+ All Categories
Home > Documents > Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1:...

Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1:...

Date post: 30-Jan-2018
Category:
Upload: votuong
View: 221 times
Download: 0 times
Share this document with a friend
28
Sec1: Vir Sec2: 2-dim subalg Sec3: The set V (r) 1o3IoŒ˘ Finite-dimensional subalgebras of the Virasoro algebra ~u * uHn˘Œ˘˘ 2015c0720F * Email: [email protected] ¥I22²U«˚·381uHn˘Œ˘˘4¢§510640 1 / 26
Transcript
Page 1: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

1�o3�Io�ê¬Æ

Finite-dimensional subalgebrasof the Virasoro algebra

~�u∗

uHnó�ÆêÆÆ�

2015c07�20F

∗Email: [email protected]¥I2À�2²½Uà«Êì´381ÒuHnó�ÆêÆÆ�4Ò¢§510640

1 / 26

Page 2: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Contents

A quick review on the centreless Virasoro algebra

A description on two-dimensional subalgebras of d

Properties of the parameters

2 / 26

Page 3: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

The centreless Virasoro algebra

The centreless Virasoro algebra (Witt algebra) d:

is the derivation Lie algebra of the Laurent polynomial algebra.

d = DerC(C[t±1]).

I d has a standard basis:{Lm := tm+1 d

dt

∣∣∣∣m ∈ Z}

satisfying

[Lm,Ln] = (n −m)Lm+n, for m,n ∈ Z.

3 / 26

Page 4: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

The Virasoro algebra

The Virasoro algebra Vir

is the universal central extension of d.

Facts

I d is an infinite-dimensional simple Lie algebra.

I d has no finite-dimensional subalgebra of dimension > 4.

I d has no commutative subalgebra of dimension > 2.

S. Ng, E. J. Taft, Classification of the Lie bialgebra structures on the Witt andVirasoro algebras. J. Pure Appl. Algebra 151 (2000): 67õ88.

Y. Su, K. Zhao, Generalized Virasoro and super-Virasoro algebras and modulesof the intermediate series. J. Algebra, 252(2002):1–19.

4 / 26

Page 5: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Three dimensional subalgebras of d

Proposition

Every three dimensional subalgebra of d is of the form

spanC{L−m,L0,Lm}

for some positive integer m.

Y. Su, K. Zhao, Generalized Virasoro and super-Virasoro algebras and modulesof the intermediate series. J. Algebra, 252(2002):1–19.

5 / 26

Page 6: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Two dimensional subalgebras of d

Every two-dimensional subalgebra of d

I is non-commutative,and hence

I has a basis {X ,Y} such that

[X ,Y ] = cY .

Questions

I How do such X and Y look like?

I Can we find all two-dimensional subalgebras of d?

6 / 26

Page 7: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Known results

I Trivial examples of two-dimensional Lie subalgebras of d

z(m) := spanC{L0,Lm},

where m is a non-zero integer.I Non-trivial examples obtained by Y. Su and Z. Zhao:

Y. Su, K. Zhao, Generalized Virasoro and super-Virasoro algebras andmodules of the intermediate series. J. Algebra, 252(2002):1–19.

X := L0 + αL−m, and Y := exp(αm · adL−m

)Lnm

span a two dimensional subalgebra of d.

7 / 26

Page 8: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Known results

I Trivial examples of two-dimensional Lie subalgebras of d

z(m) := spanC{L0,Lm},

where m is a non-zero integer.I Non-trivial examples obtained by Y. Su and Z. Zhao:

Y. Su, K. Zhao, Generalized Virasoro and super-Virasoro algebras andmodules of the intermediate series. J. Algebra, 252(2002):1–19.

X := L0 + αL−m, and Y := exp(αm · adL−m

)Lnm

span a two dimensional subalgebra of d.2015

-07-

17 A quick review on the centreless Virasoro alge-

bra

Known results

In Su and Zhao’s example,

X = L0 + αL−m = t−m+1(tm + α)ddt,

Y = exp( α

m· adL−m

)Lnm

=n+1∑k=0

(n + 1

k

)αk Lnm−km

=

(n+1∑k=0

(n + 1

k

)αk tnm−km+1

)ddt

= t−m+1(tm + α)n+1 ddt.

Page 9: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Known results

D. Yu, C. Lu, Results for Virasoro subalgebra. Acta Math. Sinica (Chin. Ser.),49(2006):632–638.

I If X = a−mL−m + · · ·+ anLn and Y = b−mL−m + · · ·+ bnLn

satisfy [X ,Y ] = Y , then there coefficients a−m, . . . ,an,

b−m, . . . ,bn satisfy certain equation system.

I One expects to obtain two dimensional subalgebras of d

through solving these equations, but it is very difficult.I A few examples are given, for instance,

X =5

768L−2 +

548

L−1 +18

L0 − L1 − L2,

Y =1

256L−2 +

116

L−1 +38

L0 + L1 + L2,

satisfy [X ,Y ] = Y .

