140
String Theory Demystifi ed
Mode Expansions and Boundary Conditions
The fi nal step in putting together the classical physics of the RNS superstring
follows the program used in the bosonic case—we need to apply boundary conditions
and write down the mode expansions. Specifi cally, we need to apply boundary
conditions for the fermionic fi elds. It is simplest to continue working in light-cone
coordinates and vary the fermionic part of the action. Before doing this, it can be
helpful to review some elementary calculus.
Recall integration by parts:
fx
dg
dx
dx fg
df
dx
gxdx
a
b
a
b
a
b
() ()=−
∫∫
The product
fg
is called the boundary term. When we vary the fermionic action, we
are going to obtain boundary terms for the fi elds
ψ
, so we need to specify boundary
conditions so that the variation in the action vanishes. The fermionic part of the
action in light-cone coordinates, modulo a few constants and ignoring the space-
time index is
Sd
F
=∂+∂
∫
−+ − +− +
2
σψ ψ ψ ψ
()
(7.23)
For simplicity, let’s consider one piece of this expression and vary it. We obtain
δ σψψ σδψψψ δψ
dd
22
∫∫
+− + +− + +− +
∂= ∂+∂[()]
Following the usual procedure applied in fi eld theory, we want to move the derivative
off the
δψ
+
term. This can be done using integration by parts. When this is done, we
pick up a boundary term:
ddd
2
0
2
σ ψ δψ τ ψ δψ σ ψ δψ
σ
σπ
∫∫∫
+− +
−∞
∞
++=
=
−+ +
∂= −∂()
A similar expression arises from the variation of the other term. All together, the
boundary terms obtained by varying the action are
δ τ ψδψ ψδψ ψδψ ψδψ
σπ
Sd
F
=−−−
−∞
∞
++ −−= ++ −−
∫
()()
σσ
=
{}
0
(7.24)