April 2, 1992

Stony
Stony Brook
Stony Brook
Brook
ITP-SB-92-13

THE N=4 STRING IS THE

SAME AS THE N=2 STRING
arXiv:hep-th/9204005v2 28 May 1992

W. Siegel1

Institute for Theoretical Physics

State University of New York, Stony Brook, NY 11794-3840

ABSTRACT

We redo the quantization of the N=4 string, taking into account the re-
ducibility of the constraints. The result is equivalent to the N=2 string, with
critical dimension D=4 and signature (++−−). The N=4 formulation has
several advantages: the σ-model field equations are implied classically, rather
than by quantum/β-function calculations; self-duality/chirality is one of the
super-Virasoro constraints; SO(2,2) covariance is manifest. This reveals that
the theory includes fermions, and is apparently spacetime supersymmetric.

1
Work supported by National Science Foundation grant PHY 90-08936.
Internet address: [email protected].
1. Summary

Recently there has been renewed interest in the N=2 string [1] as a description
of self-dual Yang-Mills and gravity. The latter theories can be found by S-matrix cal-
culations [2], or as σ-models (on the string) whose equations of motion (self-duality
conditions) follow from requiring the vanishing of the β-function in quantum calcu-
lations [3]. A major drawback of this formalism is that the SO(2,2) covariance of the
spacetime is obscure, since the N=2 world-sheet supersymmetry is not SO(2,2) co-
variant. The one state in such self-dual theories is represented by a “scalar” [4] which
is actually the one surviving component of a tensor on-shell in an appropriate gauge.
(For self-dual Yang-Mills, this is the scalar of Yang, and for self-dual gravity it’s the
Kähler scalar. In SO(2,2)=Sp(2)⊗Sp(2) spacetime, irreducible representations of the
Poincaré group are self-dual or chiral, and have only one real physical component on
shell.) For example, the Yang-Mills field can be expressed in terms of a Hertz poten-
tial as Aa = ∂ b Hab + O(H 2 ). If the Hertz potential is constrained to be anti-self-dual,
then the self-duality equation on the field strength becomes H + O(H 2 ) = 0. Then
the on-shell gauge invariance δHab = (∂[a λb] − dual) + O(H) can be used to gauge
away 2 of the 3 components of H. (Similar remarks apply for self-dual gravity, where
the Kähler scalar is one component of a tensor in the same Lorentz representation
as the anti-self-dual part of the Weyl tensor.) Extra ghosts are required to compen-
sate for the fact that the usual gauge fields are derivatives of this “scalar.” Since
this component has nonvanishing helicity, it transforms nontrivially under Lorentz
transformations even on shell, but these transformations have not been analyzed in
the string theory. (For self-dual/chiral representations of SO(2,2), the little group is
scale transformations GL(1), instead of compact U(1).)

The N=4 string [5] has drawn little attention because its critical dimension was
thought to be negative. However, the reducibility of the constraints was unnoticed:
In fact, if one first solves the N=2 subset of the N=4 constraints, the remaining
constraints are redundant. (Thus, for example, arguments for the critical dimension
based on the conformal anomaly will fail if they neglect the higher-generation ghosts
resulting from this reducibility.) This should not be surprising, since the N=2 con-
straints in the critical dimension eliminate all excitations, reducing the string to a
particle. As a result, the N=2 string can be treated in the N=4 formalism. The N=4
constraints, unlike the N=2 ones, are SO(2,2) covariant. The conformal-weight 1 con-
straints form the affine Lie algebra for Sp(2); this Sp(2) is half of the SO(2,2) spin,
and its vanishing is exactly the constraint of self-duality/chirality. N=4 σ-models are

