Microarchitecture sensitive empirical models for compiler optimizations.

This paper proposes the use of empirical modeling techniques for building microarchitecture sensitive models for compiler optimizations. The models we build relate program performance to settings of compiler optimization flags, associated heuristics and key microarchitectural parameters. Unlike traditional analytical modeling methods, this relationship is learned entirely from data obtained by measuring performance at a small number of carefully selected compiler/microarchitecture configurations. We evaluate three different learning techniques in this context viz. linear regression, adaptive regression splines and radial basis function networks. We use the generated models to a) predict program performance at arbitrary compiler/microarchitecture configurations, b) quantify the significance of complex interactions between optimizations and the microarchitecture, and c) efficiently search for'optimal' settings of optimization flags and heuristics for any given microarchitectural configuration. Our evaluation using benchmarks from the SPEC CPU2000 suits suggests that accurate models (< 5% average error in prediction) can be generated using a reasonable number of simulations. We also find that using compiler settings prescribed by a model-based search can improve program performance by as much as 19% (with an average of 9.5%) over highly optimized binaries.

Neutral-point balancing of neutral-point-clamped three-level inverter with a front end switched rectifier DC source for the full modulationn range

A switched rectifier DC voltage source three-level neutral-point-clamped (NPC) converter topology is proposed here to alleviate the inverter from capacitor voltage balancing in three-level drive systems. The proposed configuration requires only one DC link with a voltage of half of that needed in a conventional NPC inverter. To obtain a rated DC link voltage, the rectifier DC source is alternately connected in parallel to one of the two series capacitors using two switches and two diodes with device voltage ratings of half the total DC bus voltage. The frequency at which the voltage source is switched is independent of the inverter and will not affect its operation since the switched voltage source in this configuration balances the capacitors automatically. The proposed configuration can also be used as a conventional two-level inverter in the lower modulation index range, thereby increasing the reliability of the drivesystem. A space-vector-based PWM scheme is used to verify this proposed topology on a laboratory system.

On the performance analysis of a class of neuron circuits

A new performance metric, Peak-Error Ratio (PER) has been presented to benchmark the performance of a class of neuron circuits to realize neuron activation function (NAF) and its derivative (DNAF). Neuron circuits, biased in subthreshold region, based on the asymmetric cross-coupled differential pair configuration and conventional configuration of applying small external offset voltage at the input have been compared on the basis of PER. It is shown that the technique of using transistor asymmetry in a cross-coupled differential pair performs on-par with that of applying external offset voltage. The neuron circuits have been experimentally prototyped and characterized as a proof of concept on the 1.5 mu m AMI technology.

Studies on strontium titanate/barium zirconate superlattices

Artificial superlattices of SrTiO3 and BaZrO3 were grown epitaxially with different periodicities on SrRuO3 coated (00 1) SrTiO3 substrates by pulsed excimer laser ablation. Superlattices were structurally characterized by X-Ray theta-2 theta diffraction data. Electrical characterization was done in metal-insulation-metal configuration. Capacitance-voltage measurements showed limited amount of tunability. The DC field induced tunability has been observed to be sensitive to the periodicity of the superlattices, hence the effective strain present in the layers. Hysteretic behaviour in capacitance-voltage (C-V) and polarization versus electric field (P-E) results from the superlattices also indicate the sensitivity of the interfaces. Interfacial strain is supposed to be the most probable cause for such a behaviour which is also manifested in the variation of dielectric constant with individual layer thicknesses. (c) 2007 Elsevier Ltd. All rights reserved.

Synthesis of neoglycopeptides and analyses of their biodistributionin vivo to identify tissue specific uptake and novel putative membrane lectins

Complex typeN-linked oligosaccharides derived from fetuin, fibrinogen and thyroglobulin were coupled to acetyltyrosine affording a series of neoglycopeptides with retention of terminal structures and the beta-anomeric configuration of their reducing endN-acetylglycosamine residue. The neoglycopeptides thus synthesized could be labelled to high specific activities with125I in the aromatic side chain of tyrosine. Analysis of the fate of these neoglycopeptides in conjunction with inhibition with asialofetuin and oligosaccharides of defined structure in micein vivo revealed the uptake of galactosylated biantennary compound by kidneys, in addition to the known itinerary of triantennary galactosylated complex oligosaccharide from fetuin to liver and the galactosylated biantennary chain with fucosylation in the core to bone marrows. On the other hand, the agalacto, aglucosamino biantennary chains with and without fucosylation in the core region are taken up by submaxillary glands while the conserved trimannosyl core with fucose is primarily concentrated in stomach tissue. These studies thus define new routes for the uptake of complexN-linked glycans and also subserve to identify lectins presumably involved in their recognition.

