
We propose a class of intrinsic Gaussian processes (inGPs) for
interpolation, regression and classification on manifolds with a primary focus
on complex constrained domains or irregular shaped spaces arising as subsets or
submanifolds of R, R2, R3 and beyond. For example, inGPs can accommodate
spatial domains arising as complex subsets of Euclidean space. inGPs respect
the potentially complex boundary or interior conditions as well as the
intrinsic geometry of the spaces. The key novelty of the proposed approach is
to utilise the relationship between heat kernels and the transition density of
Brownian motion on manifolds for constructing and approximating valid and
computationally feasible covariance kernels. This enables inGPs to be
practically applied in great generality, while existing approaches for
smoothing on constrained domains are limited to simple special cases. The broad
utilities of the inGP approach is illustrated through simulation studies and
data examples.

Gaussian processes (GPs) are very widely used for modeling of unknown
functions or surfaces in applications ranging from regression to classification
to spatial processes. Although there is an increasingly vast literature on
applications, methods, theory and algorithms related to GPs, the overwhelming
majority of this literature focuses on the case in which the input domain
corresponds to a Euclidean space. However, particularly in recent years with
the increasing collection of complex data, it is commonly the case that the
input domain does not have such a simple form. For example, it is common for
the inputs to be restricted to a nonEuclidean manifold, a case which forms the
motivation for this article. In particular, we propose a general extrinsic
framework for GP modeling on manifolds, which relies on embedding of the
manifold into a Euclidean space and then constructing extrinsic kernels for GPs
on their images. These extrinsic Gaussian processes (eGPs) are used as prior
distributions for unknown functions in Bayesian inferences. Our approach is
simple and general, and we show that the eGPs inherit fine theoretical
properties from GP models in Euclidean spaces. We consider applications of our
models to regression and classification problems with predictors lying in a
large class of manifolds, including spheres, planar shape spaces, a space of
positive definite matrices, and Grassmannians. Our models can be readily used
by practitioners in biological sciences for various regression and
classification problems, such as disease diagnosis or detection. Our work is
also likely to have impact in spatial statistics when spatial locations are on
the sphere or other geometric spaces.

Chiral differential operators (CDOs) are closely related to string geometry
and the quantum theory of twodimensional sigma models. This paper investigates
two topics about CDOs on smooth manifolds. In the first half, we study how a
Lie group action on a smooth manifold can be lifted to a `formal loop group
action' on an algebra of CDOs; this turns out to be a condition on the
equivariant first Pontrjagin class. The case of a principal bundle receives
particular attention and gives rise to a type of vertex algebras of great
interest. In the second half, we introduce a construction of modules over CDOs
using the said `formal loop group actions' and semiinfinite cohomology.
Intuitively, these modules should have a geometric meaning in terms of `formal
loop spaces'. The first example we study leads to a new conceptual construction
of an arbitrary algebra of CDOs. The other example, called the spinor module,
may be useful for a geometric theory of the Witten genus.

The first part of this paper provides a new formulation of chiral
differential operators (CDOs) in terms of global geometric quantities. The main
result is a recipe to define all sheaves of CDOs on a smooth csmanifold; its
ingredients consist of an affine connection and an even 3form that trivializes
the first Pontrjagin form. With the connection fixed, two suitable 3forms
define isomorphic sheaves of CDOs if and only if their difference is exact.
Moreover, conformal structures are in onetoone correspondence with even
1forms that trivialize the first Chern form.
Applying our work in the first part, we construct what may be called "chiral
Dolbeault complexes" of a complex manifold M, and analyze conditions under
which these differential vertex superalgebras admit compatible conformal
structures or extra gradings (fermion numbers). When M is compact, their
cohomology computes (in various cases) the Witten genus, the twovariable
elliptic genus and a spinc version of the Witten genus. This part contains
some new results as well as provides a geometric formulation of certain known
facts from the study of holomorphic CDOs and sigma models.

This article is the first report of an ongoing project aimed at finding a
geometric interpretation of the Witten genus and other tmf classes. Section 2
reviews the sheaves of chiral differential operators (CDOs) over a complex
manifold, including their construction, obstructions and relation with the
Witten genus. In section 3, the structure of each sheaf of CDOs is reorganized
in terms of modules over the sheaf of holomorphic functions. This invokes the
notion of a differential graded vertex algebroid. The construction of sheaves
of CDOs is due to Gorbounov, Malikov and Schechtman, and so is the notion of a
vertex algebroid; the differential graded version is first introduced here.
Section 4 contains the main result, namely the construction of a sheaf of
differential graded conformal vertex algebras that provides a fine resolution
of a sheaf of CDOs. This `infinite dimensional Dolbeault complex' plays a role
for the Witten genus similar to that of the Dolbeault complex for the Todd
genus.

This is the Ph.D. dissertation of the author. The project has been motivated
by the conjecture that the HopkinsMiller tmf spectrum can be described in
terms of `spaces' of conformal field theories. In this dissertation, spaces of
field theories are constructed as classifying spaces of categories whose
objects are certain types of field theories. If such a category has a symmetric
monoidal structure and its components form a group, by work of Segal, its
classifying space is an infinite loop space and defines a cohomology theory.
This has been carried out for two classes of field theories: (i) For each
integer n, there is a category SEFT_n whose objects are the StolzTeichner
(11)dimensional super Euclidean field theories of degree n. It is proved that
the classifying space SEFT_n represents degreen K or KO cohomology,
depending on the coefficients of the field theories. (ii) For each integer n,
there is a category AFT_n whose objects are a kind of (21)dimensional field
theories called `annular field theories,' defined using supergeometric versions
of circles and annuli only. It is proved that the classifying space AFT_n
represents the degreen elliptic cohomology associated with the Tate curve. To
the author's knowledge, this is the first time the definitions of
lowdimensional supersymmetric field theories are given in full detail.