Minimizing buffer requirements under rate-optimal schedule in regular dataflow networks

Large-grain synchronous dataflow graphs or multi-rate graphs have the distinct feature that the nodes of the dataflow graph fire at different rates. Such multi-rate large-grain dataflow graphs have been widely regarded as a powerful programming model for DSP applications. In this paper we propose a method to minimize buffer storage requirement in constructing rate-optimal compile-time (MBRO) schedules for multi-rate dataflow graphs. We demonstrate that the constraints to minimize buffer storage while executing at the optimal computation rate (i.e. the maximum possible computation rate without storage constraints) can be formulated as a unified linear programming problem in our framework. A novel feature of our method is that in constructing the rate-optimal schedule, it directly minimizes the memory requirement by choosing the schedule time of nodes appropriately. Lastly, a new circular-arc interval graph coloring algorithm has been proposed to further reduce the memory requirement by allowing buffer sharing among the arcs of the multi-rate dataflow graph. We have constructed an experimental testbed which implements our MBRO scheduling algorithm as well as (i) the widely used periodic admissible parallel schedules (also known as block schedules) proposed by Lee and Messerschmitt (IEEE Transactions on Computers, vol. 36, no. 1, 1987, pp. 24-35), (ii) the optimal scheduling buffer allocation (OSBA) algorithm of Ning and Gao (Conference Record of the Twentieth Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages, Charleston, SC, Jan. 10-13, 1993, pp. 29-42), and (iii) the multi-rate software pipelining (MRSP) algorithm (Govindarajan and Gao, in Proceedings of the 1993 International Conference on Application Specific Array Processors, Venice, Italy, Oct. 25-27, 1993, pp. 77-88). Schedules generated for a number of random dataflow graphs and for a set of DSP application programs using the different scheduling methods are compared. The experimental results have demonstrated a significant improvement (10-20%) in buffer requirements for the MBRO schedules compared to the schedules generated by the other three methods, without sacrificing the computation rate. The MBRO method also gives a 20% average improvement in computation rate compared to Lee's Block scheduling method.

Structural investigation of combustion synthesized Cu/CeO2 catalysts by EXAFS and other physical techniques: Formation of a Ce1-xCuxO2-delta solid solution

The structure and chemical environment of Cu in Cu/CeO2 catalysts synthesized by the solution combustion method have been investigated by X-ray diffraction (XRD), transmission electron microscopy (TEM), electron paramagnetic resonance (EPR) spectroscopy, X-ray photoelectron spectroscopy (XPS), cyclic voltammetry (CV), and extended X-ray fine structure (EXAFS) spectroscopy. High-resolution XRD studies of 3 and 5 atom % Cu/CeO2 do not show CuO lines in their respective patterns. The structure could be refined for the composition Ce1-xCuxO2-delta (x = 0.03 and 0.05; delta similar to 0.13 and 0.16) in the fluorite structure with 5-8% oxide ion vacancy. High-resolution TEM did not show CuO particles in 5 atom % Cu/CeO2. EPR as well as XPS studies confirm the presence of Cu2+ species in the CeO2 matrix. Redox potentials of Cu species in the CeO2 matrix are lower than those in CuO. EXAFS investigations of these catalysts show an average coordination number of 3 around the Cu2+ ion in the first shell at a distance of 1.96 Angstrom, indicating the O2- ion vacancy around the Cu2+ ion. The Cu-O bond length also decreases compared to that in CuO. The second and third shell around the Cu2+ ion in the catalysts are attributed to -Cu2+-O2--Cu2+ - at 2.92 Angstrom and -Cu2+-O2--Ce4+- at the distance of 3.15 Angstrom, respectively. The present results provide direct evidence for the formation of a Ce1-xCuxO2-delta type of solid solution phase having -square-Cu2+-O-Ce4+- kind of linkages.

Studies on the effects of test gas on the flow field around large angle blunt cone flying at hypersonic Mach number

The effect of the test gas on the flow field around a 120degrees apex angle blunt cone has been investigated in a shock tunnel at a nominal Mach number of 5.75. The shock standoff distance around the blunt cone was measured by an electrical discharge technique using both carbon dioxide and air as test gases. The forebody laminar convective heat transfer to the blunt cone was measured with platinum thin-film sensors in both air and carbon dioxide environments. An increase of 10 to 15% in the measured heat transfer values was observed with carbon dioxide as the test gas in comparison to air. The measured thickness of the shock layer along the stagnation streamline was 3.57 +/- 0.17 mm in air and 3.29 +/- 0.26 mm in carbon dioxide. The computed thickness of the shock layer for air and carbon dioxide were 3.98 mm and 3.02 mm, respectively. The observed increase in the measured heat transfer rates in carbon dioxide compared to air was due to the higher density ratio across the bow shock wave and the reduced shock layer thickness.

