label of Stokes region which “degenerates” to a Stokes segment under the mutation of Stokes graph)

G G−δ

←→

µk

(Figures describes a part of Stokes graphs.)

• Quiver muation is compatible withmutationofB-matix: b0i j=

−bi j i=kor j=k

bi j+[bik]+bk j+bik[bk j]+ otherwise. ([a]+=max(a,0))

16 / 22

参照

関連したドキュメント

Reductive Takiff Lie Algebras and their Representations The attentive reader may have noticed that we stated and proved the stronger inequality (9.9) only for the Z 2 -gradings of

If Φ is a finite root system, and if we take H 0 to be the category of representations of an alternating quiver corresponding to Φ , then our generalized cluster complex is the

Key words and phrases: rooted trees, Lie-admissable algebras, right-symmetric algebras, Novikov algebras, vector fields algebras, identities, free basis..  c 2002, Askar

In this paper we develop an elementary formula for a family of elements {˜ x[a]} a∈ Z n of the upper cluster algebra for any fixed initial seed Σ.. This family of elements

In this paper, we take some initial steps towards illuminating the (hypothetical) p-adic local Langlands functoriality principle relating Galois representations of a p-adic field L

We then prove the existence of a long exact sequence involving the cohomology groups of a k-graph and a crossed product graph.. We finish with recalling the twisted k-graph C

This allows us to study effectively the tensor product construction for type II matrices, and a number of examples: character tables of abelian groups, Hadamard matrices of size

• characters of all irreducible highest weight representations of principal W-algebras W k (g, f prin ) ([T.A. ’07]), which in particular proves the conjecture of