8 / 26

Page 10: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Known results

D. Yu, C. Lu, Results for Virasoro subalgebra. Acta Math. Sinica (Chin. Ser.),49(2006):632–638.

I If X = a−mL−m + · · ·+ anLn and Y = b−mL−m + · · ·+ bnLn

satisfy [X ,Y ] = Y , then there coefficients a−m, . . . ,an,

b−m, . . . ,bn satisfy certain equation system.

I One expects to obtain two dimensional subalgebras of d

through solving these equations, but it is very difficult.I A few examples are given, for instance,

X =5

768L−2 +

548

L−1 +18

L0 − L1 − L2,

Y =1

256L−2 +

116

L−1 +38

L0 + L1 + L2,

satisfy [X ,Y ] = Y .

2015

-07-

17 A quick review on the centreless Virasoro alge-

bra

Known results

In Yu and Lu’s example,

X =

(5

768t−1 +

548

+18

t − t2 − t3)

ddt,

Y =

(1

256t−1 +

116

+38

t + t2 + t3)

ddt.

Then we can deduce that

X − 53

Y = −16

t(16t2 + 16t + 3

) ddt

= − 196

t(

t +14

)(t +

34

)ddt,

Y = t−1(

t +14

)4 ddt.

Page 11: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Contents

A quick review on the centreless Virasoro algebra

A description on two-dimensional subalgebras of d

Properties of the parameters

9 / 26

Page 12: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Main Idea

I Find all solutions to the equation

[X ,Y ] = Y

in the Lie algebra d.

I Write X = tF (t) ddt and Y = tG(t) d

dt .

I Equivalently, find all solutions to the ODE:

tF (t)G′(t)− tG(t)F ′(t) = G(t)

in the Laurent polynomial algebra C[t±1].

10 / 26

Page 13: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

An attempt using ODE

I The ODE

tF (t)G′(t)− tG(t)F ′(t) = G(t)

is separable:G′(t)G(t)

=1 + tF ′(t)

tF (t).

I Given arbitrary F (t) ∈ C[t±1], one can easily obtain a G(t)

G(t) = exp(∫ t

a

1 + uF ′(u)

uF (u)du)

as a complex function.

11 / 26

Page 14: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Problems of the ODE method

For a F (t) ∈ C[t±1],

G(t) = exp(∫ t

a

1 + uF ′(u)

uF (u)du)

is not necessarily a Laurent polynomial.

For instance,

I If F (t) = 1, then G(t) = ta is a Laurent polynomial.

I If F (t) = t , then G(t) = ta + exp

(1a −

1t

)is not a Laurent

polynomial.

12 / 26

Page 15: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Two dimensional subalgebras of d: new examples

LetX := t(t − a1) · (t − an)

ddt

and Y := t−r1−...−rk+n−k+1(t − a1)r1+1 · · · (t − an)

rk+1 ddt,

where n > k , a1, . . . ,an ∈ C are distinct non-zero numbers,

and r1, . . . , rk are positive integers.

Then, X and Y span a two-dimensional subalgebra of dif (a1, . . . ,an) satisfies

r1 · a1 + · · ·+ rk · ak = ak+1 + · · ·+ an,

r1 · a21 + · · ·+ rk · a2

k = a2k+1 + · · ·+ a2

n,

· · ·

r1 · an−11 + · · ·+ rk · an−1

k = an−1k+1 + · · ·+ an−1

n .

Moreover, such an equation system has a componentwise

non-zero solution only ifr1 + · · ·+ rk > n.

13 / 26

Page 16: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Notation

I For n > k ∈ N, let

Γ(n, k) := {(r1, . . . , rn) ∈ Nk × {−1}n−k |r1 + · · ·+ rn > k}.

I For r = (r1, . . . , rn) ∈ Γ(n, k), we denote

V (r)× :=

(a1, . . . ,an) ∈ (C×)n

∣∣∣∣∣∣n∑

j=1

rjaij = 0, for i = 1, . . . ,n − 1

.

I Σ := {(n, k , r,a)|n > k , r ∈ Γ(n, k),a ∈ V (r)}.

14 / 26

Page 17: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Main Result I

Proposition

I Every quadruple µ := (n, k , r,a) ∈ Σ determinesa two-dimensional subalgebra of d:

s(µ) = spanC

{t(t − a1) · · · (t − an)

ddt, t−r1−···−rn+1(t − a1)

r1+1 · · · (t − an)rn+1 d

dt

}.

I s(µ) is not equal to any z(m) = spanC{L0,Lm}.

I Given µ = (n, k , r,a) and µ′ = (n′, k ′, r′,a′) ∈ Σ,

s(µ) = s(µ′) iff n = n′, k = k ′ and there is a permutation

σ ∈ Sn such that rσ(i) = r ′i and aσ(i) = a′i for i = 1, . . . ,n.

15 / 26

Page 18: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Concrete Examples I

I n = k = 1, r = r , then V (r)× = C×.

We obtain two-dimensional subalgebras:

spanC

{t(t − a)

ddt, t−r+1(t − a)r+1 d

dt

}.

