Approximate similarity search for online multimedia services on distributed CPU–GPU platforms

Similarity search in high-dimensional spaces is a pivotal operation for several database applications, including online content-based multimedia services. With the increasing popularity of multimedia applications, these services are facing new challenges regarding (1) the very large and growing volumes of data to be indexed/searched and (2) the necessity of reducing the response times as observed by end-users. In addition, the nature of the interactions between users and online services creates fluctuating query request rates throughout execution, which requires a similarity search engine to adapt to better use the computation platform and minimize response times. In this work, we address these challenges with Hypercurves, a flexible framework for answering approximate k-nearest neighbor (kNN) queries for very large multimedia databases. Hypercurves executes in hybrid CPU–GPU environments and is able to attain massive query-processing rates through the cooperative use of these devices. Hypercurves also changes its CPU–GPU task partitioning dynamically according to the observed load, aiming for optimal response times. In our empirical evaluation, dynamic task partitioning reduced query response times by approximately 50 % compared to the best static task partition. Due to a probabilistic proof of equivalence to the sequential kNN algorithm, the CPU–GPU execution of Hypercurves in distributed (multi-node) environments can be aggressively optimized, attaining superlinear scalability while still guaranteeing, with high probability, results at least as good as those from the sequential algorithm.     

Soap-based detergent formulations: X. Nature of detergent deposits

The cumulative deposition of detergent residue on unsoiled cotton and polyester-cotton permanent press finish cloth was determined for a variety of detergent formulations after washing 25 consecutive times in 300 ppm hard water in a laboratory Tergotometer. Included in this study were: a phosphate-built laundry detergent, two carbonate-built detergents, tallow soap and various tallow soap formulations with anionic and amphoteric lime soap dispersing agents, and a glassy sodium silicate. Sample swatches washed with each formulation were analyzed for calcium, magnesium, and organic acid content. Fabric washed with the carbonate detergents showed the highest calcium and magnesium content, while those washed with the phosphate detergent and the soap-lime soap dispersant-builder formulations had the lowest. Fabric washed with soap alone had a much higher fatty acid residue than those washed with the other detergent formulations. However, the amount of organic acids left on the fabric after washing with a soap-lime soap dispersing agent formulation was no greater than that produced by phosphate- and carbonate-built detergents. The presence or absence of deposits also was verified visually with a scanning electron microscope. Each formulation also was tested for detergency by measuring the soil buildup in a multiwash procedure. Generally, the buildup of soil paralleled the deposit of detergent residue on the unsoiled cloths.     

Effects of apitong (Dipterocarpus spp.) extractives on bond strength development and curing rate of adhesives

Emulsion polymer isocyanate ( ), polyvinyl acetate ( ) and resorcinol-formaldehyde ( ) adhesives were used to produce single lap shear specimens using resinous and non-resinous apitong (Dipterocarpus spp.) timbers. Tests showed that joints made with highly resinous apitong were about 40% weaker than similar joints made with non-resinous apitong. The resinous apitong was treated with different solvents to yield five different extractives which were characterized by infra-red analysis. Apitong extractives were then added to and adhesives and joints made with buna (Fagus crenata), a timber known to be low in extractives and easy to bond. Joints prepared using extractive-containing adhesives were generally weaker than those made with the unmodified adhesives. RF adhesives containing extractives cured more slowly than unmodified . It is thought that the acidic nature of the extractives changes the pH of the system sufficiently to affect the curing mechanism     

Behavior of the western pine beetle during host colonization

After living ponderosa pines were baited with either female-infested bolts or synthetic pheromones,Dendroctonus brevicomis were caught on sticky screens throughout trapping periods of 15–46 days; however, large numbers of beetles were trapped during only a small portion (5–10 days) of these trapping periods. The most attractive portions of trees attacked contained 3–6 beetles dm², in galleries ca. 2 cm long. Catch increased following addition of males to female-infested bolts, supporting the hypothesis that male-produced frontalin is an attractive pheromone of the western pine beetle. Catch at bolts removed from trees under attack was strongly dependent upon levels of boring activity. We found no evidence of interruption of the response to attractants during host colonization.Coleoptera: Scolytidae. Trade names and commercial enterprises or products are mentioned solely for information and do not constitute endorsement by the U.S. Department of Agriculture or University of California.     