Systematic trends in structural and thermodynamic properties of ternary oxides in the systems Ln-Pd-O (Ln = lanthanide element)

Isothermal sections of the phase diagrams for the systems Ln-Pd-O (Ln = lanthanide element) at 1223 K indicate the presence of two inter-oxide compounds Ln(4)PdO(7) and Ln(2)Pd(2)O(5) for Ln = La, Pr, Nd, Sm, three compounds Ln(4)PdO(7), Ln(2)PdO(4) and Ln(2)Pd(2)O(5) for Ln = Eu, Gd and only one compound of Ln(2)Pd(2)O(5) for Ln = Tb to Ho. The lattice parameters of the compounds Ln(4)PdO(7), Ln(2)PdO(4) and Ln(2)Pd(2)O(5) show systematic nonlinear variation with atomic number. The unit cell volumes decrease with increasing atomic number. The standard Gibbs energies, enthalpies and entropies of formation of the ternary oxides from their component binary oxides (Ln(2)O(3) and PdO) have been measured recently using an advanced version of the solid-state electrochemical cell. The Gibbs energies and enthalpies of formation become less negative with increasing atomic number of Ln. For all the three compounds, the variation in Gibbs energy and enthalpy of formation with atomic number is markedly non-linear. The decrease in stability with atomic number is most pronounced for Ln(2)Pd(2)O(5), followed by Ln(4)PdO(7) and Ln(2)PdO(4). This is probably related to the repulsion between Pd2+ ions on the opposite phases Of O-8 cubes in Ln(2)Pd(2)O(5), and the presence of Ln-filled O-8 cubes that share three faces with each other in Ln4PdO7. The values for entropy of formation of all the ternary oxides from their component binary oxides are relatively small. Although the entropies of formation show some scatter, the average value for Ln = La, Pr, Nd is more negative than the average value for the other lanthanide elements. From this difference, an average value for the structure transformation entropy of Ln(2)O(3) from C-type to A-type is estimated as 0.87 J.mol(-1).K-1. The standard Gibbs energies of formation of these ternary oxides from elements at 1223 K are presented as a function of lanthanide atomic number. By invoking the Neumann-Kopp rule for heat capacity, thermodynamic properties of the inter-oxide compounds at 298.15 K are estimated. (C) 2002 Elsevier Science Ltd. All rights reserved.

Tailoring the microstructure and mechanical properties of arrays of aligned multiwall carbon nanotubes by utilizing different hydrogen concentrations during synthesis

We synthesize vertically aligned arrays of carbon nanotubes (CNTs) in a chemical vapor deposition system with floating catalyst, using different concentrations of hydrogen in the gas feedstock. We report the effect of different hydrogen concentrations on the microstructure and mechanical properties of the resulting material. We show that a lower hydrogen concentration during synthesis results in the growth of stiffer CNT arrays with higher average bulk density. A lower hydrogen concentration also leads to the synthesis of CNT arrays that can reach higher peak stress at maximum compressive strain, and dissipate a larger amount of energy during compression. The individual CNTs in the arrays synthesized with a lower hydrogen concentration have, on average, larger outer diameters (associated with the growth of CNTs with a larger number of walls), but present a less uniform diameter distribution. The overall heights of the arrays and their strain recovery after compression have been found to be independent of the hydrogen concentration during growth. (C) 2011 Elsevier Ltd. All rights reserved.

