D-branes and Closed String Field Theory
Koichi Murakami (KEK) This talk is based on Yutaka Baba, Nobuyuki Ishibashi and K.M., JHEP05(2006)029 [hep-th/0603152],
and a work in preparation (talk at KEK Theory Workshop 2007, March 14, 2007)
D-branes and Closed String Field Theory Koichi Murakami (KEK) This - - PowerPoint PPT Presentation
D-branes and Closed String Field Theory Koichi Murakami (KEK) This talk is based on Yutaka Baba, Nobuyuki Ishibashi and K.M., JHEP05(2006)029 [hep-th/0603152] , and a work in preparation (talk at KEK Theory Workshop 2007, March 14, 2007) 1.
Koichi Murakami (KEK) This talk is based on Yutaka Baba, Nobuyuki Ishibashi and K.M., JHEP05(2006)029 [hep-th/0603152],
and a work in preparation (talk at KEK Theory Workshop 2007, March 14, 2007)
1
I Motivation: How can we capture the solitonic nature of D-branes in string theory?
⇒ Non-perturbative formulation of string should be necessary. In this work, we focus on a traditional approach: String Field Theory (SFT)
(Sen’s conjecture, VSFT, analytic solutions, etc...)
“What are D-branes in closed SFT ?” 2 I Motivation: How can we capture the solitonic nature of D-branes in string theory?
⇒ Non-perturbative formulation of string should be necessary. In this work, we focus on a traditional approach: String Field Theory (SFT)
(Sen’s conjecture, VSFT, analytic solutions, etc...)
“What are D-branes in closed SFT ?” 2
I Boundary state |Bi
|Bi ⇒ make a single hole in a closed-string worldsheet D-brane ⇒ make an arbitrary number of holes in a closed-string worldsheet ⇒ One may guess that the D-brane state in the second-quantization looks like ∼ exp [|Bi] ??? ⇒ We will see that this is the case in a certain sense. (Of course, we will make this (meaningless) expression into a precise form later.) 3
I Clues to D-branes in closed SFT (i): D-branes in SFT for noncritical strings
H = Z ∞ dl µ 3δ00(l) − 3 4μδ(l) ¶ ψ†(l) + Z ∞ dl1dl2 £ (l1 + l2)ψ†(l1 + l2)ψ(l1)ψ(l2) + g2
sl1l2ψ†(l1)ψ†(l2)ψ(l1 + l2)
¤
Z ∞ dlψ(l) (lT(l) + ρ(l)) = Z ∞ dlψ†(l)D(l)
[ψ(l) , ψ†(l0)] = δ(l − l0) ,
Z l dl0 ψ(l0)ψ(l − l0) + g2
s
Z ∞ dl0 l0ψ†(l0)ψ(l + l0) , ρ(l) = 3δ00(l) − 3 4μδ(l)
4
⇒ loop equation: hw(l1) · · · w(ln)i = h0|ψ(l1) · · · ψ(ln)|Ψi
= the operators which commute with D(l), i.e. T(l) (⇐ Virasoro constraint)
(Fukuma-Yahikozawa, Hanada-Hayakawa-Ishibashi-Kawai-Kuroki-Matsuo-Tada)
Z dζV±(ζ) = Z dζ exp µ ±gs Z ∞ dle−ζlψ†(l) ¶ exp µ ∓ 2 gs Z ∞ dl l eζlψ(l) ¶ ⇒ Z dζV±(ζ) is a creation op of ½ D-instanton ghost D-brane (Okuda-Takayanagi) |Ψi(n + 1)-inst. = C Z dζV+(ζ) |Ψin-inst. (n ≥ 0)
We will show that a similar construction of solitonic operators in SFT for critical strings is possible in the OSp invariant SFT. 5
I Clues to D-branes in closed SFT (ii): boundary state in SFT for critical strings
type closed SFT.
|Bi ∗ |Bi ∝ |Bi 6
I What we did
⇐ states in which D-branes (or ghost D-branes) are excited
⇒ the vacuum cylinder amplitude (×4) for the D-branes is reproduced.
