We briefly describe the special number field sieve integer factoring algorithm, emphasizing the polynomial selection, and tell how we have used it to factor large integers on many workstations.

We report the factorization of a 135-digit integer by the three large prime variation of the multiple polynomial quadratic sieve, the largest factorization ever performed with MPQS. We show that it is worthwhile to use three large primes, contrary to previous work.

We have completely factored the numerators of the first 76 Bernoulli numbers and the first 44 Euler numbers. We studied the results seeking new theorems about the prime factors of these numbers and rediscovered two nearly-forgotten congruences for the Euler numbers.

The goal of the Clouds project at Georgia Tech is the implementation of a fault-tolerant distributed operating system based on the notions of objects and actions, which will provide an environment for the construction of reliable applications.  As part of the Clouds project, the author designed and implemented a high-level language in which those levels of the Clouds system above the kernel level are being implemented.  The Aeolus language provides access to the synchronization and recovery features of Clouds.  It also provides a framework within which to study programming methodologies suitable for action-object systems such as Clouds. This dissertation describes programming methodologies appropriate to the design of fault-tolerant servers needed in the Clouds system.  Among the properties needed by these objects are resilience and availability.  As part of this research, several case studies - that will serve as designs for actual Cloud servers - have been developed in Aeolus.  Among the issues examined using these case studies are: the use of knowledge about the semantics of an object, as opposed to automatic provisions, in designing for resilience and availability; the tradeoffs between consistency and availability for such objects; the support from the Aeolus runtime system and from the Clouds kernel needed for providing fault tolerance; and high-level language features for resilience and availability which may be derived from experience with programming in Aeolus.

One of the problems encountered in distributed systems is how to find the location of the resources needed by a computation.  In many situations the location may have to be found at run time, when the resource is accessed, thus the efficiency of the location algorithm will affect the performance of the system.  In general, the larger the distributed system, the more the number of processors at which a resource may reside at the time it is accessed.  The general problem of resource location in distributed systems has not been addressed adequately, and most of the systems have adopted ad hoc solutions without a careful study of the performance of algorithms used.  In this thesis it is studied the problem of finding the location of resources in order to get a better understanding of the factors affecting the cost of a location algorithm.  This study will make it possible to judge proposed algorithms as well as to come up with new ones, optimized for particular systems.

The Linda shared space model and its derivatives provide great flexibility for building parallel and distributed applications composed if independent processes.  However, the shared space model does not provide protection against untrustworthy processes.  Linda processes communicate by reading and writing messages in a globally visible data space, so a malicious process can launch any number of security attacks.  This paper presents the design of a new coordination model which extends Linda with fine grained access control.  The semantics of the model which is presented in the context of a process calculus.  A prototype of our model, called SecOS, has been implemented in JAVA.

Executing computatutations in a single instance of safe language virtual machine can improve performance and overall platform scalability.  It also poses various challenges.  One of them is providing a fast inter-application communication mechanism.  In addition for being efficient, such a mechanism should not violate any functional and non-functional properties of its environment, and should also support enforcement of application-specific security policies.  This paper explores the design and implementation of a communication substrate for applications executing within a single Java virtual machine modified to enable safe and interference-free execution of isolated computations.  Designing an efficient extension that des not break isolation properties and at the same time pragmatically offers an intuitive API has proven non-trivial.  This paper demonstrates a set of techniques that lead to at least an eight-fold performance improvement over the in-process inter-application communication using standard mechanisms offered by the Java platform.

Object-oriented languages provide little support for encapsulating objects.  Reference semantics allows objects to escape their defining scope.  The pervasive aliasing that ensues remains a major source of software defects.  This paper introduces Kacheck/J a tool for inferring object encapsulation properties in large Java programs.  Our goal is to develop practical tools to assist software engineers, thus we focus on simple and scalable techniques.  Kacheck/J is able to infer confinement for Java classes.  A class and its subclasses are confined if all of their instances are encapsulated in their defining package.  This simple property can be used to identify accidental leaks of sensitive objects.  The analysis is scalable and efficient; Kacheck/J is able to infer confinement on a corpus of 46,000 classes (115 MB) in 6 minutes.