The Geometry/Gauge Theory Duality and the Dijkgraaf-Vafa Conjecture

Abstract

In this dissertation we discuss various issues concerning application of the Dijkgraaf-Vafa (DV) conjecture to the study of supersymmetric gauge theories. The DV approach is very powerful in that it provides a systematic way of computing the nonperturbative, often even exact, superpotential of the system, which was possible only on a case-by-case basis in the more traditional approach based on holomorphy and symmetry. This conjecture has been checked for many nontrivial examples, but the range of its applicability remained unclear. We give an explicit example, Sp(N) theory with antisymmetric tensor, which reveals the subtleties in applying the conjecture. We show that, the superpotential obtained by a straightforward application of the DV approach starts to disagree with the standard gauge theory result at N/2+1 loops. The same discrepancy is reproduced in the generalized Konishi anomaly method. In order to look for the physical origin of the discrepancy, we consider the string theory realization of the gauge theories by Calabi-Yau compactifications. By closely analyzing the physics that accompanies the geometric transitions involved, we clarify the prescription regarding when to include a glueball field as the physical field, and when to not. In particular, the aforementioned discrepancy is resolved if we follow this prescription and introduce a glueball field for the "Sp(0)" group. Furthermore, we generalize the prescription to include flavors and demonstrate that the matrix model computations with the generalized prescription correctly reproduce the gauge theory results.