I n = k = 2, r = (r1, r2) ∈ N2, then V (r)× = {(ar2,−ar1)|a ∈ C×}.We obtain two-dimensional subalgebras:

spanC

{t(t − ar2)(t + ar1)

ddt, t−r1−r2+1(t − ar2)r1+1(t + ar1)r2+1 d

dt

}.

16 / 26

Page 19: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Concrete Examples II

I n = k ∈ N and r = (r , . . . , r) ∈ Γ(n, k).

Let ζn be a primitive n-th root of unity,

then (ζn, ζ2n , . . . , ζ

nn ) ∈ V (r)×.

We obtain two-dimensional subalgebra

spanC

{t(tn − 1)

ddt, t−rn+1(tn − 1)r+1 d

dt

}.

I n = 3, k = 2, r = (r ,1,−1) with r > 2.

Then V (r)× = {a(2,1− r ,1 + r)|a ∈ C×}.We obtain two-dimensional subalgebras

spanC

{t(t − 2)(t − 1 + r)(t − 1− r)

ddt, t−r+1(t − 2)r+1(t − 1 + r)2 d

dt

}.

17 / 26

Page 20: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Main Result II

TheoremEvery two-dimensional subalgebra of d is equal to

either z(m) = span{L0,Lm} for some non-zero integer m

or s(µ) for some µ ∈ Σ.

18 / 26

Page 21: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Sketch of Proof I

I For X = (tα1 + ·+ tαs )t ddt with α1 < · · · < αs, we write

deg1 X = αs, and deg2 X = α1.

I If [X ,Y ] = cY then [X − aY ,Y ] = cY for any a ∈ C.

Hence, we may assume deg2 X 6= deg2 Y .

I [X ,Y ] = cY and deg2 X 6= deg2 Y implies that

deg2 X = 0 and deg1 X = deg1 Y .

19 / 26

Page 22: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Sketch of Proof II

I We may assume both X and Y are monic, and then write

X = tF (t)ddt

and Y = ts+1G(t)ddt,

where F (t),G(t) ∈ C[t ], F (0) 6= 0 and G(0) 6= 0.

I [X ,Y ] = cY implies thatI every root of G(t) is a root of F (t);I F (t) has no multiple root;I G(t) has no simple root.

20 / 26

Page 23: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Sketch of Proof III

I We may further write

X = t(t−a1) · · · (t−an)ddt, and Y = ts+1(t−a1)r1+1 · · · (t−ak )rk+1,

and denote rk+1 = · · · = rn = 0.

I deg1 X = deg1 Y implies that s = −(r1 + · · ·+ rn).

I Consider the automorphism of d:

ω : t l+1 ddt7→ −t−l+1 d

dt.

ω(X ) and ω(Y ) also span a two-dimensional subalgebra of

d, from which we obtain

r1 + · · ·+ rn > k .

I Using [X ,Y ] = cY again, we deduce (a1, . . . ,an) ∈ V (r)×.

21 / 26

Page 24: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Contents

A quick review on the centreless Virasoro algebra

A description on two-dimensional subalgebras of d

Properties of the parameters

22 / 26

Page 25: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

The algebraic set V (r)

Our description of two-dimensional subalgebras of d depends

on componentwise nonzero points in V (r), where V (r) is an

affine algebraic set defined by

r1x1 + · · ·+ rnxn = 0,

r1x21 + · · ·+ rnx2

n = 0,

· · ·

r1xn−11 + · · ·+ rnxn−1

n = 0.

(∗)

QuestionCan we find all componentwise nonzero solutions to (*)?

23 / 26

Page 26: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

The projective variety V (r)

Observing that (*) is homogeneous and hence determines a

projective variety V (r).

Results on V (r):

I Let a be a componentwise nonzero solution to (*) and a its

canonical image in V (r). Then a has multiplicity 1 in V (r).

I The subset V (r)× of V (r) consisting of all componentwise

nonzero elements contains at most (n − 1)! elements.

I Let r = (r1, . . . , rk ,−1, . . . ,−1). If

ri > n − k + 1, i = 1, . . . , k ,

then V (r)× contains exactly (n − 1)! elements.24 / 26

Page 27: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

A conjecture

Conjecture

Let r = (r1, . . . , rk ,−1, . . . ,−1) ∈ Nk × {−1}n−k . If

r1 + · · ·+ rk > n,

then V (r)× is non-empty, i.e., the equation system

r1x1 + · · ·+ rnxn = 0,

r1x21 + · · ·+ rnx2

n = 0,

· · ·

r1xn−11 + · · ·+ rnxn−1

n = 0.

(∗)

has a componentwise nonzero solution.

25 / 26

Page 28: Sec1: Finite-dimensional subalgebras of the Virasoro …zhchang/Research/PT20150720P.pdf · Sec1: Vir Sec2: 2-dim subalg Sec3: The set V(r) 1ło3˝IofiŒ‹˘ Finite-dimensional

Sec1:Vir

Sec2:2-dimsubalg

Sec3:The setV (r)

Thank you!

Questions?

26 / 26


Recommended