Separation and determination of mono-, Di-, and triglycerides in monoglyceride concentrates

Summary Monoglyceride concentrates are quantitatively separated into mono-, di-, and triglyceride components on silica gel columns by an adsorption chromatographic technique. The separated glycerides are determined gravimetrically. The adsorption on silica gel is dependent on the number of hydroxyl groups in the molecule, and the influence of unsaturation and chain length is minimized. Combinations of benzene and ethyl ether are used for elution, thus preventing isomerization, which frequently results when polar solvents such as alcohols are used. The procedure for the chromatographic separation is simple and straightforward. The equipment used is easily obtainable. The silica gel adsorbent requires only adjustment of its water content before use. Mixtures of both saturated and unsaturated glycerides and those having different fatty acid chain length have been separated. Known mixtures have also been analyzed. For a known mixture containing 15% tripalmitin, 45% distearin, and 38% monopalmitin the average deviations from the calculated values were 0.4%, 0.2%, and 0.6%, respectively     

Soap-based detergent formulations: VI. Alkylaryl sulfonamide derivatives as lime soap dispersing agents

Alkylbenzenes, such as industrial detergent alkylates, as well as pure 1-phenylalkanes whose side chain lengths varied C₈−C₁₂, were converted into the corresponding alkylbenzenensulfonyl chlorides with chlorosulfonic acid. Surface active sulfonamides were obtained from the reaction of the sulfonyl chlorides with various low mol wt aminosulfonic acids, such as N-methyltaurine, or with aminoalkyl esters of sulfuric acid, such as 2-aminoethyl hydrogen sulfate. The hydrolytic stability of the resulting surface active sulfonamide derivatives was investigated. As expected, the sulfonates were quite resistant to acid or alkaline hydrolysis, while the sulfates were more susceptible to hydrolysis. Hydrolytic stability of the sulfonamides was compared with that of the analogous fatty acid amide sulfactants. All of the compounds were excellent lime soap dispersing agents giving Borhetty-Bergman values of 4–10. The compounds were evaluated for detergency either alone or formulated either with tallow soap or with tallow soap and sodium silicate (Na₂O/SiO₂ ratio of 1∶1.6) The derivatives of the pure hydrocarbons which gave the best overall detergency were those of 1-phenyldecane and 1-phenyldodecane, whereas those of 1-phenyloctane exhibited poor detergency. This ranking was observed when the compounds were tested alone as well as when formulated. The sulfonamide derivatives of the detergent alkylate type of alkylbenzenes exhibited excellent detergency characteristics and showed substantial potentiation of detergency when mixed with soap or with a soap-sodium silicate blend. The detergency performance of some of these formulated detergents was equal to that of a commercial household detergent used as a control.     

Introductory remarks

Surface active agents are used in the textile industry during the processing of fibers into yarns and fabrics, the finishing of leather products and during the removal of soil from textile products by laundering or dry cleaning. In these introductory remarks to the symposium, attention is directed to the phenomena occurring at solid-liquid interfaces during these processes. The properties of the surface active agents and the solvents used, the surface free energy and structure of the polymeric textile products and other, often poorly defined, surface chemical and physical characteristics of the products influence the interactions of these agents. The interactions of surface active agents, often derived from fats and oils, dissolved or dispersed in liquids, with textile materials, including natural products such as cotton, wool and leather, as well as man-made fibers, are considered in this symposium. One of eight papers being published from the Symposium “Surface Active Agents in the Textile Industry,” presented at the AOCS meeting in New Orleans, April 1970. So. Utiliz. Res. Dev. Div., ARS, USDA.     

Effects of crack depth on elastic-plastic fracture toughness

Short crack test specimens (a/W 0.50) are frequently employed when conventional deep crack specimens are either inappropriate or impossible to obtain, for example, in testing of particular microstructures in weldments and in-service structures containing shallow surface flaws. Values of elastic-plastic fracture toughness, here characterized by the crack tip opening displacement (CTOD), are presented for square (cross-section) three-point bend specimens with a/W ratios of 0.15 and 0.50 throughout the lower-shelf and lower-transition regions. Three dimensional, finite-element analyses are employed to correlate the measured load and crack mouth opening displacement (CMOD) values to the corresponding CTOD values, thus eliminating a major source of experimental difficulty in previous studies of shallow crack specimens. In the lower-transition region, where extensive plasticity (but no ductile crack growth) precedes brittle fracture, critical CTOD values for short crack specimens are significantly larger (factor of 2–3) than the CTOD values for deep crack specimens at identical temperatures. Short crack specimens are shown to exhibit increased toughness at the initiation of ductile tearing and decreased brittle-to-ductile transition temperatures. Numerical analyses for the two a/W ratios reveal large differences in stress fields ahead of the crack tip at identical CTOD levels which verify the experimentally observed differences in critical CTOD values. Correlations of the predicted stresses with measured critical CTOD values demonstrate the limitations of single-parameter fracture mechanics (as currently developed) to characterize the response.     

Characterizing the deformation of cold-rolled copper sheet

An advanced x-ray diffraction technique allows the development of texture components to be monitored and provides supplementary information about the cold work and residual stresses in cold-rolled copper sheet. The new technique confirms the results of an ultrasonic test, except for the ultrasonic test’s sensitivity to grain size. The new x-ray diffraction technique could be adapted to provide in-process, nondestructive evaluation of thin copper sheet, even for material moving at high speeds.