Wilson loops, which satisfy the same boundary conditions.
We observe that for the Poincar´e-AdS solutions regularity requires a quantization
condition on the Wilson loops which was missed in [12]. This arises from considering
the uplift of the configuration to a M4×S7solution of M theory, where the Wilson
loops parametrise a shift on the S7as we go around the ϕcircle.
For fixed flux boundary conditions, the coexistence of the soliton and the Poincar´e-
AdS and domain wall solutions leads to a degeneracy of supersymmetric solutions as
in [12]. We also consider the alternate boundary condition of fixed currents on the
boundary – we will in the sequel refer to these boundary conditions as fixed charge –
although this is actually correct only for the Euclidean continuation, as the currents
are spacelike in the Lorentzian solution. We find that, for supersymmetry-preserving
fixed charge boundary conditions, there are two distinct soliton solutions, leading to
a new kind of degeneracy of supersymmetric solutions (the Poincar´e-AdS and domain
wall solutions do not satisfy these boundary conditions, so there is no degeneracy at
fixed charge in the previous case studied in [12]).
Finally, we find that the non-supersymmetric solutions of [12] also satisfy the bound-
ary conditions which give supersymmetric solutions for fixed charge, and one branch of
those solutions has lower energy than the supersymmetric solutions. This is surprising
as we would expect the supersymmetric solutions to saturate a BPS bound, forbidding
the existence of solutions with lower energy. The BPS bound for these alternate bound-
ary conditions has however not been explicitly worked out as far as we are aware. We
discuss this result in light of the positive energy theorem [15,16], which implies that the
energy of a supersymmetry preserving solution is lower than the energy of any other
solution satisfying the same boundary conditions. We show that our result is not in
contradiction with this general property. The central point of the argument is that a
necessary condition for the positive energy theorem to apply is the existence, for the
non-supersymmetric solution, of an asymptotic Killing spinor which coincides, up to
O(1/r2) terms at radial infinity, with the Killing spinor of the supersymmetric one.
Since the latter has antiperiodic boundary conditions along the circle at infinity, in or-
der for the positive energy theorem to apply, the non-supersymmetric solutions should
admit an asymptotic Killing spinor with the same property at the boundary. As we
shall prove, this is the case only if the charges at infinity have specific values, for which
the energy of the non-supersymmetric solution exceeds that of the supersymmetric one.
In summary, there is no contradiction with the positive energy theorem if we include
among the boundary conditions those applying to the asymptotic Killing spinor.
Another important direction for future work is to find a more general understanding
of the degeneracy of the susy solutions. We believe this is a generic feature of such
boundary conditions; we intend to provide a general proof in a forthcoming paper.
It is also possible to construct black holes in this theory [17–23] and endow them
with Wilson lines along the lines of [24]. This will provide an even more complete phase
4