Our solitonic ops create two D-branes (or ghost D-branes) and not a single. (← We do not know why...)
string tachyons. For this purpose, ¤ define BRST invariant observables in the OSp invariant SFT ¤ Green’s functions, S-matrix As a bonus, determine the sign ambiguity between D-brane and ghost D-brane 7
Plan of the talk §1. Introduction §2. OSp invariant SFT §3. Boundary state and Solitonic operators §4. Observables §5. Green’s functions and S-matrix elements §6. Disk amplitude §7. Summary and Discussions 8
9
I Procedure for covariantizing the LC gauge SFT : OSp extension (Siegel) LC gauge SFT OSp invariant SFT (t = X+, α = 2p+, Xi)
(i = 1, . . . , 24)
− → OSp extension
(t, α, XM) XM = (Xi, X25, X26; C, ¯ C) linear symmetry O(24) OSp(26|2) S-matrix symmetry O(25,1) OSp(27, 1|2)
C(τ, σ): spinless Grassmann odd variables (ghost) ( C(τ, σ) = C0 + 2iπ0τ − i P
n6=0 1 n
¡ γne−n(τ+iσ) + ˜ γne−n(τ−iσ)¢ ¯ C(τ, σ) = ¯ C0 − 2i¯ π0τ + i P
n6=0 1 n
¡ ¯ γne−n(τ+iσ) + ˜ ¯ γne−n(τ−iσ)¢
“transverse directions”: δij − → ηMN =
C ¯ C C ¯ C
⎛ ⎜ ⎜ ⎜ ⎜ ⎝ δμν −i i ⎞ ⎟ ⎟ ⎟ ⎟ ⎠ 10
Thus we may say: OSp inv. SFT is LC SFT with XM = (Xi, X25, X26; C, ¯ C) as transverse directions . I String field action S = Z dt " 1 2 Z d1d2 hR(1, 2) |Φi1 Ã i ∂ ∂t − L(2) + ˜ L(2) − 2 α2 ! |Φi2 + 2g 3 Z d1d2d3 V 0
3 (1, 2, 3)
¯ ¯ Φi1|Φi2|Φi3 ¸
⇒ momentum zero-mode integration: dr = αrdαr 2 d26pr (2π)26 id¯ π(r)
0 dπ(r)
11
I Relationship between S-matrix in LC SFT and S-matrix in OSp invariant SFT
SLC ¡ pLC, ²LC¢ = δLC(pLC) F ¡ pLC
r
· pLC
s
, pLC
r
· ²LC
s
, ²LC
r
· ²LC
s
¢ pLC
r
= (kr,0, kr,1, pr,i) (k+ = 2α) ²LC
r : polarizations
SOSp ¡ pOSp, ²OSp¢ = δOSp(pOSp) F ¡ pOSp
r
· pOSp
s
, pOSp
r
· ²OSp
s
, ²OSp
r
· ²OSp
s
¢ pOSp
r
= (pLC
r
; pr,25, pr,26 ; π(r)
0 , ¯
π(r)
0 ) , ²OSp r
= (²LC
r
; ²r,25, ²r,26 ; ²r,C, ²r, ¯
C)
Because of the OSp(27, 1|2) invariance, the function F is common to the two theories.
Using Pairisi-Sourlas formula, for the S-matrix elements of the external states with ²OSp
∓
= ²OSp
C
= ²OSp
¯ C
= 0, we have (after k0 → k2 = −ik0, p26 → p0 = ip26), Z Y
r
∙dkr,1dk2 2π id¯ π(r)
0 dπ(r)
¸ SOSp(pOSp, ²OSp) µ = SOSp(pOSp, ²OSp) ¯ ¯ ¯ ¯ ¶ = SLC(pμ, ²μ) . 12
I BRST symmetry = J−C of the OSp(27, 1|2),
(Siegel-Zwiebach, Bengtsson-Linden)
δBΦ = QBΦ + gΦ ∗ Φ
QB = C0 2α(L0 + ˜ L0 − 2) − iπ0 ∂ ∂α + i α
∞
X
n=1
à γ−nLn − L−nγn n + ˜ γ−n ˜ Ln − ˜ L−n˜ γn n !
Ln, ˜ Ln: Virasoro generators Ln ≡ 1 2 X
m
n+mαN −mηMN
˜ Ln ≡ 1 2 X
m
αM
n+m ˜
αN
−mηMN
4 pμpμ +
∞
X
m=1
αμ
−mαmμ + iπ0¯
π0 + i
∞
X
m=1
(γ−m¯ γm − ¯ γ−mγm) , ˜ L0 = . . .