On the performance analysis of a class of neuron circuits

A new performance metric, Peak-Error Ratio (PER) has been presented to benchmark the performance of a class of neuron circuits to realize neuron activation function (NAF) and its derivative (DNAF). Neuron circuits, biased in subthreshold region, based on the asymmetric cross-coupled differential pair configuration and conventional configuration of applying small external offset voltage at the input have been compared on the basis of PER. It is shown that the technique of using transistor asymmetry in a cross-coupled differential pair performs on-par with that of applying external offset voltage. The neuron circuits have been experimentally prototyped and characterized as a proof of concept on the 1.5 mu m AMI technology.

Fracture analysis of stiffened panels under combined tensile, bending, and shear loads

The objective is to present the formulation of numerically integrated modified virtual crack closure integral technique for concentrically and eccentrically stiffened panels for computation of strain-energy release rate and stress intensity factor based on linear elastic fracture mechanics principles. Fracture analysis of cracked stiffened panels under combined tensile, bending, and shear loads has been conducted by employing the stiffened plate/shell finite element model, MQL9S2. This model can be used to analyze plates with arbitrarily located concentric/eccentric stiffeners, without increasing the total number of degrees of freedom, of the plate element. Parametric studies on fracture analysis of stiffened plates under combined tensile and moment loads have been conducted. Based on the results of parametric,studies, polynomial curve fitting has been carried out to get best-fit equations corresponding to each of the stiffener positions. These equations can be used for computation of stress intensity factor for cracked stiffened plates subjected to tensile and moment loads for a given plate size, stiffener configuration, and stiffener position without conducting finite element analysis.

Dimethyl sulphoxide and dimethyl formamide complexes of Mn(III) perchlorate

Dimethyl sulphoxide (DMSO) and dimethyl formamide (DMF) complexes of Mn(III) perchlorate have been prepared and their conductivity, magnetic susceptibility and i.r. and electronic spectra studied. The complexes behave as uni-trivalent electrolytes in acetonitrile. Their magnetic moments of 5·1 B.M. show them to be of high spin type. Infra-red spectra show that oxygen is the donor atom in both complexes. The spin allowed electronic transition for d4 system, around 20,000 cm−1, ascribable to the 5Eg → 5T2g transition, suggests an octahedral configuration for these complexes

Syntheses of the trans,meso- and racemic C11 acids, degradation products of diterpene acids

Syntheses of the isomers of the C11 acid, 1(a),3(a)- dimethylcyclohexane-1 (e),2(e),3(e)-tricarboxylic acid (A) and 1(a),3(e)-dimethylcyclohexane-1(e),2(e),3(a)-tricarboxylic acid (B), the latter by two different routes, are reported. Two of the four possible isomers of the precursor triester, trimethyl 1-methylcyclohexane-1,2,3-tricarboxylate, on individual methylation followed by hydrolysis, gave the trans,meso-acid (A), identified by comparison with an authentic sample, and the cis,trans-form (B) whose structure and configuration were proved by comparison with a specimen obtained by the unambiguous and highly stereoselective second synthesis. This demonstrated that methylation of the triester isomers occurs stereospecifically and exclusively at C-3. In the second sequence, it has been possible to assign definite conformations to four key intermediates and the final product, directly from n.m.r. spectra, from changes in these spectra accompanying specific steps, and from chemical evidence. Comparison of the n.m.r. spectra of the isomeric triesters (A) and (B) has provided unequivocal proof of the accepted trans,meso configuration for the abietic acid degradation product (A).

Electric-polarization and structure of molecular complexes of the metallic halides of group II of the periodic table

Dielectric measurements have been made on a number of molecular complexes of beryllium, zinc, cadmium and mercuric halides. The polarizations observed have been interpreted in terms of a tetrahedral configuration for the undissociated beryllium, zinc and cadmium halide complexes. In other cases the observed polarization has been shown to be due to the dissociation of the complex in solution.

Studies in sesquiterpenes-XX. Acetoxymethylation of longifolene

Longifolene in Prins reaction with formaldehyde yielded the expected ω-acetoxymethyl longifolene, which was transformed into a number of interesting derivatives. Configuration of the Prins product has been arrived at by NMR measurements. The UV absorption of these derivatives show a considerable bathochromic shift with respect to those in the camphene series and this could be attributed to the slight twisting of the ethylenic linkage in longifolene and its derivatives.

