Improved customer choice predictions using ensemble methods

In this paper various ensemble learning methods from machine learning and statistics are considered and applied to the customer choice modeling problem. The application of ensemble learning usually improves the prediction quality of flexible models like decision trees and thus leads to improved predictions. We give experimental results for two real-life marketing datasets using decision trees, ensemble versions of decision trees and the logistic regression model, which is a standard approach for this problem. The ensemble models are found to improve upon individual decision trees and outperform logistic regression.     

High finesse opto-mechanical cavity with a movable thirty-micron-size mirror

We report on the demonstration of a high finesse micro-optomechanical system and identify potential applications ranging from optical cooling to weak force detection to massive quantum superpositions. The system consists of a high quality diameter flat dielectric mirror cut from a larger substrate with a focused ion beam and attached to an atomic force microscope cantilever. Cavity ring-down measurements performed on a 25 mm long Fabry-Pérot cavity with the 30 microm mirror at one end show an optical finesse of 2100. Numerical calculations show that the finesse is not diffraction limited and that orders of magnitude higher finesse should be possible. A mechanical quality factor of more than 10(5) at pressures below 10(-3) mbar is demonstrated for the cantilever with a mirror attached.     

Combining approximate solutions for linear discrete ill-posed problems

Linear discrete ill-posed problems of small to medium size are commonly solved by first computing the singular value decomposition of the matrix and then determining an approximate solution by one of several available numerical methods, such as the truncated singular value decomposition or Tikhonov regularization. The determination of an approximate solution is relatively inexpensive once the singular value decomposition is available. This paper proposes to compute several approximate solutions by standard methods and then extract a new candidate solution from the linear subspace spanned by the available approximate solutions. We also describe how the method may be used for large-scale problems.     

Subspace methods for three‐parameter eigenvalue problems

We propose subspace methods for three‐parameter eigenvalue problems. Such problems arise when separation of variables is applied to separable boundary value problems; a particular example is the Helmholtz equation in ellipsoidal and paraboloidal coordinates. While several subspace methods for two‐parameter eigenvalue problems exist, their extensions to a three‐parameter setting seem challenging. An inherent difficulty is that, while for two‐parameter eigenvalue problems, we can exploit a relation to Sylvester equations to obtain a fast Arnoldi‐type method, such a relation does not seem to exist when there are three or more parameters. Instead, we introduce a subspace iteration method with projections onto generalized Krylov subspaces that are constructed from scratch at every iteration using certain Ritz vectors as the initial vectors. Another possibility is a Jacobi–Davidson‐type method for three or more parameters, which we generalize from its two‐parameter counterpart. For both approaches, we introduce a selection criterion for deflation that is based on the angles between left and right eigenvectors. The Jacobi–Davidson approach is devised to locate eigenvalues close to a prescribed target; yet, it often also performs well when eigenvalues are sought based on the proximity of one of the components to a prescribed target. The subspace iteration method is devised specifically for the latter task. The proposed approaches are suitable especially for problems where the computation of several eigenvalues is required with high accuracy. MATLAB implementations of both methods have been made available in the package MultiParEig (see http://www.mathworks.com/matlabcentral/fileexchange/47844-multipareig ).     

Catalytic Gas‐Phase Production of Lactide from Renewable Alkyl Lactates

A new route to lactide, which is a key building block of the bioplastic polylactic acid, is proposed involving a continuous catalytic gas‐phase transesterification of renewable alkyl lactates in a scalable fixed‐bed setup. Supported TiO₂/SiO₂ catalysts are highly selective to lactide, with only minimal lactide racemization. The solvent‐free process allows for easy product separation and recycling of unconverted alkyl lactates and recyclable lactyl intermediates. The catalytic activity of TiO₂/SiO₂ catalysts was strongly correlated to their optical properties by DR UV/Vis spectroscopy. Catalysts with high band‐gap energy of the supported TiO₂ phase, indicative of a high surface spreading of isolated Ti centers, show the highest turnover frequency per Ti site.     