2
finite [6-8]; as a result, the self-duality equations for Yang-Mills and gravity are found
already at the classical level, required by the N=4 supersymmetry [6,7]. The explicit
SO(2,2) also reveals a difference between Neveu-Schwarz and Ramond sectors: For
example, for the open string, just as for N=1 (except for the self-duality), the Neveu-
Schwarz sector describes self-dual Yang-Mills while the Ramond sector describes a
Weyl Majorana spinor, so the whole string describes self-dual N=1 super Yang-Mills.
(In the N=2 formalism the SO(2,2) transformations of these 1-component objects are
obscure, while the σ-model approach is incapable of showing the existence of spinor
fields.)
The N=4 string shares several properties with the Green-Schwarz formalism: (1)
The critical dimension appears classically, since for D>4 there is no SO(D) symmetry.
In Green-Schwarz the condition D=3,4,6,10 is found classically, which is not as strong
as the quantum condition D=10, but more than for classical N=0 or 1 spinning strings,
which say nothing, or the N=2 string, where even for D=4 the SO(2,2) symmetry is
hidden. (2) σ-model field equations appear classically. In Green-Schwarz, some of the
field equations appear classically, as a consequence of κ-symmetry, which is analogous
to world-sheet supersymmetry. (3) The constraints are reducibile. However, they are
not also second-class, as in Green-Schwarz. (4) GSO projection is unnecessary (there
is no tachyon, and the Sp(2) constraint already makes the spinor chiral).

2. Constraints

The N=4 constraints can be written as

′ ′′
A ≡ 21 P αβ Pαβ ′ − i 12 Γαβ Γ′αβ ′′ = 0, Bα′ β ′′ ≡ P γ α′ Γγβ ′′ = 0, Cα′′ β ′′ ≡ Γγ (α′′ Γγβ ′′ ) = 0

(We work with holomorphic variables P ≡ ∂X and Γ, suitable for describing the
open string. As usual, the closed string requires a second set of such variables for the
other handedness, while the heterotic string requires bosonic variables of the other
handedness.) The N=2 subset of these constraints is given by A (Virasoro), B+′ −′′ ,
B−′ +′′ , and C+′′ −′′ (GL(1)). (Note that the appropriate reality conditions for SO(2,2)
imply that the N=4 gauged group is Sp(2)=SL(2)=SU(1,1), not SU(2), while the
subgroup for N=2 is real scale transformations GL(1), not U(1).) Since we consider
both the N=2 and N=4 string in D=4, the same set of physical variables (P and Γ,
4 of each) is used.
To analyze the constraints, we can consider the usual Gupta-Bleuler-like analysis
in terms of the physical oscillators, imposing the non-negative modes of the constraints

3
on the states, and using the negative modes as generators of gauge transformations.
In the N=2 theory, this analysis (as do other analyses) implies that the physical states
are just the ground states, constructed without the oscillator modes of P and Γ, and
they satisfy the massless Klein-Gordon equation p2 = 0. (The BRST proof [9] is
equivalent to this way of phrasing Gupta-Bleuler. In the Ramond sector, there are
also fermionic constraints in terms of the zero-modes γ and p.)

In the N=4 theory, we can make the same type of analysis, but consider the
N=2 subset of the constraints first, where the analysis is the same, since the physical
variables Γ and P are the same, and this subset has the same representation in terms
of them as in the N=2 theory. Since the physical states are again restricted to ground
states, the positive modes of the remaining N=4 constraints B+′ +′′ , B−′ −′′ , C+′′ +′′ ,
and C−′′ −′′ then automatically vanish on the physical states, so these constraints are
redundant, and the negative modes generate no gauge transformations of the ground
states. In the Neveu-Schwarz sector, having eliminated all the non-zero modes of P
and all the modes of Γ (which is half-integrally moded), we are left with the constraint
p2 = 0.

In the Ramond sector, the N=2 constraints again eliminate all non-zero modes
of P and Γ, making all non-zero modes of the remaining N=4 constraints redundant.
This leaves only the zero modes of the N=4 constraints, to be imposed on states
which are a representation of just the zero modes of P and Γ, i.e. a Dirac spinor:
′′
In Sp(2)⊗Sp(2) notation, the representation of γαα′′ is |Λi = λα |αi + λα |α′′ i, where
γαα′′ |βi = Cαβ |α′′i and γαα′′ |β ′′ i = Cα′′ β ′′ |αi. (The C’s are the antisymmetric, her-
mitian metrics of the Sp(2)’s.) We are thus left with an N=4 spinning particle. The
zero mode of A is again the massless Klein-Gordon equation. The zero mode of C is
′′
one of the two Sp(2)’s of spin, and it kills one of the two Weyl spinors (λα ) making
up the Dirac spinor. Finally, the zero mode of B imposes the massless Dirac equation
on the surviving Weyl spinor (and sets to zero the gradient of the other Weyl spinor,
which was already eliminated by the C constraint). If we had instead solved just the
N=2 constraints, we would have been left with an N=2 spinning particle (as defined
by the zero modes of the N=2 string), and the results would be the same (though
not in covariant form) with an appropriate choice of the normal-ordering constant in
C+′′ −′′ .

