Style Algorithm Research Articles

A Distillation Algorithm for Floating-Point Summation

The addition of two or more floating-point numbers is fundamental to numerical computations. This paper describes an efficient "distillation" style algorithm which produces a precise sum by exploiting the natural accuracy of compensated cancellation. The algorithm is applicable to all sets of data but is particularly appropriate for ill-conditioned data, where standard methods fail due to the accumulation of rounding error and its subsequent exposure by cancellation. The method uses only standard floating-point arithmetic and does not rely on the radix used by the arithmetic model, the architecture of specific machines, or the use of accumulators.

Adaptive cancellation of multiple interfering sinusoids

The adaptive sinusoidal interference canceller (ASIC) is a least-mean-square (LMS)-style algorithm for eliminating a sinewave of known frequency from an observed signal. The ASIC is generalised to cancel multiple interfering sinusoids. The performance of the algorithm, particularly when the interference frequencies are not exactly known, is analysed. It is also shown that the ASIC can provide a significant improvement in the signal-to-noise ratio.

COMPARISON AND EVALUATION OF A CLASS OF IDA* ALGORITHMS

In this paper, we study the performance of various IDA*-style searches and investigate methods to improve their performance by predicting in each stage the threshold to be used for pruning. Without loss of generality, we consider minimization problems in this paper. We first present three models to approximate the distribution of the number of search nodes by lower bounds: exponential, geometric, and linear, and illustrate these distributions based on some well-known combinatorial search problems. Based on these distributions, we show the performance of an ideal IDA* algorithm and identify reasons why existing IDA*-style algorithms perform well. In practice, we will be able to know from experience the type of distribution for a given problem instance, but will not be able to know the parameters of this distribution until the instance is solved. Hence, we develop RIDA*, a method that estimates dynamically the parameters of the distribution, and predicts the best threshold to be used in each stage. Finally, we compare the performance of several IDA*-style algorithms—Korf’s IDA* and RBFS, RIDA*, IDA*_CR and DFS*—on several application problems, and identify conditions under which each of these algorithms will perform well.

Successive combination jet algorithm for hadron collisions.

Jet finding algorithms, as they are used in ${e}^{+}{e}^{\ensuremath{-}}$ and hadron collisions, are reviewed and compared. It is suggested that a successive combination style algorithm, similar to that used in ${e}^{+}{e}^{\ensuremath{-}}$ physics, might be useful also in hadron collisions, where cone style algorithms have been used previously.

Exploiting locality to provide adaptive routing of real-time flows in global internets (abstract)

This paper addresses the role of routing in supporting real-time applications across large integrated services internetworks. The increasing speed of underlying transmission media is enabling new, and perhaps more importantly, integrated applications. However, because of the size and decentralized nature of the global internet, we cannot assume that the network will be homogeneously capable of all new services. Therefore, routing must identify routes with specific type-of-service (TOS) capabilities. Moreover, we suggest that TOS-routing for performance-sensitive applications (e.g., real-time videoconference, real-time interaction with high-bandwidth sensor or experimental data) should be sensitive to significant load-changes, as well as to topology changes. At the same time, global scale means that we must minimize reliance on global consistency of routing databases and global distribution of dynamic routing information. Today’s internet routing does not adapt to load shifts for two sound reasonsstability and overhead. Loaddependent algorithms can become unstable, simultaneously directing all nodes to respond to load changes in the same way, causing oscillation. Further, distributing fine grain load information throughout the internet (a requirement for loop-free routing in today’s hop-by-hop routed Internet) would introduce excessive overhead. The Internet avoids global distribution of fine grained information by using hierarchical aggregation. If it distributes load information, it does so only within local domains, hiding fine detail between domains. Hierarchical aggregation is one of the most powerful ways of dealing with scaleZ. However, hierarchical aggregation does not provide information about the availability of real time routes across a wide-area, heterogeneous internet. For example, if the internetwork routing updates only contain information about configured, inter-domain, topology, then multiple routes to each destination will be computed and selected based only on the relatively-static information. If at call setup time the selected route is not available, or has insufficient resources to carry the flow, an alternate route will be tried. However, such a trial-and-error approach may introduce intolerable session-startup delays, especially in a high-speed network environment where many data transfers could be completed during the several round-trip times that it might take to discover a route. On the other hand, the overhead of flooding dynamic link or path status information is unacceptable in a large network. Our routing architecture addresses this problem by (a) advertising a different level of detail for those routes that are currently carrying real-time multimedia applications, than for routes that are carrying more generic (less performance sensitive) traffic, and (b) distributing this information to current users of the resource in question, not to the global network. We thereby exploit the route locality of internetwork traffic. Route locality is measured by the mapping of each source to the set of network resources traversed to reach the destinations with which it is communicating. The proposed mechanisms are used to complement, not replace, traditional hierarchical-aggregation techniques (i.e., inter-domain routing). We present what might appear to be an unsupportable requirement, i.e., different levels of routing detail from and to different nodes in the network. However, our proposed solutions can be made to work in the context of source routing. Our source routing uses a link-state style algorithm to compute routes 4. Domain-level,

Optoelectronic implementation of multilayer neural networks in a single photorefractive crystal

We present a novel, versatile optoelectronic neural network architecture for implementing supervised learning algorithms in photorefractive materials. The system is based on spatial multiplexing rather than the more commonly used angular multiplexing of the interconnect gratings. This simple, single-crystal architecture implements a variety of multilayer supervised learning algorithms including mean field theory, back-propagation, and Marr-Albus-Kanerva style algorithms. Extensive simulations show how beam depletion, rescattering, absorption, and decay effects of the crystal are compensated for by suitably modified supervised learning algorithms.

A new fault-tolerant algorithm for clock synchronization

We describe a new fault-tolerant algorithm for solving a variant of Lamport's clock synchronization problem. The algorithm is designed for a system of distributed processes that communicate by sending messages. Each process has its own read-only physical clock whose drift rate from real time is very small. By adding a value to its physical clock time, the process obtaines its local time. The algorithm solves the problem of maintaining closely synchronized local times, assuming that processes' local times are closely synchronized initially. The algorithm is able to tolerate the failure of just under one-third of the participating processes. It maintains synchronization to within a small constant, whose magnitude depends upon the rate of clock drift, the message delivery time and its uncertainty, and the initial closeness of synchronization. We also give a characterization of how far the clocks drift from real time. Reintegration of a repaired process can be accomplished using a slight modification of the basic alborithm. A similar style algorithm can also be used to achieve synchronization initially.