On the necessity of explicit cross-layer data formats in near-data processing systems

Distributed and Parallel Databases - Massive data transfers in modern data-intensive systems resulting from low data-locality and data-to-code system design hurt their performance and scalability....     

CeO<Subscript>2</Subscript> and Co<Subscript>3</Subscript>O<Subscript>4</Subscript>–CeO<Subscript>2</Subscript> nanoparticles: effect of the synthesis method on the structure and catalytic properties in COPrOx and methanation reactions

CeO₂ and Co₃O₄–CeO₂ nanoparticles were synthesized, thoroughly characterized, and evaluated in the COPrOx reaction. The CeO₂ nanoparticles were synthesized by the diffusion-controlled precipitation method with ethylene glycol. A notably higher yield was obtained when H₂O₂ was used in the synthesis procedure. For comparison, two commercial samples of CeO₂ nanoparticles (Nyacol^®)—one calcined and the other sintered—were also studied. Catalytic results of bare CeO₂ calcined at 500 °C showed a strong influence of the method of synthesis. Despite having similar BET area values, the CeO₂ synthesized without H₂O₂ was the most active sample. Co₃O₄–CeO₂ catalysts with three different Co/(Co + Ce) atomic ratios, 0.1, 0.3, and 0.5, were prepared by the wet impregnation of the CeO₂ nanoparticles. TEM and STEM observations showed that impregnation produced mixed oxides composed of small CeO₂ nanoparticles located both over the surface and inside the Co₃O₄ crystals. The mixed oxide catalysts prepared with a cobalt atomic ratio of 0.5 showed methane formation, which started at 200 °C due to the reaction between CO₂ and H₂. However, above 250 °C, the reaction between CO and H₂ became important, thus contributing to CO elimination with a small H₂ loss. As a result, CO could be totally eliminated in a wide temperature range, from 200 to 400 °C. The methanation reaction was favored by the reduction of the cobalt oxide, as suggested by the TPR experiments. This result is probably originated in Ce–Co interactions, related to the method of synthesis and the surface area of the mixed oxides obtained.     

Bulk mass flow in a vibratory finisher: mechanisms and effect of process parameters

Vibratory finishing is a widely-used manufacturing process in which a vibrating container filled with granular media becomes fluidized. The resulting bulk flow entrains workpieces and exposes their surfaces to the impacts resulting from the small-scale media vibrations. The bulk flow is responsible for entrainment and mixing, while the media vibration does work on the surfaces. The selection of machine vibration parameters is commonly based on experience due to the difficulty in predicting the fluidized bed behavior. In this work, a discrete element method was used to investigate how the bulk flow in an actual tub finisher filled with steel balls depends on the tub motion parameters through a parametric study. The underlying mechanisms that create and drive the bulk flow were identified by examining the relationships between the bulk flow rates and the wall forces. Finally, the connection between the wall motion and the wall forces was investigated. The tub frequency was the most effective control parameter and there was an optimal phase difference between the horizontal and vertical vibrations to maximize bulk flow. The relationship between the media packing at the walls and the tangential forces between the walls and the media explained the formation and speed of the bulk flow. Lastly, it was shown that the tangential wall forces, unlike the normal forces, cannot be obtained from the known wall motion alone since they also depend on the media velocities relative to the walls.     

Influence of process parameters on average particle speeds in a vibratory finisher

Vibratory finishing (VF) employs vibrationally-fluidized granular media to finish the surfaces of workpieces that are entrained in the flowing media. Its application has been based mostly on experience and trial-and-error due to the complexity of the granular material behavior. The present study used discrete element modeling (DEM) to investigate how the movement of a commercial two-dimensional tub finisher influenced the average particle speed of the media in a bed of smooth, steel, spherical particles, and thus the work that would be done on an entrained workpiece. The parameters governing the tub wall motion (frequency, in-plane amplitudes, and phases of vibration) and the coefficient of friction between the media and the wall were systematically varied in 71 three-dimensional DEM simulations. The average particle speed was affected mostly by the vertical amplitude of tub motion rather than by the frequency, and was mostly independent of other parameters of motion and of the wall friction. A strong relationship was found between the average particle speed and the work done by the wall per cycle of vibration. The normal force on the wall was also found to correlate strongly with the normal component of the wall velocity. Together, these relationships offer the potential to enable the analytical prediction of the average particle speed based on the motion parameters of the tub alone. The paper provides a set of practical guidelines for the control of the average particle speed in VF that are explained by the forces between the media and walls of the tub finisher.     

Nonlinear Versus Linear Learning Devices: A Procedural Perspective

We provide a discussion of bounded rationality learning behind traditional learning mechanisms, i.e., Recursive Ordinary Least Squares and Bayesian Learning . These mechanisms lack for many reasons a behavioral interpretation and, following the Simon criticism, they appear to be substantively rational. In this paper, analyzing the Cagan model, we explore two learning mechanisms which appear to be more plausible from a behavioral point of view and somehow procedurally rational: Least Mean Squares learning for linear models and Back Propagation for Artificial Neural Networks . The two algorithms look for a minimum of the variance of the error forecasting by means of a steepest descent gradient procedure. The analysis of the Cagan model shows an interesting result: non-convergence of learning to the Rational Expectations Equilibrium is not due to the restriction to linear learning devices; also Back Propagation learning for Artificial Neural Networks may fail to converge to the Rational Expectations Equilibrium of the model.     

Genetic programming with one-point crossover and subtree mutation for effective problem solving and bloat control

Genetic programming (GP) is one of the most widely used paradigms of evolutionary computation due to its ability to automatically synthesize computer programs and mathematical expressions. However, because GP uses a variable length representation, the individuals within the evolving population tend to grow rapidly without a corresponding return in fitness improvement, a phenomenon known as bloat. In this paper, we present a simple bloat control strategy for standard tree-based GP that achieves a one order of magnitude reduction in bloat when compared with standard GP on benchmark tests, and practically eliminates bloat on two real-world problems. Our proposal is to substitute standard subtree crossover with the one-point crossover (OPX) developed by Poli and Langdon (Second online world conference on soft computing in engineering design and manufacturing, Springer, Berlin (1997)), while maintaining all other GP aspects standard, particularly subtree mutation. OPX was proposed for theoretical purposes related to GP schema theorems, however since it curtails exploration during the search it has never achieved widespread use. In our results, on the other hand, we are able to show that OPX can indeed perform an effective search if it is coupled with subtree mutation, thus combining the bloat control capabilities of OPX with the exploration provided by standard mutation.