0% found this document useful (0 votes)
6 views

STR 04

Uploaded by

2078238514
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
6 views

STR 04

Uploaded by

2078238514
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

Chapter 4

The String Spectrum

Following “String Theory” by J. Polchinski, Vol.I.


Notes written by students (work still in progress).
For more information contact George Siopsis
[email protected]

I. BRST quantization
We have a number of different types of quantization schemes which have been proposed over the decades, each
with their own merits and drawbacks.
1) The canonical quantization method:
At the level of QED, the canonical quantization method is not too difficult, but the canonical quantization of
more complicated theories, such as non-Abelian theories, is often prohibitively tedious.
2) The Gupta-Bleuler or covariant quantization method:
Contrary to canonical quantization, it maintains full Lorentz symmetry, which is a great advantage. The
disadvantage of this approach is that ghosts or unphysical states of negative norm are allowed to propagete in
the theory.
3) The path integral method:
This is perhaps the most elegant and powerful of all quantization programs. One advantage is that one can easily
go back and forth between many of the other quantization programs to see the relationships between them. The
disadvantage of the path integral approach is that functional integration is a mathematically delicate operation
that may not even exist in Minkowski space.
4) The Becchi-Rouet-Stora-Tyupin(BRST) quantization method:
This is one of the most convenient and practical covariant approaches used for gauge theories. Like the covariant
quantization program, negative norm states or ghosts are allowed to propagate and are eliminated by applying
the BRST condition onto the state vectors. All the information is contained in a single operator, making this
a very attractive formalism.
We will especially discuss the BRST quantization method in this chapter.

II. BRST quantization of the point particle


The action for a point particle of rest mass m is
Z  
1 1 µ
S= dτ Ẋ Ẋµ − ηm2 . (4.1)
2 η

Using the equations of motion


∂L 1 ∂L
Pµ = = Ẋµ , Pη = =0 (4.2)
∂ Ẋ µ η ∂ η̇

1
2 CHAPTER 4. THE STRING SPECTRUM

where η is not dynamic variable, the action becomes


Z
1  
S = dτ Pµ Ẋ µ − ηm2
2
Z  
µ 1 µ 1 2
= dτ Pµ Ẋ − Pµ Ẋ − ηm
2 2
Z h η i
= dτ Pµ Ẋ µ − (Pµ P µ + m2 )
2
Z
= dτ (Pµ Ẋ µ − ηχ) (4.3)

where χ = 12 (Pµ P µ + m2 ).
Now, we consider the following transformation

X −→ X + δX, P −→ P + δP.

From the quantization condition [X µ , P ν ] = iη µν , we obtain

δX µ = i[X µ , χ]
i µ
= [X , Pν P ν + m2 ]
2
i
= (Pν [X µ , P ν ] + [X µ , Pν ]P ν )
2
i
= (iPν η µν + iηνµ P ν )
2
= −P µ (4.4)
δPµ = i[Pµ , χ]
i
= [Pµ , P ν Pν ]
2
= 0 (4.5)

By giving dynamics to anticommuting variable  like

 −→ −c(τ )

we can obtain the BRST transformation;

δB X µ = cP µ
δB P µ = 0
δB b = −χ
δB c = 0 (4.6)

where new coordinate c and new momentum b are anticommuting fields.


Under this BRST transformation (4.6), the action can be invariant by adding the additional term associated
with new fields b and c to the old action. That is, the new action can be written as

S 0 = S + Sb,c (4.7)
R
where Sb,c = dτ bċ. Then, the variation of action is
Z
δB S 0 = δ(S + Sb,c ) = δ dτ (Pµ Ẋ µ + bċ)
3
Z
= dτ (Pµ δ Ẋ µ + δbċ + bδ ċ)
Z  
d
= dτ Pµ  (cP µ ) − χċ

Z h i
= dτ ċ(Pµ P µ − m2 ) + cPµ Ṗ µ
2
Z     
1 µ 1 2 d 1 µ 1 2
= dτ ċ Pµ P − m + c Pµ P − m
2 2 dτ 2 2
Z   
d 1 1
= dτ  c Pµ P µ − m2
dτ 2 2
= 0. (4.8)
c

Define the BRST charge as QB ≡ 2 Pµ P µ + m2 = cχ where c is anticommuting. Using {c, c} = 0, we obtain

Q2B = c2 χ2 = 0. (4.9)

That is, the BRST charge is nilpotent. The physical state condition is given by

QB |Ψi = 0. (4.10)

The states that satisfy this condition are the physical states of the system and physical states must be BRST
invariant. We now have a compact and elegant statement of the physical state condition.