An experimental study of reverse transition in two-dimensional channel flow

An experimental investigation on reverse transition from turbulent to laminar flow in a two-dimensional channel was carried out. The reverse transition occurred when Reynolds number of an initially turbulent flow was reduced below a certain value by widening the duct in the lateral direction. The experiments were conducted at Reynolds numbers of 625, 865, 980 and 1250 based on half the height of the channel and the average of the mean velocity. At all these Reynolds numbers the initially turbulent mean velocity profiles tend to become parabolic. The longitudinal and vertical velocity fluctuations ($\overline{u^{\prime 2}}$ and $\overline{v^{\prime 2}}$) averaged over the height of the channel decrease exponentially with distance downstream, but $\overline{u^{\prime}v^{\prime}} $ tends to become zero at a reasonably well-defined point. During reverse transition $\overline{u^{\prime}}\overline{v^{\prime}}/\sqrt{\overline{u^{\prime 2}}}\sqrt{\overline{v^{\prime 2}}}$ also decreases as the flow moves downstream and Lissajous figures taken with u’ and v’ signals confirm this trend. There is approximate similarly between $\overline{u^{\prime 2}} $ profiles if the value of $\overline{u^{\prime 2}_{\max}} $ and the distance from the wall at which it occurs are taken as the reference scales. The spectrum of $\overline{u^{\prime 2}} $ is almost similar at all stations and the non-dimensional spectrum is exponential in wave-number. All the turbulent quantities, when plotted in appropriate co-ordinates, indicate that there is a definite critical Reynolds number of 1400±50 for reverse transition.

Particle dynamics in a turbulent particle–gas suspension at high Stokes number. Part 2. The fluctuating-force model

A fluctuating-force model is developed for representing the effect of the turbulent fluid velocity fluctuations on the particle phase in a turbulent gas–solid suspension in the limit of high Stokes number, where the particle relaxation time is large compared with the correlation time for the fluid velocity fluctuations. In the model, a fluctuating force is incorporated in the equation of motion for the particles, and the force distribution is assumed to be an anisotropic Gaussian white noise. It is shown that this is equivalent to incorporating a diffusion term in the Boltzmann equation for the particle velocity distribution functions. The variance of the force distribution, or equivalently the diffusion coefficient in the Boltzmann equation, is related to the time correlation functions for the fluid velocity fluctuations. The fluctuating-force model is applied to the specific case of a Couette flow of a turbulent particle–gas suspension, for which both the fluid and particle velocity distributions were evaluated using direct numerical simulations by Goswami & Kumaran (2010). It is found that the fluctuating-force simulation is able to quantitatively predict the concentration, mean velocity profiles and the mean square velocities, both at relatively low volume fractions, where the viscous relaxation time is small compared with the time between collisions, and at higher volume fractions, where the time between collisions is small compared with the viscous relaxation time. The simulations are also able to predict the velocity distributions in the centre of the Couette, even in cases in which the velocity distribution is very different from a Gaussian distribution.

A dynamical instability due to fluid–wall coupling lowers the transition Reynolds number in the flow through a flexible tube

A flow-induced instability in a tube with flexible walls is studied experimentally. Tubes of diameter 0.8 and 1.2 mm are cast in polydimethylsiloxane (PDMS) polymer gels, and the catalyst concentration in these gels is varied to obtain shear modulus in the range 17–550 kPa. A pressure drop between the inlet and outlet of the tube is used to drive fluid flow, and the friction factor $f$ is measured as a function of the Reynolds number $Re$. From these measurements, it is found that the laminar flow becomes unstable, and there is a transition to a more complicated flow profile, for Reynolds numbers as low as 500 for the softest gels used here. The nature of the $f$–$Re$ curves is also qualitatively different from that in the flow past rigid tubes; in contrast to the discontinuous increase in the friction factor at transition in a rigid tube, it is found that there is a continuous increase in the friction factor from the laminar value of $16\ensuremath{/} Re$ in a flexible tube. The onset of transition is also detected by a dye-stream method, where a stream of dye is injected into the centre of the tube. It is found that there is a continuous increase of the amplitude of perturbations at the onset of transition in a flexible tube, in contrast to the abrupt disruption of the dye stream at transition in a rigid tube. There are oscillations in the wall of the tube at the onset of transition, which is detected from the laser scattering off the walls of the tube. This indicates that the coupling between the fluid stresses and the elastic stresses in the wall results in an instability of the laminar flow.

