We advocate a framework for constructing perturbative closed string compactifications which do not have large-radius limits. The idea is to augment the class of vacua which can be described as fibrations by enlarging the monodromy group around the singular fibers to include perturbative stringy duality symmetries. As a controlled laboratory for testing this program, we study in detail six-dimensional (1,0) supersymmetric vacua arising from two-torus fibrations over a two-dimensional base. We also construct some examples of two-torus fibrations over four-dimensional bases, and comment on the extension to other fibrations.August, 2002
Weakly coupled supersymmetric string vacua without geometryIn this paper we will examine a new method for constructing supersymmetric, nongeometric string theories. The examples on which we focus most closely make up a class of solutions to supergravity in 7 + 1 dimensions with 32 supercharges. These solutions will involve nontrivial behavior of the metric and Neveu-Schwarz (NS) B-field, but not of any of the Ramond-Ramond fields, nor of the eight dimensional dilaton. (The ten-dimensional dilaton will vary, but only in such a way that the eight dimensional effective coupling is held fixed.) We will argue that these backgrounds are likely to represent sensible backgrounds for string propagation on which the dynamics of string worldsheets are determined by a two-dimensional conformal field theory of critical central charge, with a controlled genus expansion whose expansion parameter can be made arbitrarily small.Almost all known examples of perturbative string backgrounds are descibed by nonlinear sigma models, that is, by field theories containing scalar fields X µ parametrizing a topologically and geometrically nontrivial target space. The lagrangian for these theories, for weak curvatures R αβγδ R αβγδ << 1 in string units. Therefore the condition for conformal invariance is approximately the same as the Einstein equation for the target space metric. So a nonlinear sigma model whose target space is a smooth Ricci-flat manifold at large volume will always be approximately conformal, an approximation which improves if one scales up the manifold G µν → ΛG µν .The existence of a large-volume limit of a family of solutions or approximate solutions gives rise to the moduli problem: there are usually several massless scalars in the lowerdimensional effective field theory corresponding to Ricci-flat deformations of the target space. There is always at least one light scalar, or approximate modulus -namely, the overall size of the compact space -if flat ten-dimensional space is an exact solution to the string equations of motion at the quantum level. In critical superstring theory, to which we will restrict our attention exclusively in this paper, flat ten dimensional space is a solution to the equations of motion at the quantum level.1 So the overall volume modulus is an intrinsic difficulty for geometric compactifications of superstring theory. Even if potentials are generated for the volu...