13
14
I Boundary state in OSp invariant theory
(as usual)
C(0, σ)|B0i = 0 I def |B0(l)i = |B0iδ(α − l) ⇒ α|B0(l)i = l|B0(l)i I Regularization |B0(l)i → |B0(l)i² = e− L0+ ˜
L0−2 |α|
|B0(l)i
² α ³ L0 + ˜ L0 − 2 ´ = {QB , 2²¯ π0}
²hB0(l)|B0(l0)i² = π e
π2|l| ²
l0δ(l + l0)(1 + O(e− 1
² )
I def |n(l)i = |B0(l)i² e− π2
2² |l|
I Remark: Naively
R hB0(l)|B0(l0)i = 0, due to the lack of π0 and ¯
π0. However, for |B0(l)i² these zero modes are supplied by e−
² |α| (L0+˜
L0−2).
15
I An expansion of the string field |Φi = |ψi + | ¯ ψi = Z ∞ dl £ |n(l)iφ(l) + |n(−l)i¯ φ(l) + (orthogonal complement) ¤
h |ψir , | ¯ ψis i = |R(r, s)i ⇒ |ψi: α > 0 mode → annihilation ; | ¯ ψi: α < 0 mode → creation of a string ⇒ This yields h φ(l) , ¯ φ(l0) i = 1 πlδ(l − l0)
i = h h0|h ¯ ψ| = 0, h h0|0i i = 1 I similarity of the above part and SFT for c = 0 noncritical strings critical string noncritical string φ(l) ∼ r 2 π 1 gslψ(l) ¯ φ(l) ∼ gs √ 2π ψ†(l) BRST invariance ∼ Virasoro constraints This similar structure suggests ⇒ we may be able to construct solitonic
16
I Solitonic state |D±i i ≡ Z dζV±(ζ)|0i i V±(ζ) = λ exp ∙ ± √ 2π Z ∞ dl e−ζl ¯ φ(l) ¸ exp ∙ ∓ √ 2π Z ∞ dl0 eζl0φ(l0) ¸ e
±
C²2 √ 2π g
³ ζ+ π2
2²
´2
i is BRST invariant: δB|D±i i = 0
∙µZ dζV(ζ) ¶n |0i i ¸ = 0 (n = 1, 2, . . . )
In the ζ integration, contribution at ζ ∼ − π2
2² is dominant. ⇒ saddle pt approx.
|D±i i ' λ0 exp ∙ ± √ 2π Z ∞ dl e
π2 2² l ¯
φ(l) ¸ |0i i = λ0 exp " ± Z ∞ dl l r 2 π Z h ¯ ψ|B0(l)i² # |0i i ⇐ what we guessed!!
( |D+i i for g > 0 |D−i i for g < 0 ⇐ “unstable” 17
I Vacuum amplitude h hD±|e−iT ˆ
H|D±i
i ' exp h 22 × (cylinder amplitude for a single D-brane) i
H is BRST exact
i should be identified with the state in which two D-branes or ghost D-branes are excited 18
19
I on-shell physical states
à i ∂ ∂t − L0 + ˜ L0 − 2 α ! | i = 0 , physical: QB| i = 0 Hamiltonian L0+˜
L0−2 α
is QB-exact ⇒ We may set t = 0 in the QB-cohomology I Relationship between QB in OSp invariant theory and Kato-Ogawa’s QKO
B
no counter part in (b.c) system C0 = 2αc+ ¯ π0 =
1 2αb+
¯ C0 , π0 , α γn = inαcn ¯ γn = 1
αbn
˜ γn = inα˜ cn ˜ ¯ γn = 1
α˜
bn = ⇒ QB = QKO
B
− iπ0 µ ∂ ∂α + 1 α ¶ I QB cohomology: Solve QB| i = 0 ⇒ | i = 1 α|physi + ¯ π0|physi + g1(α)π0|physi + g2(α)π0¯ π0|physi |physi: QKO
B -physical ,
g1(α), g2(α): ∀ function of α 20
I boundary condition for α direction (C, ¯ C-representation) For the bra or ket states, 0 < |α| < ∞ ⇒ introduce ω s.t. α = eω, then −∞ < ω < ∞ ⇒ integration measure: Z ∞ αdα = Z ∞
−∞
dωe2ω ⇒ The wave functions should be expand with respect to exp (−ω + iωxω) ⇒ • the wave functions are delta-function normalizable
etc... We should focus on the state | i = 1 α|physi + 1 απ0¯ π0|physi %
physical external states I Observable associated with |primaryiX = |primaryiX(2π)26δ26(pμ − kμ) O(t, k) = i 2 Z ∞
−∞
dα Z d¯ π0dπ0 C ¯
Ch0| ⊗ Xhprimary|Φ(t, α, π0, ¯
π0, k)i 21
22
I Two-point Green’s function for Or(tr) (r = 1, 2): |primary1iX = |primary2iX with
Xhprimary|primaryiX = 1 ;
mass M D D ˜ O1(E1) ˜ O2(E2) E E ≡ Z dt1dt2eiE1t1+iE2t2h h0|TO1(t1)O2(t2)|0i i =
2
Y
r=1
µ i 2 Z dαrd¯ π(r)
0 dπ(r)
¶ iδ(1, 2)2πδ(E1 + E2) α1E1 − p2
1 − M 2 − 2iπ(1) 0 ¯
π(1) + i² . . .
