New support vector algorithms

We propose a new class of support vector algorithms for regression and classification. In these algorithms, a parameter nu lets one effectively control the number of support vectors. While this can be useful in its own right, the parameterization has the additional benefit of enabling us to eliminate one of the other free parameters of the algorithm: the accuracy parameter epsilon in the regression case, and the regularization constant C in the classification case. We describe the algorithms, give some theoretical results concerning the meaning and the choice of nu, and report experimental results.     

Metal extraction from ores by heap leaching

Heap leaching low-grade ores has become a major contributor to the extraction of economically important metals, notably copper and gold. The state-of-the-art in heap leaching is reviewed with emphasis on process engineering. Rock leaching, including rock pore diffusion and mineral kinetics, solution flow, and retention in ore heaps during percolation leaching, and bio-oxidation of sulfidic ores are covered. Oxygen transport into heaps by gaseous diffusion, natural convection, and forced air ventilation is discussed. Strategies for optimizing heap leaching include ore crushing, ore agglomeration, low-cost air ventilation of sulfide ore heaps undergoing bio-oxidation using fans, and the use of aggregate metal extraction rate constants in making metallurgical business decisions about heap leaching. The Extraction and Processing Lecturer Award honors an outstanding scientific leader in the field of nonferrous extractive metallurgy with an invitation to present a comprehensive lecture at the TMS Annual Meeting. Robert W. Bartlett, presently dean of the College of Mines and Earth Resources and professor of metallurgical engineering at the University of Idaho, has spent half of his career in industry, Anaconda, ARCO, and Kennecott, and half in academia, Stanford and Idaho. He earned his degrees at the University of Utah. His lecture is based on research in industry and at the University of Idaho. Dr. Bartlett has been a member of TMS for over 40 years and was president in 1989. Earlier, he was a member of the TMS Board of Directors representing, for 3 years, what is now the Extraction and Processing Division. Dr. Bartlett consults primarily on processing in the mining and metals industries. He is a recipient of the TMS Extraction and Processing Technology Award, the Turner Award of the Electrochemical Society, the Wadsworth Award of the Society of Mining, Metallurgy and Exploration, and the McConnell Award of AIME. He is a member of the National Academy of Engineering.     

Extracorporeal membrane oxygenator support for cardiopulmonary failure. Experience in 28 cases

We have used extracorporeal membrane oxygenation (ECMO) for 28 patients (14 children and 14 adults) over a 5 year period. Nine patients improved on ECMO and 5 were long-term survivors. ECMO was used for pulmonary insufficiency in 24 patients. Initially, only moribund patients were treated, but recently the combination of open lung biopsy and pulmonary insufficiency index (PII) has been used to select patients. The best results have been obtained in newborn cases and the adult capillary leak syndromes; the major problem has been progression to fibrosis despite ECMO support. ECMO was used for cardiac failure in 4 patients. Children with postoperative cardiac failure did the best; profound shock was not reversed with venoarterial bypass. ECMO support is lifesaving in selected cases of pulmonary insufficiency. Initial trials in cardiac failure and the infant age group in this series suggest that ECMO will have an even greater role in those applications.     

Diffusivity and solubility of oxygen in platinum and Pt-Ni alloys

The diffusivity and solubility of oxygen in metal specimens were determined from measurements of permeation through thin wall tubes containing oxygen and heated by electrical resistance. The permeating oxygen desorbed in vacuum as monoatomic oxygen and the flux was monitored mass-spectrometrically. A known helium leak rate and experimentally-determined sensitivities of the two gases were used for flow calibration. The diffusion coefficient of oxygen in pure platinum, calculated from the time lag to reach a steady state flux, is: {fx065-01} in the temperature range of 1435∮ to 1504°. The solubility of oxygen in pure platinum was obtained from the steady-state flux using the previously determined diffusivity. The solubility is proportional to p^1/2O₂ and at 1 atm of oxygen the solubility is: C^s _O = (0.2 ± 0.1) × 10¹² exp {fx065-02}, wt pct. Small amounts of nickel, less than required for internal oxidation, had a negligible effect on the oxygen solubility and diffusivity in platinum alloys. Formerly Graduate Student, Department of Mineral Engineering, Stanford University, Stanford, Calif. This paper is based upon a thesis submitted by L. R. VELHO in partial fulfillment of the requirements of the degree of Doctor of Philosophy at Stanford University.     

