Engineering PapersSearch

SEARCH · Engineering Papers

Results for “Natarajan's algorithm”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

Sparse Regression as a Sparse Eigenvalue Problem

We extend the l0-norm "subspectral" algorithms for sparse-LDA [5] and sparse-PCA [6] to general quadratic costs such as MSE in linear (kernel) regression. The resulting "Sparse Least Squares" (SLS) problem is also NP-hard, by way of its equivalence to a rank-1 sparse eigenvalue problem (e.g., binary sparse-LDA [7]). Specifically, for a general quadratic cost we use a highly-efficient technique for direct eigenvalue computation using partitioned matrix inverses which leads to dramatic x103 speed-ups over standard eigenvalue decomposition. This increased efficiency mitigates the O(n4) scaling behaviour that up to now has limited the previous algorithms' utility for high-dimensional learning problems. Moreover, the new computation prioritizes the role of the less-myopic backward elimination stage which becomes more efficient than forward selection. Similarly, branch-and-bound search for Exact Sparse Least Squares (ESLS) also benefits from partitioned matrix inverse techniques. Our Greedy Sparse Least Squares (GSLS) generalizes Natarajan's algorithm [9] also known as Order-Recursive Matching Pursuit (ORMP). Specifically, the forward half of GSLS is exactly equivalent to ORMP but more efficient. By including the backward pass, which only doubles the computation, we can achieve lower MSE than ORMP. Experimental comparisons to the state-of-the-art LARS algorithm [3] show forward-GSLS is faster, more accurate and more flexible in terms of choice of regularization

Exact Sparse Least Squares (ESLS)

Mechanical verification of a schematic Byzantine clock synchronization algorithm

Schneider generalizes a number of protocols for Byzantine fault tolerant clock synchronization and presents a uniform proof for their correctness. The authors present a machine checked proof of this schematic protocol that revises some of the details in Schneider's original analysis. The verification was carried out with the EHDM system developed at the SRI Computer Science Laboratory. The mechanically checked proofs include the verification that the egocentric mean function used in Lamport and Melliar-Smith's Interactive Convergence Algorithm satisfies the requirements of Schneider's protocol.

Shankar, Natarajan

Scheduling real-time, periodic jobs using imprecise results

A process is called a monotone process if the accuracy of its intermediate results is non-decreasing as more time is spent to obtain the result. The result produced by a monotone process upon its normal termination is the desired result; the error in this result is zero. External events such as timeouts or crashes may cause the process to terminate prematurely. If the intermediate result produced by the process upon its premature termination is saved and made available, the application may still find the result unusable and, hence, acceptable; such a result is said to be an imprecise one. The error in an imprecise result is nonzero. The problem of scheduling periodic jobs to meet deadlines on a system that provides the necessary programming language primitives and run-time support for processes to return imprecise results is discussed. This problem differs from the traditional scheduling problems since the scheduler may choose to terminate a task before it is completed, causing it to produce an acceptable but imprecise result. Consequently, the amounts of processor time assigned to tasks in a valid schedule can be less than the amounts of time required to complete the tasks. A meaningful formulation of this problem taking into account the quality of the overall result is discussed. Three algorithms for scheduling jobs for which the effects of errors in results produced in different periods are not cumulative are described, and their relative merits are evaluated.

Liu, Jane W. S.

The influence of NO and ClO variations at twilight on the interpretation of solar occultation measurements

Measurement of short-lived photochemically-produced species in the stratosphere by solar occultation is difficult because the rapid variation of such species near the terminator introduces ambiguities in interpreting the measured absorption in terms of meaningful atmospheric abundances. These variations produce tangent path concentrations that are asymmetric relative to the tangent point, as opposed to the symmetrical distribution usually assumed in most inversion algorithms. Neglect of this asymmetry may yield an inverted profile that deviates significantly from the true sunset/sunrise profile. In the present paper, the influence of this effect on solar occultation measurements of ClO and NO is examined. The results show that average inhomogeneity factors, which measure the concentration variation along the tangent path and which can be calculated from a photochemical model, can indicate which species require more careful data analysis.