Stability of the flow of a fluid through a flexible tube at high Reynolds number

The stability of the Hagen-Poiseuille flow of a Newtonian fluid in a tube of radius R surrounded by an incompressible viscoelastic medium of radius R < r < HR is analysed in the high Reynolds number regime. The dimensionless numbers that affect the fluid flow are the Reynolds number Re = (ρVR / η), the ratio of the viscosities of the wall and fluid ηr = (ηs/η), the ratio of radii H and the dimensionless velocity Γ = (ρV2/G)1/2. Here ρ is the density of the fluid, G is the coefficient of elasticity of the wall and Vis the maximum fluid velocity at the centre of the tube. In the high Reynolds number regime, an asymptotic expansion in the small parameter ε = (1/Re) is employed. In the leading approximation, the viscous effects are neglected and there is a balance between the inertial stresses in the fluid and the elastic stresses in the medium. There are multiple solutions for the leading-order growth rate do), all of which are imaginary, indicating that the fluctuations are neutrally stable, since there is no viscous dissipation of energy or transfer of energy from the mean flow to the fluctruations due to the Reynolds strees. There is an O(ε1/2) correction to the growth rate, s(1), due to the presence of a wall layer of thickness ε1/2R where the viscous stresses are O(ε1/2) smaller than the inertial stresses. An energy balance analysis indicates that the transfer of energy from the mean flow to the fluctuations due to the Reynolds stress in the wall layer is exactly cancelled by an opposite transfer of equal magnitude due to the deformation work done at the interface, and there is no net transfer from the mean flow to the fluctuations. Consequently, the fluctuations are stabilized by the viscous dissipation in the wall layer, and the real part of s(1) is negative. However, there are certain values of Γ and wavenumber k where s(l) = 0. At these points, the wail layer amplitude becomes zero because the tangential velocity boundary condition is identically satisfied by the inviscid flow solution. The real part of the O(ε) correction to the growth rate s(2) turns out to be negative at these points, indicating a small stabilizing effect due to the dissipation in the bulk of the fluid and the wall material. It is found that the minimum value of s(2) increases [is proportional to] (H − 1)−2 for (H − 1) [double less-than sign] 1 (thickness of wall much less than the tube radius), and decreases [is proportional to] (H−4 for H [dbl greater-than sign] 1. The damping rate for the inviscid modes is smaller than that for the viscous wall and centre modes in a rigid tube, which have been determined previously using a singular perturbation analysis. Therefore, these are the most unstable modes in the flow through a flexible tube.

Particle dynamics in a turbulent particle–gas suspension at high Stokes number. Part 1. Velocity and acceleration distributions

The effect of fluid velocity fluctuations on the dynamics of the particles in a turbulent gas–solid suspension is analysed in the low-Reynolds-number and high Stokes number limits, where the particle relaxation time is long compared with the correlation time for the fluid velocity fluctuations, and the drag force on the particles due to the fluid can be expressed by the modified Stokes law. The direct numerical simulation procedure is used for solving the Navier–Stokes equations for the fluid, the particles are modelled as hard spheres which undergo elastic collisions and a one-way coupling algorithm is used where the force exerted by the fluid on the particles is incorporated, but not the reverse force exerted by the particles on the fluid. The particle mean and root-mean-square (RMS) fluctuating velocities, as well as the probability distribution function for the particle velocity fluctuations and the distribution of acceleration of the particles in the central region of the Couette (where the velocity profile is linear and the RMS velocities are nearly constant), are examined. It is found that the distribution of particle velocities is very different from a Gaussian, especially in the spanwise and wall-normal directions. However, the distribution of the acceleration fluctuation on the particles is found to be close to a Gaussian, though the distribution is highly anisotropic and there is a correlation between the fluctuations in the flow and gradient directions. The non-Gaussian nature of the particle velocity fluctuations is found to be due to inter-particle collisions induced by the large particle velocity fluctuations in the flow direction. It is also found that the acceleration distribution on the particles is in very good agreement with the distribution that is calculated from the velocity fluctuations in the fluid, using the Stokes drag law, indicating that there is very little correlation between the fluid velocity fluctuations and the particle velocity fluctuations in the presence of one-way coupling. All of these results indicate that the effect of the turbulent fluid velocity fluctuations can be accurately represented by an anisotropic Gaussian white noise.

Measurement of Heat Transfer Rate on Backward-Facing Steps at Hypersonic Mach Number

Two backward-facing models with step heights of 2 and 3 mm are used to measure the convective surface heat transfer rates by using platinum thin-film gauges, deposited on Macor inserts. Heat transfer rates have been theoretically calculated along the flat plate portion of a model using the Eckert reference temperature method. The experimentally determined surface heat transfer rate distributions are compared with theoretical and numerical estimations. Experimental heat flux distribution over a flat plate model showed good agreement with the reference temperature method at stagnation enthalpy range of 0.8-2 MJ/kg. Theoretical analysis has been used for downstream of a backward-facing step using Gai's nondimensional analysis. It has been found from the present study that approximately 10 and 8 step heights are required for the flow to reattach for 2 and 3 mm step height backward-facing step models, respectively, at a nominal Mach number of 7.6.

