Algebraic Geometry

We describe interdependencies among the quantum cohomology associativity relations. We strengthen the first reconstruction theorem of Kontsevich and Manin by identifying a subcollection of the associativity relations which implies the full system of WDVV equations. This provides a tool for identifying non-geometric solutions to WDVV.

We construct some families of automorphic forms on Grassmannians which have singularities along smaller sub Grassmannians, using Harvey and Moore's extension of the Howe (or theta) correspondence to modular forms with poles at cusps. Some of the applications are as follows. We construct families of holomorphic automorphic forms which can be written as infinite products, which give many new examples of generalized Kac-Moody superalgebras. We extend the Shimura and Maass-Gritsenko correspondences to modular forms with singularities. We prove some congruences satisfied by the theta functions of positive definite lattices, and find a sufficient condition for a Lorentzian lattice to have a reflection group with a finite volume fundamental domain. We give some examples suggesting that these automorphic forms with singularities are related to Donaldson polynomials and to mirror symmetry for K3 surfaces.

After the work of Bloch and Srinivas on correspondences and algebraic cycles we begin the study of a birational class of algebraic varieties determined by the property that a multiple of the diagonal is rationally equivalent to a cycle supported on proper subschemes.

I give a necessary and sufficient condition for a nef and big line bundle in positive characteristic to be semi-ample, and then give two applications: I show that the relative dualizing sheaf of the universal curve is semi-ample, in positive characteristic, and give a simple example which shows that this, and the semi-ampleness criterion, fail in characteristic zero. My second application is to Mori's program for minimal models of 3-folds. I prove a version of the Basepoint Free Theorem (for big line bundles on 3-folds of positive characteristic) and a simple Cone Theorem, for 3-folds of positive Kodaira dimension.

We give a 'recursive' formula (in terms of reducible limits) for counting rational curves on a variety moving in any sufficiently large and well-behaved family. Our approach is completely elementary and makes no use of moduli spaces for maps or the like.

Extending the method of Part I (alg-geom/9704004), we give recursive formulae for: the genus-1 Severi degree (formula first found by Getzler), the degree of the variety of 1-cuspidal curves of genus 0 or 1, and the linear (sectional) geometric genus of the genus-0 and genus-1 Severi varieties. The proof is short and elementary.

After reviewing Bertini's life story, a fascinating drama, we make a critical examination of the old statements and proofs of Bertini's two fundamental theorems, the theorem on variable singular points and the theorem on reducible linear systems. We explain the content of the statements in a way that is accessible to a nonspecialist, and we develop versions of the old proofs that are complete and rigorous by current standards. In particular, we prove a new extension of Bertini's first theorem, which treats variable r -fold points for any r .

We construct a (smooth, projective) surface over the field of rational numbers, which is a counterexample to the Hasse principle not accounted for by the Manin obstruction. The construction relies on the classical 4-descent on elliptic curves. By combining the Manin obstruction with descent we define a refinement of the Manin obstruction which for surfaces of the type considered suffices to explain the absence of a rational point.

Let X and Y be two smooth projective n-dimensional algebraic varieties X and Y over C with trivial canonical line bundles. We use methods of p-adic analysis on algebraic varieties over local number fields to prove that if X and Y are birational, they have the same Betti numbers.

We give a method to construct stable vector bundles whose rank divides the degree over curves of genus bigger than one. The method complements the one given by Newstead. Finally, we make some systematic remarks and observations in connection with rationality of moduli spaces of stable vector bundles.