Evaluation of dynamic voltage and frequency scaling for stream programs

Dynamic Voltage and Frequency Scaling (DVFS) offers a huge potential for designing trade-offs involving energy, power, temperature and performance of computing systems. In this paper, we evaluate three different DVFS schemes - our enhancement of a Petri net performance model based DVFS method for sequential programs to stream programs, a simple profile based Linear Scaling method, and an existing hardware based DVFS method for multithreaded applications - using multithreaded stream applications, in a full system Chip Multiprocessor (CMP) simulator. From our evaluation, we find that the software based methods achieve significant Energy/Throughput2(ET−2) improvements. The hardware based scheme degrades performance heavily and suffers ET−2 loss. Our results indicate that the simple profile based scheme achieves the benefits of the complex Petri net based scheme for stream programs, and present a strong case for the need for independent voltage/frequency control for different cores of CMPs, which is lacking in most of the state-of-the-art CMPs. This is in contrast to the conclusions of a recent evaluation of per-core DVFS schemes for multithreaded applications for CMPs.

Limit laws for coverage in backbone-sensor networks

We study coverage in sensor networks having two types of nodes, namely, sensor nodes and backbone nodes. Each sensor is capable of transmitting information over relatively small distances. The backbone nodes collect information from the sensors. This information is processed and communicated over an ad hoc network formed by the backbone nodes, which are capable of transmitting over much larger distances. We consider two models of deployment for the sensor and backbone nodes. One is a PoissonPoisson cluster model and the other a dependently thinned Poisson point process. We deduce limit laws for functionals of vacancy in both models using properties of association for random measures.

Scaling of turbulent flame speed for expanding flames with Markstein diffusion considerations

In this paper we clarify the role of Markstein diffusivity, which is the product of the planar laminar flame speed and the Markstein length, on the turbulent flame speed and its scaling, based on experimental measurements on constant-pressure expanding turbulent flames. Turbulent flame propagation data are presented for premixed flames of mixtures of hydrogen, methane, ethylene, n-butane, and dimethyl ether with air, in near-isotropic turbulence in a dual-chamber, fan-stirred vessel. For each individual fuel-air mixture presented in this work and the recently published iso-octane data from Leeds, normalized turbulent flame speed data of individual fuel-air mixtures approximately follow a Re-T,f(0.5) scaling, for which the average radius is the length scale and thermal diffusivity is the transport property of the turbulence Reynolds number. At a given Re-T,Re-f, it is experimentally observed that the normalized turbulent flame speed decreases with increasing Markstein number, which could be explained by considering Markstein diffusivity as the leading dissipation mechanism for the large wave number flame surface fluctuations. Consequently, by replacing thermal diffusivity with the Markstein diffusivity in the turbulence Reynolds number definition above, it is found that normalized turbulent flame speeds could be scaled by Re-T,M(0.5) irrespective of the fuel, equivalence ratio, pressure, and turbulence intensity for positive Markstein number flames.

A divide and conquer strategy for scaling weather simulations with multiple regions of interest

Accurate and timely prediction of weather phenomena, such as hurricanes and flash floods, require high-fidelity compute intensive simulations of multiple finer regions of interest within a coarse simulation domain. Current weather applications execute these nested simulations sequentially using all the available processors, which is sub-optimal due to their sub-linear scalability. In this work, we present a strategy for parallel execution of multiple nested domain simulations based on partitioning the 2-D processor grid into disjoint rectangular regions associated with each domain. We propose a novel combination of performance prediction, processor allocation methods and topology-aware mapping of the regions on torus interconnects. Experiments on IBM Blue Gene systems using WRF show that the proposed strategies result in performance improvement of up to 33% with topology-oblivious mapping and up to additional 7% with topology-aware mapping over the default sequential strategy.

Terminal Impact Angle Constrained Guidance Laws Using Variable Structure Systems Theory

In this brief, variable structure systems theory based guidance laws, to intercept maneuvering targets at a desired impact angle, are presented. Choosing the missile's lateral acceleration (latax) to enforce sliding mode, which is the principal operating mode of variable structure systems, on a switching surface defined by the line-of-sight angle leads to a guidance law that allows the achievement of the desired terminal impact angle. As will be shown, this law does not ensure interception for all states of the missile and the target during the engagement. Hence, additional switching surfaces are designed and a switching logic is developed that allows the latax to switch between enforcing sliding mode on one of these surfaces so that the target can be intercepted at the desired impact angle. The guidance laws are designed using nonlinear engagement dynamics for the general case of a maneuvering target.