III. BRST quantization of the string


The action for the string is Z
1 ¯ µ
S= d2 z∂X µ ∂X (4.11)
2πα0
and the classical equation of motion is
¯ µ (z, z̄) = 0.
∂ ∂X
Now, we can consider the new action by adding an additional term Sb,c and it becomes

S = Sold + Sbc
Z Z
1 2 µ¯ 1 ¯
= d z∂X ∂Xµ + d2 zb∂c (4.12)
2πα0 2π

When a transformation is given by

z → z + v(z) =⇒ δX µ = −v∂X µ

then, the BRST transformation is

δB X µ = ic∂X µ
δB b = iT + iT bc
δB c = ic∂c (4.13)

where T = α10 : ∂X µ ∂Xµ : and T bc =: (∂b)c : −λ∂(: bc :). Here, T bc is added for invariance of S. Under the
BRST transformation, the resulting change in S is
Z
δB S = d2 ∂j¯B (4.14)
4 CHAPTER 4. THE STRING SPECTRUM

where the BRST current is


1 3
jB = cT + : cT bc : + ∂ 2 c
2 2
3 2
= cT + : bc∂c : + ∂ c.
2
Here, λ = 2 was chosen because we will have a holomorphic bc theory with λ = 2 in string theory. The BRST
operator is given by
I
dz
QB = jB
2πi
I  
dz 1 3
= cT + : cT bc : + ∂ 2 c . (4.15)
2πi 2 2
−n+1
and T (z) = n Ln z −n−2 , the BRST charge in string can be written
P P
Using mode expansion c(z) = n cn z
as
∞ 
X 1 bc
QB = : cn L−n + L−n − δn0 : (4.16)
n=−∞
2

where Lbc
P
−n = m (m − n)cm b−m−n . In the previous discussion(BRST quantization of the point-particle), we
found that Q2B = 0, but we now wish to ask whether this is true in case of the string. Using the Eq.(4.16),

1
Q2B = {QB , QB }
2
1X
cn cm [LT−n
ot
, LT−m
ot
] − (m − n)LT−n−m
ot

=
2 n,m
D(m3 − m) cbc (m3 − m)
 
1X
= cn cm δm+n,0 + δm+n,0 (4.17)
2 n,m 12 12

λ=2
where LTn ot = Ln + Lbc bc 2 2
n and c = 1 − 3(2λ − 1) −→ −26. Therefore, QB = 0 for D = 26.
Like a point particle, we can find the BRST invariant physical state

QB |physi = 0. (4.18)

When we define a new state such as

|physinew = |physi + QB |Ai , (4.19)

then we can find


   
new phys1 phys2 new
= hphys1 | + hA1 | QB · |phys2 i + QB |A2 i
= phys1 phys2 + A1 QB phys2 + phys1 QB A2 + A1 QB QB A2
= phys1 phys2 . (4.20)

Therefore, |physi + QB |Ai is a same physical state as |physi:


 
QB |physi + QB |Ai = 0. (4.21)

These two physical states, |physinew and |physi, have the same inner products with all physical states and are
therefore physically equivalent. So we identify the true physical space with a set of equivalence classes. This is
a natural construction for a nilpotent operator, and is known as the cohomology of QB .
5

equivalence classes

Hphys = { |physi : QB |physi = 0 }


Hexact = { |Ai : |Ai = QB |ϕi }
⇒ Hexact ⊆ Hphys
Hphys
=⇒ HBRST = (4.22)
Hexact

The complete Virasoro generators are given by

LTmot = Lm + Lbc
m = {QB , bm }. (4.23)

Then, from Q2B = 0 we obtain


[LTmot , QB ] = [{QB , bm }, QB ] = 0. (4.24)
In a similar manner, we can see that the algebra of the complete Virasoro generators Lm closes using Jacobi
identity:

[LTmot , LTn ot ] = [LTmot , {QB , bn }] = {QB , [LTmot , bn ]}


= (m − n){QB , bm+n } = (m − n)LTm+n ot
(4.25)