Stereoselectivity in the addition of hydrogen chloride to cyclohex-1-enecarbonitrile: Possible role of chlorine lone pair-nitrile π-orbital pseudoallylic A1, 3 interaction

The (overall trans) addition of hydrogen chloride to cyclohex-1- enecarbonitrile in anhydrous alcoholic media proceeds to give cis-2-chlorocyclohexanecarboxylate (together with some cis-2- chlorocyclohexanecarboxamide): no corresponding products with the trans-configuration are detectable. In anhydrous ether the addition proceeds to give a single isomer, presumably cis-, of 2-chlorocyclohexanecarbonitrile, indicating that the configuration of the products may not be equilibrium-controlled in alcoholic media. An examination of the steric factors indicates that the transition state for protonation of the presumed intermediate, 2-chlorocyclohexylidenemethylideneimine, leading to cis-product is favoured if interaction between the lateral π-orbital of the C-N double bond and the lone-pairs on the chlorine atom at the 2-position is large. Consideration of interactions in the transition states meets Zimmerman's criticism that invoking A1, 3 interaction existing in ground states to explain product configuration takes insufficient account of the Curtin-Hammett principle.

Investigations of a two-stage pulse tube cryocooler operating down to 2.5K

A two-stage pulse tube cryocooler (PTC) which produces a no-load temperature of similar to 2.5 K in its second stage at an operating frequency of 1.6 Hz has been designed and fabricated. The second stage of the system provides a refrigeration power of similar to 250 mW at 5.0 K. The system uses stainless steel meshes (mesh size 200) along with lead (Pb) granules and combinations of Pb, Er3Ni, and HoCu2 as the first and second stage regenerator materials, respectively. Experimental studies have been carried out on different pulse tube configurations by varying the dimensions of the pulse tubes and regenerators to arrive at the best one, which leads to the lowest no-load second stage cold head temperature. Using this configuration, detailed experimental studies have been conducted by varying the volume percentage ratios of the second stage regenerator materials such as HoCu2, Er3Ni, and Pb (with an average grain size of similar to 250 mu m). This article presents the results of our experimental studies on cryocoolers with the regenerator material arranged in layered structures. Comparative studies have also been presented for specific cases where the regenerator materials are arranged as a homogeneous mixture in the second stage. The experimental results clearly indicate that the design of PTCs should use only layered structures of regenerator materials and not homogenous mixtures.

Contribution of different physical forces to the disjoining pressure of a thin water film being pressed by an oil droplet

With an objective to understand the nature of forces which contribute to the disjoining pressure of a thin water film on a steel substrate being pressed by an oil droplet, two independent sets of experiments were done. (i) A spherical silica probe approaches the three substrates; mica, PTFE and steel, in a 10 mM electrolyte solution at two different pHs (3 and 10). (ii) The silica probe with and without a smeared oil film approaches the same three substrates in water (pH = 6). The surface potential of the oil film/water was measured using a dynamic light scattering experiment. Assuming the capacity of a substrate for ion exchange the total interaction force for each experiment was estimated to include the Derjaguin-Landau-Verwey-Overbeek (DLVO) force, hydration repulsion, hydrophobic attraction and oil-capillary attraction. The best fit of these estimates to the force-displacement characteristics obtained from the two sets of experiment gives the appropriate surface potentials of the substrates. The procedure allows an assessment of the relevance of a specific physical interaction to an experimental configuration. Two of the principal observations of this work are: (i) The presence of a surface at constant charge, as in the presence of an oil film on the probe, significantly enhances the counterion density over what is achieved when both the surfaces allow ion exchange. This raises the corresponding repulsion barrier greatly. (ii) When the substrate surface is wettable by oil, oil-capillary attraction contributes substantially to the total interaction. If it is not wettable the oil film is deformed and squeezed out. (C) 2010 Elsevier Inc. All rights reserved.

Relation between part position and kinematic freedom: An expository survey

This paper makes explicit the relation between relative part position and kinematic freedom of the parts which is implicitly available in the literature. An extensive set of representative papers in the areas of assembly and kinematic modelling is reviewed to specifically identify how the ideas in the two areas are related and influencing the development of each other. The papers are categorised by the approaches followed in the specification, representation, and solution of the part relations. It is observed that the extent of the part geometry is not respected in modelling schemes and as a result, the causal flow of events (proximity–contact–mobility) during the assembling process is not realised in the existing modelling paradigms, which are focusing on either the relative positioning problem or the relative motion problem. Though an assembly is a static description of part configuration, achievement of this configuration requires availability of relative motion for bringing parts together during the assembly process. On the other hand, the kinematic freedom of a part depends on the nature of contacting regions with other parts in its static configuration. These two problems are thus related through the contact geometry. The chronology of the approaches that significantly contributed to the development of the subject is also included in the paper.