Plume structure in high-Rayleigh-number convection

Near-wall structures in turbulent natural convection at Rayleigh numbers of $10^{10}$ to $10^{11}$ at A Schmidt number of 602 are visualized by a new method of driving the convection across a fine membrane using concentration differences of sodium chloride. The visualizations show the near-wall flow to consist of sheet plumes. A wide variety of large-scale flow cells, scaling with the cross-section dimension, are observed. Multiple large-scale flow cells are seen at aspect ratio (AR)= 0.65, while only a single circulation cell is detected at AR= 0.435. The cells (or the mean wind) are driven by plumes coming together to form columns of rising lighter fluid. The wind in turn aligns the sheet plumes along the direction of shear. the mean wind direction is seen to change with time. The near-wall dynamics show plumes initiated at points, which elongate to form sheets and then merge. Increase in rayleigh number results in a larger number of closely and regularly spaced plumes. The plume spacings show a common log–normal probability distribution function, independent of the rayleigh number and the aspect ratio. We propose that the near-wall structure is made of laminar natural-convection boundary layers, which become unstable to give rise to sheet plumes, and show that the predictions of a model constructed on this hypothesis match the experiments. Based on these findings, we conclude that in the presence of a mean wind, the local near-wall boundary layers associated with each sheet plume in high-rayleigh-number turbulent natural convection are likely to be laminar mixed convection type.

Defining the ‘modified Griffin plot’ in vortex-induced vibration: revealing the effect of Reynolds number using controlled damping

In the present work, we study the transverse vortex-induced vibrations of an elastically mounted rigid cylinder in a fluid flow. We employ a technique to accurately control the structural damping, enabling the system to take on both negative and positive damping. This permits a systematic study of the effects of system mass and damping on the peak vibration response. Previous experiments over the last 30 years indicate a large scatter in peak-amplitude data ($A^*$) versus the product of mass–damping ($\alpha$), in the so-called ‘Griffin plot’. A principal result in the present work is the discovery that the data collapse very well if one takes into account the effect of Reynolds number ($\mbox{\textit{Re}}$), as an extra parameter in a modified Griffin plot. Peak amplitudes corresponding to zero damping ($A^*_{{\alpha}{=}0}$), for a compilation of experiments over a wide range of $\mbox{\textit{Re}}\,{=}\,500-33000$, are very well represented by the functional form $A^*_{\alpha{=}0} \,{=}\, f(\mbox{\textit{Re}}) \,{=}\, \log(0.41\,\mbox{\textit{Re}}^{0.36}$). For a given $\mbox{\textit{Re}}$, the amplitude $A^*$ appears to be proportional to a function of mass–damping, $A^*\propto g(\alpha)$, which is a similar function over all $\mbox{\textit{Re}}$. A good best-fit for a wide range of mass–damping and Reynolds number is thus given by the following simple expression, where $A^*\,{=}\, g(\alpha)\,f(\mbox{\textit{Re}})$: \[ A^* \,{=}\,(1 - 1.12\,\alpha + 0.30\,\alpha^2)\,\log (0.41\,\mbox{\textit{Re}}^{0.36}). \] In essence, by using a renormalized parameter, which we define as the ‘modified amplitude’, $A^*_M\,{=}\,A^*/A^*_{\alpha{=}0}$, the previously scattered data collapse very well onto a single curve, $g(\alpha)$, on what we refer to as the ‘modified Griffin plot’. There has also been much debate over the last three decades concerning the validity of using the product of mass and damping (such as $\alpha$) in these problems. Our results indicate that the combined mass–damping parameter ($\alpha$) does indeed collapse peak-amplitude data well, at a given $\mbox{\textit{Re}}$, independent of the precise mass and damping values, for mass ratios down to $m^*\,{=}\,1$.

Identity, energetics, dynamics and environment of interfacial water molecules in a micellar solution

The structure and energetics of interfacial water molecules in the aqueous micelle of cesium perfluorooctanoate have been investigated, using large-scale atomistic molecular dynamics simulations, with the primary objective of classifying them. The simulations show that the water molecules at the interface fall into two broad classes: bound and free, present in a ratio of 9:1. The bound water molecules can be further categorized on the basis of the number of hydrogen bonds (one or two) that they form with the surfactant headgroups. The hydrogen bonds of the doubly hydrogen-bonded species are found to be, on the average, slightly weaker than those in the singly bonded species. The environment around interfacial water molecules is more ordered than that in the bulk. The surface water molecules have substantially lower potential energy, because of interaction with the micelle. In particular, both forms of bound water have energies that are lower by �2.5-4.0 kcal/ mol. Entropy is found to play an important role in determining the relative concentration of the species.