General Algebra

k
[Li , Lj ] = ifij Lk
{ci , bj } = δji , {ci , cj } = {bi , bj } = 0 (4.26)

The general form of the BRST operator is


i k i j
QB = ci Li − fij c c bk
 2 
1
= ci Li + Lbc . (4.27)
2 i

where Lbc k j
i = −ifij c bk . Using the Eqs.(4.26), (4.27) and Jacobi identity, we obtain

1
Q2B = {QB , QB }
2
1 i 1 k m i j l
= {c Li , cj Lj } − iflm
k l m i
c c {c , bk }Li − fij fkl c c c bm
2 2
1 k m i j l
= ici cj fij
k i
Lk − iflm cl cm Li − fij fkl c c c bm
2
= 0. (4.28)

BRST cohomology of the string

Let us now focus on the open string. Physical state satisfies the additional condition

b0 |ϕi = 0. (4.29)

Using |χi = c0 |ϕi, we can see that

ϕ ϕ = ϕ ϕ = 0, χ ϕ 6= 0, hϕ| c0 |ϕi =
6 0. (4.30)
6 CHAPTER 4. THE STRING SPECTRUM

The Eq.(4.29) implies


LT0 ot |ϕi = {QB , b0 } |ϕi = 0 (4.31)
since QB and b0 both annihilate |ϕi. We denote by Ĥ the space of states satisfying the conditions Eqs.(4.29)
and (4.31);
Ĥ = { |ϕi : b0 |ϕi = 0, LT0 ot |ϕi = 0 }. (4.32)
From the commutators {QB , b0 } = L0 and [QB , LT0 ot ] = 0, it follows that QB takes Ĥ into itself. If we consider
a new state such as |Zi = QB |ϕi, then
b0 |Zi = {b0 , QB } |ϕi = LT0 ot |ϕi = 0
L0 |Zi = [LT0 ot , QB ] |ϕi = 0. (4.33)
Therefore, we can see that
|Zi = QB |ϕi ∈ Ĥ. (4.34)

Example 1.
Let us now
E work out the first levels of the D = 26 flat spacetime string. If we write the state at the lowest level
~
as ϕ; k , then we have
E E
L0 ϕ; ~k = (α0 k 2 − 1) ϕ; ~k = 0 (4.35)
E
using L0 ϕ; ~k = 0. Therefore, we obtain
1
m2 = −k 2 = − 0 . (4.36)
α
This state is invariant, E
QB ϕ; ~k = 0, (4.37)

because every term in QB contains either lowering operators or LT0 ot . This also shows that there are no exact
states at this level, so each invariant state corresponds to a cohomology class. The Eq.(4.36) means that these
are just the states of the tachyon.

Example 2.
At the next level, there are 26 + 2 states,
E   E
µ
Z, ~k = Aµ (~k)α−1 + β(~k)b−1 + γ(~k)c−1 ϕ; ~k , −k 2 = 0, (4.38)

depending on a 26-vector Aµ (~k) and two constants, β and γ. The norm of this state is
D E D    E
Z, ~k Z, k~0 = ϕ; ~k A∗µ (~k)α1µ + β ∗ (~k)b1 + γ ∗ (~k)c1 Aµ (~k)αµ,−1 + β(~k)b−1 + γ(~k)c−1 ϕ; k~0
D E
= A∗µ Aµ + β ∗ γ + βγ ∗ ϕ; ~k ϕ; k~0 (4.39)

using Aµ Aν [α1µ , α−1


ν
] = Aµ Aν η µν = Aµ Aν , {bm , cn } = δm,−n and {bm , bn } = {cm , cn } = 0. From the eq.(4.39),
we can see that there are 26 positive norm states and 2 negative norm states because |Aµ |2 > 0 for µ = 1, 2, · · · 25,
−|Aµ |2 < 0 for µ = 0, |β|2 > 0 for β = γ and −|β|2 < 0 for β = −γ. The BRST condition QB |Zi = 0 gives
 
QB |Zi = c−1 kµ α1µ + c1 kµ α−1
µ
|Zi
  E
= c−1 kµ α1µ + c1 kµ α−1 µ  µ
Aµ (~k)α−1 + β(~k)b−1 + γ(~k)c−1 ϕ; ~k
E
µ 
= c−1 kµ Aµ + βkµ α−1 ϕ; ~k
= 0 (4.40)
7