Methane-to-methanol oxidation by the hydrated iron(IV) oxo species in aqueous solution: a combined DFT and car-parrinello molecular dynamics study

Previously, we have shown that the ferryl ion ([FeIVO]2+) is easily produced from Fenton's reagent (i.e., a mixture of Fe2+ ions and H2O2 in aqueous solution), using DFT and Car-Parrinello MD calculations. To verify that the ferryl ion can indeed act as the active species in oxidation reactions with Fenton's reagent, we study in the present paper the reactivity of the ferryl ion toward an organic substrate, in particular the oxidation of methane to methanol. In the first part of this paper, we perform static DFT calculations on the reaction of CH4 with the [(H2O)5FeIVO]2+ complex in vacuo that show a strong prevalence of the oxygen-rebound mechanism over the methane coordination mechanism. This is in agreement with the static DFT results for methane oxidation by biocatalysts MMO and P450, but not with those for methane oxidation by bare metal-oxo ions, where the methane coordination mechanism prevails. The highest energy barrier in the oxygen-rebound mechanism is only 3 kcal/mol in vacuo, whereas in the methane coordination mechanism the highest barrier is 23 kcal/mol. Overall the oxidation reaction energy is downhill by 47 kcal/mol. We conclude that the ferryl ion can indeed act as the oxidative intermediate in the Fenton oxidation of organic species. In the second part of this paper, we perform a preliminary assessment of solvent effects on the oxidation by the ferryl ion in aqueous solution using the method of constrained (first principles) molecular dynamics. The free energy barrier of the H-abstraction reaction from methane by the ferryl ion (i.e., the first step in the rebound mechanism) in aqueous solution is, with 22 kcal/mol in solution, significantly higher than in vacuo. Given the fact that methane has a relatively strong C-H bond (ca. 10 kcal/mol stronger than the C-H bonds in the more typical Fenton's reagent substrates), we infer that for many organic substrates oxidation with the ferryl ion as an active intermediate may be a perfectly viable route.     

Electronic structure and solvation of copper and silver ions: a theoretical picture of a model aqueous redox reaction

Electronic states and solvation of Cu and Ag aqua ions are investigated by comparing the Cu(2+) + e(-)--> Cu(+) and Ag(2+) + e(-)--> Ag(+) redox reactions using density functional-based computational methods. The coordination number of aqueous Cu(2+) is found to fluctuate between 5 and 6 and reduces to 2 for Cu(+), which forms a tightly bound linear dihydrate. Reduction of Ag(2+) changes the coordination number from 5 to 4. The energetics of the oxidation reactions is analyzed by comparing vertical ionization potentials, relaxation energies, and vertical electron affinities. The model is validated by a computation of the free energy of the full redox reaction Ag(2+) + Cu(+) --> Ag(+) + Cu(2+). Investigation of the one-electron states shows that the redox active frontier orbitals are confined to the energy gap between occupied and empty states of the pure solvent and localized on the metal ion hydration complex. The effect of solvent fluctuations on the electronic states is highlighted in a computation of the UV absorption spectrum of Cu(+) and Ag(+).     

Evaluation of a purge-and-trap injection system for capillary gas chromatography-microwave induced plasma-atomic emission spectrometry for the determination of volatile selenium compounds in water

A method is presented for the selective determination of the volatile selenium species dimethylselenide and dimethyldiselenide, using a commercially available purge-and-trap injection system coupled to capillary gas chromatography-microwave induced plasma-atomic emission spectrometry. The efficiency of the purging step was evaluated and the parameters affecting the purge and trap processes were optimized. The method was applied to the determination of volatile selenium compounds in lake water. Relative detection limits of 2ng/l for dimethylselenide and dimethyldiselenide, corresponding to an absolute detection limit of 10 pg, were achieved.     

Synthesis and elaboration of functionalised carbohydrate-derived spiroketals

The scope of a stereoselective three-step approach for the synthesis of sugar derived spiroketals is presented. The methodology consists of Grignard addition of vinyl- or allylmagnesium bromide to a carbohydrate lactone, followed by K-10 clay mediated glycosidation with a terminal alkenol and subsequent ring-closing metathesis of the resulting diene. The generality of this procedure is demonstrated by the synthesis of various pyranose- and furanose-derived spiroketals, as well as more advanced tricyclic spiroketal derivatives. It is shown that functionalisation of the double bond in the resulting spiroketals leads to fused polycyclic ethers.     

Nanoparticulate Pd supported catalysts: size-dependent formation of Pd(I)/Pd(0) and their role in CO elimination

A combination of time-resolved X-ray absorption spectroscopy (XAS), hard X-ray diffraction (HXRD), diffuse reflectance infrared spectroscopy (DRIFTS), and mass spectrometry (MS) reveals a series of size-dependent phenomena at Pd nanoparticles upon CO/(NO+O(2)) cycling conditions. The multitechnique approach and analysis show that such size-dependent phenomena are critical for understanding Pd CO elimination behavior and, particularly, that different Pd(I) and Pd(0) centers act as active species for a size estimated by XAS to be, respectively, below and above ca. 3 nm. The relative catalytic performance of these two noble metal species indicates the intrinsic higher activity of the Pd(I) species.