Boughner, R.

Priority in Process Algebras

This paper surveys the semantic ramifications of extending traditional process algebras with notions of priority that allow for some transitions to be given precedence over others. These enriched formalisms allow one to model system features such as interrupts, prioritized choice, or real-time behavior. Approaches to priority in process algebras can be classified according to whether the induced notion of preemption on transitions is global or local and whether priorities are static or dynamic. Early work in the area concentrated on global pre-emption and static priorities and led to formalisms for modeling interrupts and aspects of real-time, such as maximal progress, in centralized computing environments. More recent research has investigated localized notions of pre-emption in which the distribution of systems is taken into account, as well as dynamic priority approaches, i.e., those where priority values may change as systems evolve. The latter allows one to model behavioral phenomena such as scheduling algorithms and also enables the efficient encoding of real-time semantics. Technically, this paper studies the different models of priorities by presenting extensions of Milner's Calculus of Communicating Systems (CCS) with static and dynamic priority as well as with notions of global and local pre- emption. In each case the operational semantics of CCS is modified appropriately, behavioral theories based on strong and weak bisimulation are given, and related approaches for different process-algebraic settings are discussed.

Cleaveland, Rance

Designing a fuzzy scheduler for hard real-time systems

In hard real-time systems, tasks have to be performed not only correctly, but also in a timely fashion. If timing constraints are not met, there might be severe consequences. Task scheduling is the most important problem in designing a hard real-time system, because the scheduling algorithm ensures that tasks meet their deadlines. However, the inherent nature of uncertainty in dynamic hard real-time systems increases the problems inherent in scheduling. In an effort to alleviate these problems, we have developed a fuzzy scheduler to facilitate searching for a feasible schedule. A set of fuzzy rules are proposed to guide the search. The situation we are trying to address is the performance of the system when no feasible solution can be found, and therefore, certain tasks will not be executed. We wish to limit the number of important tasks that are not scheduled.

Yen, John

On the Quality of the Nimbus 7 LIMS Version 6 Water Vapor Profiles and Distributions

This report describes the quality of the Nimbus 7 Limb Infrared Monitor of the Stratosphere (LIMS) water vapor (H2O) profiles of 1978/79 that were processed with a Version 6 (V6) algorithm and archived in 2002. The V6 profiles incorporate a better knowledge of the instrument attitude for the LIMS measurements along its orbits, leading to improvements for its temperature profiles and for the registration of its water vapor radiances with pressure. As a result, the LIMS V6 zonal-mean distributions of H2O exhibit better hemispheric symmetry than was the case from the original Version 5 (V5) dataset that was archived in 1982. Estimates of the precision and accuracy of the V6 H2O profiles are developed and provided. Individual profiles have a precision of order 5% and an estimated accuracy of about 19% at 3 hPa, 14% at 10 hPa, and 26% at 50 hPa. Profile segments within about 2 km of the tropopause are often affected by emissions from clouds that appear in the finite field-of-view of the detector for the LIMS H2O channel. Zonally-averaged distributions of the LIMS V6 H2O are compared with those from the more recent Microwave Limb Sounder (MLS) satellite experiment for November, February, and May of 2004/2005. The patterns and values of their respective distributions are similar in many respects. Effects of a strengthened Brewer-Dobson circulation are indicated in the MLS distributions of the recent decade versus those of LIMS from 1978/79. A tropical tape recorder signal is present in the 7-month time series of LIMS V6 H2O with lowest values in February 1979, and the estimated, annually-averaged "entry-level" H2O is 3.5 to 3.8 ppmv. It is judged that this historic LIMS water vapor dataset is of good quality for studies of the near global-scale chemistry and transport for pressure levels from 3 hPa to about 70 to 100 hPa.

Remsberg, E. E.