The usual approach to deal with noise present in many real-world optimization problems is to take an arbitrary number of samples of the objective function and use the sample average as an estimate of the true objective value. The number of samples is typically chosen arbitrarily and remains constant for the entire optimization process. This paper studies an adaptive sampling technique that varies the number of samples based on the uncertainty of deciding between two individuals. Experiments demonstrate the effect of adaptive sampling on the final solution quality reached by a genetic algorithm and the computational cost required to find the solution. The results suggest that the adaptive technique can effectively eliminate the need to set the sample size a priori, but in many cases it requires high computational costs.

QR and LU factorizations for dense matrices are important linear algebra computations that are widely used in scientific applications. To efficiently perform these computations on modern computers, the factorization algorithms need to be blocked when operating on large matrices to effectively exploit the deep cache hierarchy prevalent in today's computer memory systems. Because both QR (based on Householder transformations) and LU factorization algorithms contain complex loop structures, few compilers can fully automate the blocking of these algorithms. Though linear algebra libraries such as LAPACK provides manually blocked implementations of these algorithms, by automatically generating blocked versions of the computations, more benefit can be gained such as automatic adaptation of different blocking strategies. This paper demonstrates how to apply an aggressive loop transformation technique, dependence hoisting, to produce efficient blockings for both QR and LU with partial pivoting. We present different blocking strategies that can be generated by our optimizer and compare the performance of auto-blocked versions with manually tuned versions in LAPACK, both using reference BLAS, ATLAS BLAS and native BLAS specially tuned for the underlying machine architectures.

Many turbulent premixed flames of practical interest are statistically stationary. They occur in combustors that have anchoring mechanisms to prevent blow-off and flashback. The stabilization devices often introduce a level of geometric complexity that is prohibitive for detailed computational studies of turbulent flame dynamics. As a result, typical detailed simulations are performed in simplified model configurations such as decaying isotropic turbulence or inflowing turbulence. In these configurations, the turbulence seen by the flame either decays or, in the latter case, increases as the flame accelerates toward the turbulent inflow. This limits the duration of the eddy evolutions experienced by the flame at a given level of turbulent intensity, so that statistically valid observations cannot be made. In this paper, we apply a feedback control to computationally stabilize an otherwise unstable turbulent premixed flame in two dimensions. For the simulations, we specify turbulent in flow conditions and dynamically adjust the integrated fueling rate to control the mean location of the flame in the domain. We outline the numerical procedure, and illustrate the behavior of the control algorithm. We use the simulations to study the propagation and the local chemical variability of turbulent flame chemistry.

A solid state electrochemical sensor has been developed for hydrogen leak detection in ambient air. The sensor uses an yttria-stabilized electrolyte with a tin-doped indium oxide sensing electrode and a Pt reference electrode. Excellent sensitivity, and response time of one second or less, are reported for hydrogen gas over the concentration range of 0.03 to 5.5% in air. Cross-sensitivity to relative humidity and to CO{sub 2} are shown to be low. The response to methane, a potentially significant source of interference for such a sensor, is significantly less than that for hydrogen. The sensor shows good reproducibility and was unaffected by thermal cycling over the course of this investigation. The effects of sensing electrode thickness and thermal aging are also reported, and the sensing mechanism is discussed. The sensor is intended for use in vehicles powered by hydrogen fuel cells and hydrogen internal combustion engines. Those vehicles will use and/or store significant quantities of hydrogen, and will require safety sensor for monitoring potential hydrogen leakage in order to ensure passenger safety.

Perturbations on an interface driven by a strong blast wave grow in time due to a combination of Rayleigh-Taylor, Richtmyer-Meshkov, and decompression effects. In this paper, we present the first results from a computational study of such a system under drive conditions to be attainable on the National Ignition Facility. Using the multiphysics, AMR, higher order Godunov Eulerian hydrocode, Raptor, we consider the late nonlinear instability evolution for multiple amplitude and phase realizations of a variety of multimode spectral types. We show that compressibility effects preclude the emergence of a regime of self-similar instability growth independent of the initial conditions by allowing for memory of the initial conditions to be retained in the mix width at all times. The loss of transverse spectral information is demonstrated, however, along with the existence of a quasi-self-similar regime over short time intervals. Certain aspects of the initial conditions, including the rms amplitude, are shown to have a strong effect on the time to transition to the quasi-self-similar regime.

This article is a reply to a comment by A. Helmstetter and D. Sornette about the article 'Power-Law Time Distribution of Large Earthquakes' from 2003.