Observation of an undecagold cluster compound in the scanning transmission electron microscope

A water-soluble undecagold cluster compound has been shown to be visible in moderate dose images (10⁴ electrons/nm²) obtained with the scanning transmission electron microscope (STEM). The properties of this compound render it potentially useful as a marker for specific sites within biological molecules.     

Transport and cathode deposition of reducible impurities during electrorefining with nonpolarizable electrodes

A general equation for cathode impurity concentration is developed for steady-state electrorefining with an impurity that is more readily reduced than the primary metal. The results depend on the equilibrium cathode/electrolyte distribution coefficient and the transport properties of the electrolyte. Applicability is restricted to situations where electrode kinetics are not rate limiting, such as most fused salt systems. Experimental results for copper impurity in a Sn/fused SnCl₂/Sn system are compared with calculated cathode impurities as a function of cell voltage.     

Two-dimensional "Circle criterion" on the parameter plane

A graphical procedure using parameter plane theory is given for studying the stability of systems consisting of a linear time-invariant subsystem with two feedback time-varying gains.     

Growth and properties of CdxHg1−xTe crystals

It has been found that there are two main factors complicating the preparation of Cd_xHg_1–xTe. The marked difference between the liquidus and solidus curves of the CdTe-HgTe pseudo-binary system gives the expected problems of segregation of CdTe with respect to HgTe during growth but there are also problems due to the segregation of any excess Te in the melt. A 2% excess of Te can give rise to pronounced constitutional supercooling effects. To avoid this, careful control of melt stoichiometry is required. This is made difficult by the high vapour pressure of mercury over the melt, the value of which is not known with great accuracy.The conditions of melt stoichiometry required for crystal growth do not necessarily give material of the required type and resistivity and this must be adjusted after growth by annealing at a controlled mercury pressure at a fixed temperature.     

A field comparison of four fungal aerosol sampling instruments: inter-sampler calibrations and caveats

Four bioaerosol samplers (Reuter Centrifugal, Andersen N6 Single Stage, Surface Air System Super 90, and Air-o-Cell) were used to take c. 300 side-by-side measurements at 75 public building sites. Regression models were developed to examine the relationships between each method pair. The models demonstrate that measurements from these instruments are not directly comparable, requiring inter-instrument calibration. Sampling location (indoor vs. outdoor) was a confounder in all the pairwise comparisons between samplers. In addition, the slopes of the relationships between all method pairs except one differed in indoor vs. outdoor locations. These results emphasize that direct comparisons between methods should not be undergone without prior calibration. Where measurement circumstances are similar to those of this study, the regression models might serve as a basis to convert measurements made with one instrument to those made with another. However, the robustness and generalizability of the models in different measurement settings needs to be assessed. PRACTICAL IMPLICATIONS: Many different bioaerosol sampling devices are in common use for indoor air quality studies. If data from research studies are to be compared, an approximation of the relationships between the equipment would be useful. A comparison of three culturable sampling devices (Andersen N6, SAS 90, RCS) and one particulate sampling device (Air-o-Cell) collecting simultaneous samples under field conditions showed high linear correlations between methods. However, while direct comparisons between sampling data were not possible, the regression models reported here explained 60-85% of the variance in fungal concentrations, and underscored the importance of the effect of environment on measurement.     

Practical implications of neutron survey instrument performance

Improvements have been made to the Monte Carlo modelling used to calculate the response of the neutron survey instruments most commonly used in the UK, for neutron energies up to 20 MeV. The improved modelling of the devices includes the electronics and battery pack, allowing better calculations of both the energy and angle dependence of response. These data are used to calculate the response of the instruments in rotationally and fully isotropic, as well as unidirectional fields. Experimental measurements with radionuclide sources and monoenergetic neutron fields have been, and continue to be made, to test the calculated response characteristics. The enhancements to the calculations have involved simulation of the sensitivity of the response to variations in instrument manufacture, and will include the influence of the user and floor during measurements. The practical implications of the energy and angle dependence of response, variations in manufacture, and the influence of the user are assessed by folding the response characteristics with workplace energy and direction distributions.