Turbulence in the two-dimensional Fourier-truncated Gross-Pitaevskii equation

We undertake a systematic, direct numerical simulation of the twodimensional, Fourier-truncated, Gross-Pitaevskii equation to study the turbulent evolutions of its solutions for a variety of initial conditions and a wide range of parameters. We find that the time evolution of this system can be classified into four regimes with qualitatively different statistical properties. Firstly, there are transients that depend on the initial conditions. In the second regime, powerlaw scaling regions, in the energy and the occupation-number spectra, appear and start to develop; the exponents of these power laws and the extents of the scaling regions change with time and depend on the initial condition. In the third regime, the spectra drop rapidly for modes with wave numbers k > kc and partial thermalization takes place for modes with k < kc; the self-truncation wave number kc(t) depends on the initial conditions and it grows either as a power of t or as log t. Finally, in the fourth regime, complete thermalization is achieved and, if we account for finite-size effects carefully, correlation functions and spectra are consistent with their nontrivial Berezinskii-Kosterlitz-Thouless forms. Our work is a natural generalization of recent studies of thermalization in the Euler and other hydrodynamical equations; it combines ideas from fluid dynamics and turbulence, on the one hand, and equilibrium and nonequilibrium statistical mechanics on the other.

On the k(1)(-1) scaling in sink-flow turbulent boundary layers

Scaling of the streamwise velocity spectrum phi(11)(k(1)) in the so-called sink-flow turbulent boundary layer is investigated in this work. The present experiments show strong evidence for the k(1)(-1) scaling i.e. phi(11)(k(1)) = Lambda(1)U(tau)(2)k(1)(-1), where k(1)(-1) is the streamwise wavenumber and U-tau is the friction velocity. Interestingly, this k(1)(-1) scaling is observed much farther from the wall and at much lower flow Reynolds number (both differing by almost an order of magnitude) than what the expectations from experiments on a zero-pressure-gradient turbulent boundary layer flow would suggest. Furthermore, the coefficient A(1) in the present sink-flow data is seen to be non-universal, i.e. A(1) varies with height from the wall; the scaling exponent -1 remains universal. Logarithmic variation of the so-called longitudinal structure function, which is the physical-space counterpart of spectral k(1)(-1) scaling, is also seen to be non-universal, consistent with the non-universality of A(1). These observations are to be contrasted with the universal value of A(1) (along with the universal scaling exponent of 1) reported in the literature on zero-pressure-gradient turbulent boundary layers. Theoretical arguments based on dimensional analysis indicate that the presence of a streamwise pressure gradient in sink-flow turbulent boundary layers makes the coefficient A(1) non-universal while leaving the scaling exponent -1 unaffected. This effect of the pressure gradient on the streamwise spectra, as discussed in the present study (experiments as well as theory), is consistent with other recent studies in the literature that are focused on the structural aspects of turbulent boundary layer flows in pressure gradients (Harun etal., J. Flui(d) Mech., vol. 715, 2013, pp. 477-498); the present paper establishes the link between these two. The variability of A(1) accommodated in the present framework serves to clarify the ideas of universality of the k(1)(-1) scaling.

Sliding mode control based terminal impact angle constrained guidance laws using dual sliding surfaces

In this paper, sliding mode control theory based guidance laws to intercept non-maneuvering targets at a desired impact angle are presented. The desired impact angle, defined in terms of a desired line-of-sight (LOS) angle, is achieved by selecting the missile's lateral acceleration (latax) to enforce sliding mode on a sliding surface based on this LOS angle. As will be shown, this guidance law does not ensure interception for all states of the missile and the target during the engagement. Hence, to satisfy the requirement of interception at the desired impact angle, a second sliding surface is designed and a switching logic, based on the conditions necessary for interception, is presented that allows the latax to switch between enforcing sliding mode on one of these surfaces so that the target can be intercepted at the desired impact angle. The guidance laws are designed using non-linear engagement dynamics.

Nonlinear gravity-wave interactions in stratified turbulence

To investigate the dynamics of gravity waves in stratified Boussinesq flows, a model is derived that consists of all three-gravity-wave-mode interactions (the GGG model), excluding interactions involving the vortical mode. The GGG model is a natural extension of weak turbulence theory that accounts for exact three-gravity-wave resonances. The model is examined numerically by means of random, large-scale, high-frequency forcing. An immediate observation is a robust growth of the so-called vertically sheared horizontal flow (VSHF). In addition, there is a forward transfer of energy and equilibration of the nonzero-frequency (sometimes called ``fast'') gravity-wave modes. These results show that gravity-wave-mode interactions by themselves are capable of systematic interscale energy transfer in a stratified fluid. Comparing numerical simulations of the GGG model and the full Boussinesq system, for the range of Froude numbers (Fr) considered (0.05 a parts per thousand currency sign Fr a parts per thousand currency sign 1), in both systems the VSHF is hardest to resolve. When adequately resolved, VSHF growth is more vigorous in the GGG model. Furthermore, a VSHF is observed to form in milder stratification scenarios in the GGG model than the full Boussinesq system. Finally, fully three-dimensional nonzero-frequency gravity-wave modes equilibrate in both systems and their scaling with vertical wavenumber follows similar power-laws. The slopes of the power-laws obtained depend on Fr and approach -2 (from above) at Fr = 0.05, which is the strongest stratification that can be properly resolved with our computational resources.