In both sectors we are left with one physical polarization: In the Neveu-Schwarz
sector we have what appears to be a scalar but is actually the one surviving component

4
on-shell of self-dual Yang-Mills; in the Ramond sector we have the one on-shell com-
ponent of a Weyl Majorana spinor. In the N=4 formalism the Lorentz covariance of
the spinor is manifest because, as for the N=1 string, the ground state of the Ramond
sector is itself the spinor. On the other hand, in this analysis of the Neveu-Schwarz
sector in terms of just the physical operators P and Γ, the Lorentz transformations
of the vector are just as obscure in the N=4 formalism as in the N=2. This is related
to the fact that, as for the N=0 and N=1 strings, the vector is not the naive ground
state, but is created by acting with oscillators on a “tachyon.” This problem will be
resolved when the Lorentz covariant ghosts are understood, since the true ground
state in the Neveu-Schwarz sector is the zero-momentum Yang-Mills ghost, which is
in the BRST cohomology at unphysical ghost number. In the N=2 formalism the
SO(2,2) transformations of the states are obscure, so the one state of the Ramond
sector is confused with the one state of the Neveu-Schwarz sector. (Without Lorentz
transformations, statistics is also obscure.) This “equivalence” is usually explained by
the ambiguity in redefining A by adding an arbitrary constant times the derivative of
C+′′ −′′ ; but this redefinition is forbidden in the N=4 formulation by Sp(2) covariance.

Thus, at least at the free level, we have a spacetime-supersymmetric theory, self-

dual super Yang-Mills, consisting of self-dual Yang-Mills and a real chiral spinor (for
the open string; similar remarks for self-dual supergravity apply for the closed/heterotic
case, whose spectrum can be obtained by direct products of open string spectra). The
description is the same as for SO(3,1), except that the two types of Weyl spinor (in-
dex) are now both real, instead of being complex conjugates of each other. Thus,
anti-chiral superfield strengths can consistently be set to vanish while keeping the
corresponding chiral superfield. So, an N=1 spacetime supersymmetric multiplet
consists of real component field strengths A(α1 ...αn ) and B(α1 ...αn+1 ) , totally symmetric
′
in their indices, and killed by pαα contracted with any index, so each has one physical
polarization. The chirality of the fermions leads to the same kind of problem writing
′
actions as self-duality for the bosons: The usual action λα pαα′ λα would require both
chiralities. There is a corresponding difficulty in defining norms, but this should not
be surprising in a theory with 2 time dimensions. (Notice that norms are not nec-
essary in finding the physical states in the way we have defined the Gupta-Bleuler
procedure; analogously, the definition of BRST cohomology does not make use of
the norm.) Since super Yang-Mills in D=4 is also spacetime conformal, we have the
interesting result that a world-sheet superconformal theory results in a spacetime
superconformal theory. For SO(2,2), the superconformal group is SL(4|N) (for N su-
persymmetries; SL(4)=SO(3,3)); a manifestly SL(4|N) formulation of these self-dual

5
theories might be useful. Supersymmetry at the interacting level is expected to follow
from the type of construction used for N=1 strings [10]; however, a Lorentz covariant
supersymmetry generator will require the N=4 formalism and its ghosts.
Although the above demonstration of the redundancy of the N=4 constraints
is sufficient to show equivalence to N=2, an equivalent covariant analysis is needed
for a covariant treatment of ghosts and BRST. The reducibility of the constraints
(at least at the first level) is most easily formulated with harmonic superspace [8]
(although we use a slightly different harmonic superspace than in [8]). In addition
′ ′′ ′′
to the usual world-sheet coordinates z and θα α , we introduce uα . Then the usual
′
supersymmetric derivative dα′ α′′ and superfield Ψαα′′ = Γαα′′ +θα α′′ Pαα′ +z-derivative
terms are replaced with
′′ ′′
Ψα ≡ uα Ψαα′′ , dα′ ≡ uα dα′ α′′ , dα′ Ψα = 0