An invariant state therefore satisfies


kµ Aµ = 0, β=0 (4.41)
which get rid of negative norm states:

k0 A0 = 0, k0 6= 0 =⇒ A0 = 0

β=0 =⇒ β = γ = 0.
Therefore, there are 24 linearly independent states remaining, which have positive norm, being orthogonal to
all physical states including themselves. In the Eq.(4.40), we already see that
E E
µ
QB Aµ α−1 ϕ; ~k = c−1 kµ Aµ ϕ; ~k
E 1 µ
E
=⇒ c−1 ϕ; ~k = QB Aµ α−1 ϕ; ~k = exact.
k·A
E
Thus the ghost state c−1 ϕ; ~k is BRST-exact.

IV. The no-ghost theorem


Define the transverse Hilbert space H⊥ to consist of states in Ĥ that have no longitudinal (X 0 , X 1 , b, or c)
excitations:
H⊥ = { |ϕi : no longitudinal (X 0 , X 1 , b, c) mode } (4.42)
Light-cone gauge-fixing elimilates the longitudinal oscillators directly - the light-cone Hilbert space is isomorphic
to H⊥ . Generally the BRST cohomology is isomorphic to H⊥ . That is, it has the same number of states at
each mass level, and has a positive inner product. This is the no-ghost theorem.

Proof
Before we prove the no-ghost theorem, Let us find the new operators Q1 and S. Define the light-cone oscillators

± 1 0 1
αm = √ (αm ± αm ), m, n ∈ Z − {0} (4.43)
2
which satisfy
+ − + + − −
[αm , αm ] = −mδm,−n , [αm , αm ] = [αm , αm ] = 0. (4.44)
Using the quantum number

X 1 + −
N lc = α α ,
m=−∞
m −m m
m6=0

decompose the BRST generator:


QB = Q1 + Q0 + Q−1 , (4.45)
lc lc
where Qj changes N by j units, [N , Qj ] = jQj . From Q2B = 0, we obtain Q21 = 0, that is, Q1 itself is
nilpotent and has a cohomology. Explicitly

√ X

Q1 = − 2α0 k + α−m cm . (4.46)
m=−∞
m6=0

Define S ≡ {Q1 , R} where



1 X
+
R= √ α−m bm , (4.47)
0 +
2α k m=−∞
m6=0
8 CHAPTER 4. THE STRING SPECTRUM

then S can be written as



X

+
αn− − α−n αn+ .

S= nb−n cn + nc−n bn − α−n (4.48)
n=1

We can also see that

[Q1 , S] = [Q1 , {Q1 , R}]


= Q1 {Q1 , R} − {Q1 , R}Q1 = Q1 RQ1 − Q1 RQ1
= 0. (4.49)

Now, we prove the no-ghost theorem. The proof has two parts of which the first step is to find the cohomology
of a BRST operator Q1 and the second step is to show that the cohomology of the full QB is identical to that
of Q1 .
The 1st step :
Theorem : If |ϕi is Q1 -invariant with S |ϕi = s |ϕi, then |ϕi is Q1 -exact for nonzero s.
Proof

|ϕi = S |Zi = {Q1 , R} |Zi


= Q1 R |Zi + RQ1 |Zi = Q1 R |Zi . (4.50)

Therefore the Q1 cohomology can be nonzero only at s = 0.


±
S |ϕi = 0 iff |ϕi has no α−m , b−m , c−m , m > 0 ⇒ |ϕi ∈ H⊥

=⇒ Cohomology of Q1 ≡ H⊥

The 2nd step :


|ϕi ∈ the kernel of S.
Define
{QB , R} = {Q1 , R} + {QB − Q1 , R} = S + U (4.51)
where U ≡ {QB − Q1 , R}.
±
U |ϕi has no α−m modes. ⇒ U |ϕi 6∈ the kernel of S.
If |ϕi is annihilated by S, then

|Zi = 1 − S −1 U + S −1 U S −1 U − · · · |ϕi


is annihilated by (S + U ):
(S + U ) |Zi = S |ϕi = 0.

Cohomology of QB

= the kernel of (S + U ) ∼ the kernel of S

= Cohomology of Q1 .

You might also like