Scaling of pressure spectrum in turbulent boundary layers

Scaling of pressure spectrum in zero-pressure-gradient turbulent boundary layers is discussed. Spatial DNS data of boundary layer at one time instant (Re-theta = 4500) are used for the analysis. It is observed that in the outer regions the pressure spectra tends towards the -7/3 law predicted by Kolmogorov's theory of small-scale turbulence. The slope in the pressure spectra varies from -1 close to the wall to a value close to -7/3 in the outer region. The streamwise velocity spectra also show a -5/3 trend in the outer region of the flow. The exercise carried out to study the amplitude modulation effect of the large scales on the smaller ones in the near-wall region reveals a strong modulation effect for the streamwise velocity, but not for the pressure fluctuations. The skewness of the pressure follows the same trend as the amplitude modulation coefficient, as is the case for the velocity. In the inner region, pressure spectra were seen to collapse better when normalized with the local Reynolds stress (-(u'v') over bar) than when scaled with the local turbulent kinetic energy (q(2) = (u'(2)) over bar + (v'(2)) over bar + (w'(2)) over bar)

Generalized scaling of misorientation angle distributions at meso-scale in deformed materials

Scaling behaviour has been observed at mesoscopic level irrespective of crystal structure, type of boundary and operative micro-mechanisms like slip and twinning. The presence of scaling at the meso-scale accompanied with that at the nano-scale clearly demonstrates the intrinsic spanning for different deformation processes and a true universal nature of scaling. The origin of a 1/2 power law in deformation of crystalline materials in terms of misorientation proportional to square root of strain is attributed to importance of interfaces in deformation processes. It is proposed that materials existing in three dimensional Euclidean spaces accommodate plastic deformation by one dimensional dislocations and their interaction with two dimensional interfaces at different length scales. This gives rise to a 1/2 power law scaling in materials. This intrinsic relationship can be incorporated in crystal plasticity models that aim to span different length and time scales to predict the deformation response of crystalline materials accurately.

Finite size scaling in crossover among different random matrix ensembles in microscopic lattice models

Using numerical diagonalization we study the crossover among different random matrix ensembles (Poissonian, Gaussian orthogonal ensemble (GOE), Gaussian unitary ensemble (GUE) and Gaussian symplectic ensemble (GSE)) realized in two different microscopic models. The specific diagnostic tool used to study the crossovers is the level spacing distribution. The first model is a one-dimensional lattice model of interacting hard-core bosons (or equivalently spin 1/2 objects) and the other a higher dimensional model of non-interacting particles with disorder and spin-orbit coupling. We find that the perturbation causing the crossover among the different ensembles scales to zero with system size as a power law with an exponent that depends on the ensembles between which the crossover takes place. This exponent is independent of microscopic details of the perturbation. We also find that the crossover from the Poissonian ensemble to the other three is dominated by the Poissonian to GOE crossover which introduces level repulsion while the crossover from GOE to GUE or GOE to GSE associated with symmetry breaking introduces a subdominant contribution. We also conjecture that the exponent is dependent on whether the system contains interactions among the elementary degrees of freedom or not and is independent of the dimensionality of the system.

Scaling law characterizing the dynamics of the transition of HIV-1 to error catastrophe

Increasing the mutation rate, mu, of viruses above a threshold, mu(c), has been predicted to trigger a catastrophic loss of viral genetic information and is being explored as a novel intervention strategy. Here, we examine the dynamics of this transition using stochastic simulations mimicking within-host HIV-1 evolution. We find a scaling law governing the characteristic time of the transition: tau approximate to 0.6/(mu - mu(c)). The law is robust to variations in underlying evolutionary forces and presents guidelines for treatment of HIV-1 infection with mutagens. We estimate that many years of treatment would be required before HIV-1 can suffer an error catastrophe.

A recursive multi-scaling approach to regional flood frequency analysis

Scaling approaches are widely used by hydrologists for Regional Frequency Analysis (RFA) of floods at ungauged/sparsely gauged site(s) in river basins. This paper proposes a Recursive Multi-scaling (RMS) approach to RFA that overcomes limitations of conventional simple- and multi-scaling approaches. The approach involves identification of a separate set of attributes corresponding to each of the sites (being considered in the study area/region) in a recursive manner according to their importance, and utilizing those attributes to construct effective regional regression relationships to estimate statistical raw moments (SMs) of peak flows. The SMs are then utilized to arrive at parameters of flood frequency distribution and quantile estimate(s) corresponding to target return period(s). Effectiveness of the RMS approach in arriving at flood quantile estimates for ungauged sites is demonstrated through leave-one-out cross-validation experiment on watersheds in Indiana State, USA. Results indicate that the approach outperforms index-flood based Region-of-Influence approach, simple- and multi-scaling approaches and a multiple linear regression method. (C) 2015 Elsevier B.V. All rights reserved.