δ(1, 2) ≡ 2δ(α1 + α2)(2π)26δ26(p1 + p2)i(¯ π(1) + ¯ π(2)
0 )(π(1)
+ π(2)
0 )
= 2πδ(E1) 2πδ(E2) p2
1 + M 2
(2π)26δ26(p1 + p2) I def Ofr
r ≡
Z dtrfr(tr)Or(tr) with Z dtrfr(tr) = 1 , then D D Of1
1 Of2 2
E E = 1 p2
1 + M 2 (2π)26δ26(p1 + p2)
⇐ Euclidean propagator for a particle of mass M with correct normalization 23
I N-point Green’s functions ¿ ¿ N Y
r=1
˜ Or(Er) À À ≡
N
Y
r=1
µZ dtreiErtr ¶ h h0|T
N
Y
r=1
Or(tr)|0i i ≡
N
Y
r=1
à i 2 Z dαrd¯ π(r)
0 dπ(r)
i αrEr − p2
r − M 2 r − 2iπ(r) 0 ¯
π(r) ! ×δOSp µ N X
s=1
pOSp
s
¶ Gtranc.(pOSp
1
, . . . , pOSp
N
) (For p2
r + M 2 r ∼ 0)
∼ −i à N Y
r=1
2πδ(Er) p2
r + M 2 r
! (2π)26δ26 µ N X
r=1
pr ¶ Gtranc.(pOSp
r
) ¯ ¯ ¯ ¯ µ ¯ ¯ ¯ ⇔ set p2
r + M 2 r = Er = αr = π(r)
= ¯ π(r) = 0 ¶ Hence ¿ ¿ N Y
r=1
˜ Ofr
r
À À ∼ −i à N Y
r=1
1 p2
r + M 2 r
! (2π)26δ26 µ N X
r=1
pr ¶ Gtranc.(pOSp
r
) ¯ ¯ ¯ ¯ ⇐ 26 dim. Euclidean Green’s function = ⇒ LC SFT’s S-matrix elements 24
25
I Green’s function for two closed string tachyons in the presence of the soliton ¿ ¿ ˜ OT
1 (E1) ˜
OT
2 (E2)
À À
D±
≡ h hD±| ˜ OT
1 (E1) ˜
OT
2 (E2)|D±i
i h hD±|D±i i where OT : tachyon ⇔ |primaryiX = |0iX
ST T D± = ±2ig r 2 π s (2π)26 (8π2)p+1 Y
μ∈N
(2π)δ(p1,μ + p2,μ) B µ t 2 − 1, 2s − 1 ¶ s = X
μ∈N
p1,μpμ
1 =
X
μ∈N
p2,μpμ
2 ,
t =
26
X
μ=1
(p1 + p2)μ(p1 + p2)μ 26
ST T D± ∼ ∓2iDμν √ 2π s (2π)26 (8π2)p+1 −i (p1 + p2)2 ig(p1 − p2)μ(p1 − p2)ν
%
tachyon-tachyon-graviton Dμν = ( ημν for μ, ν ∈ N −δμν for μ, ν ∈ D In our convention, Gμν(x) ∼ ημν + 4ghμνeip·x. The above result means that ( |D+i i for g > 0 |D−i i for g < 0 ⇐ should be identified with the ordinary D-brane. Otherwise ghost D-brane. 27
28
¥ D-brane ⇔ BRST invariant state in OSp invariant SFT looks like bosonization ¥ D-brane or ghost D-brane D-brane state is unstable ⇐ consistent with the instability of D-brane in bosonic string theory ¥ How to include open string degrees of freedom associated with D-brane? ¥ similar construction in superstrings? ¤ single D-brane state? why two D-brane state?? etc..... 29