The constraints are then

′′ ′′
T ≡ 21 Ψα Ψα ≡ uα uβ Tα′′ β ′′ = 0

where the θ expansion of Tα′′ β ′′ gives C, B, A. The reducibility condition

Ψα T = 0

then follows from the commutation relations

′′
{Ψα (1), Ψβ (2)} = uα1 u2α′′ d4 δ 5 (z, θ)

which vanish at u1 = u2 . There is an infinite chain of such reducibilities (from

multiplying repeatedly by Ψα at each ghost level, since Ψα Ψβ ∼ Cαβ T ), as well as
new reducibility conditions which show up at the second ghost level. The reducibility
′
Ψα T = 0 includes such conditions as P αα Bα′ α′′ +· · · = 0, which is of the same form as
the infinite chain of reducibility conditions which appear for the analogous fermionic
constraint (P/ D) in the Green-Schwarz string. This suggests that an understanding
of the covariant quantization of the N=4 string might help with the Green-Schwarz
case.

ACKNOWLEDGMENTS
I thank Nathan Berkovits, Jim Gates, and Martin Roček for directing me to the
N=2 string, educating me about various points regarding it and N=4 σ-models, and
for many discussions.

6
REFERENCES
[1] M. Ademollo, L. Brink, A. D’Adda, R. D’Auria, E. Napolitano, S. Sciuto, E. Del Giu-
dice, P. Di Vecchia, S. Ferrara, F. Gliozzi, R. Musto, R. Pettorino, and J.H. Schwarz,
Nucl. Phys. B111 (1976) 77.
[2] H. Ooguri and C. Vafa, Mod. Phys. Lett. A5 (1990) 1389, Nucl. Phys. B361 (1991)
469, 367 (1991) 83;
N. Marcus, Group theory for open and closed N=2 strings, Tel Aviv preprint TAUP-
1929-91, to appear.
[3] H. Nishino and S.J. Gates, Jr., N=(2,0) superstring as the underlying theory of self-dual
Yang-Mills theory, Maryland preprint UMDEPP 92-137 (January 1992).
[4] J.F. Plebañski, J. Math. Phys. 16 (1975) 2395;
C.N. Yang, Phys. Rev. Lett. 38 (1977) 1377.
[5] M. Ademollo, L. Brink, A. D’Adda, R. D’Auria, E. Napolitano, S. Sciuto, E. Del Giu-
dice, P. Di Vecchia, S. Ferrara, F. Gliozzi, R. Musto, and R. Pettorino, Nucl. Phys.
B114 (1976) 297, Phys. Lett. 62B (1976) 105.
[6] L. Alvarez-Gaumé and D.Z. Freedman, Comm. Math. Phys. 80 (1981) 443;
U. Lindström and M. Roček, Nucl. Phys. B222 (1983) 285;
S.J. Gates, Jr., C.M. Hull, and M. Roček, Nucl. Phys. B248 (1984) 157;
A.Yu. Morozov and A. Perelomov, Nucl. Phys. B271 (1986) 620;
A. Galperin, E. Ivanov, V. Ogievetsky, and E. Sokatchev, Class. Quantum Grav. 2
(1985) 617;
C.M. Hull, Nucl. Phys. B260 (1985) 182;
L. Alvarez-Gaumé and P. Ginsparg, Comm. Math. Phys. 102 (1985) 311;
J. Grundberg, A. Karlhede, U. Lindström, and G. Theodoridis, Nucl. Phys. B282
(1987) 142.
[7] P.S. Howe and G. Papadopoulos, Nucl. Phys. B289 (1987) 264, Class. Quantum Grav.
5 (1988) 1647.
[8] E. Sokatchev and K.S. Stelle, Class. Quantum Grav. 4 (1987) 501;
F. Delduc, S. Kalitsyn, and E. Sokatchev, Class. Quantum Grav. 7 (1990) 1567.
[9] J. Bieńkowska, The generalized no-ghost theorem for N=2 SUSY critical strings, Chicago
preprint EFI 91-65 (November 1991).
[10] N. Ohta and S. Osabe, Phys. Rev. D39 (1989) 1641.