We present speculative parallelization techniques that can exploit parallelism in loops even in the presence of dynamic irregularities that may give rise to cross-iteration dependences. The execution of a speculatively parallelized loop consists of five phases: scheduling, computation, misspeculation check, result committing, and misspeculation recovery. While the first two phases enable exploi...

Agreement problems are a fundamental building block for constructing reliable distributed systems. While robust and eÆcient protocols exist in the crash-failure setting, protocols resilient against a Byzantine adversary tend to have problems with either eÆciency or security. This paper proposes an optimistic approach to the Byzantine agreement problem, combining the eÆciency of fully synchronou...

Learning methods vary in the optimism or pessimism with which they regard the informativeness of learned knowledge. Pessimism is implicit in hypothesis testing, where we wish to draw cautious conclusions from experimental evidence. However, this paper demonstrates that optimism in the utility of derived rules may be the preferred bias for learning systems themselves. We examine the continuum be...

On-line simulations play a critical role in understanding, supervising, and controlling large-scale systems. For instance, in military training applications, users are presented with complex scenarios that consist of both actual and simulated devices interacting in real-time. Troubleshooting of complex telephone networks may utilize on-line simulations to play`what if' games when evaluating alt...

Group communication systems are powerful building blocks that facilitate the development of fault-tolerant distributed applications. Such systems generally run in an asynchronous fault-prone environment, and provide semantics (called Virtual Syn-chrony) that mask the asynchrony and unreliability of the environment. In order to implement Virtual Synchrony semantics, group communication systems t...

We introduce the general and powerful scheme of predicting information re-use in optimization algorithms. This allows us to devise a computationally efficient algorithm for bandit convex optimization with new state-of-the-art guarantees for both Lipschitz loss functions and loss functions with Lipschitz gradients. This is the first algorithm admitting both a polynomial time complexity and a reg...