Computers & chemistry最新文献

A differential fraction-based kinetic model (DFKM) was proposed to simulate the reaction process in view of the extremely large number of compounds in the petroleum fraction. The results of the simulation experiment on the hydrodesulfurization (HDS) process demonstrated that without considering the exact compounds, DFKM can depict the reaction processes not only of the whole petroleum fraction but of any narrow cut we are interested in. Apparently, this kind of ability, which can be considered as the ‘zoom in/out’ function, is very useful for the further study on the processing techniques of the petroleum fraction. In the meantime, different methods for obtaining the functions needed in DFKM were investigated and satisfactory results were presented.

A new approach for estimating the chemical rank of the three-way array called the principal norm vector orthogonal projection method has been proposed. The method is based on the fact that the chemical rank of the three-way data array is equal to one of the column space of the unfolded matrix along the spectral or chromatographic mode. A vector with maximum Frobenius norm is selected among all the column vectors of the unfolded matrix as the principal norm vector (PNV). A transformation is conducted for the column vectors with an orthogonal projection matrix formulated by PNV. The mathematical rank of the column space of the residual matrix thus obtained should decrease by one. Such orthogonal projection is carried out repeatedly till the contribution of chemical species to the signal data is all deleted. At this time the decrease of the mathematical rank would equal that of the chemical rank, and the remaining residual subspace would entirely be due to the noise contribution. The chemical rank can be estimated easily by using an F-test. The method has been used successfully to the simulated HPLC-DAD type three-way data array and two real excitation–emission fluorescence data sets of amino acid mixtures and dye mixtures. The simulation with added relatively high level noise shows that the method is robust in resisting the heteroscedastic noise. The proposed algorithrn is simple and easy to program with quite light computational burden.

The application of fourth-order finite difference discretisations of the second derivative of concentration with respect to distance from the electrode, in electrochemical digital simulations, is examined further. In the bulk of the diffusion space, a central 5-point scheme is used, and 6-point asymmetric schemes are used at the edges. In this paper, four Runge–Kutta schemes have been used for the time integration. The observed efficiencies, for the Cottrell experiment and chronopotentiometry, are satisfactory, going beyond those for the 3-point scheme. However, it is third-order Runge–Kutta, rather than the fourth-order scheme, which is the most efficient, the two resulting in practically the same errors. This is probably due to the computational procedure where a constant ratio of δt/h² was used.

The Quantitative Structure–Property Relationship (QSPR) method is used to develop the correlation between structures of a great number of substituted benzenes and their critical pressure. Molecular descriptors calculated from structure alone were used to represent molecular structures. A subset of the calculated descriptors selected using forward stepwise regression was used in the QSPR model development. Multiple Linear Regression and Radial Basis Function Neural Networks are utilized to construct the linear and non-linear prediction model, respectively. To obtain good prediction ability, both topological structure and training parameters of radial basis function neural networks are optimized. The prediction result agrees well with the experimental value of these properties.

A P-stable method of algebraic order eight for the approximate numerical integration of the Schrödinger equation is developed in this paper. Since the method is P-stable (i.e. its interval of periodicity is equal to (0, ∞)), large step sizes for the numerical integration can be used. Based on this new method and on a sixth algebraic order P-stable method developed by Simos (Phys. Scripta 55 (1997) 644–650), a new variable step method is obtained. Numerical results presented for the phase-shift problem of the radial Schrödinger equation and for the coupled differential equations arising from the Schrödinger equation show the efficiency of the developed method.

The molecular geometries and internal rotational barriers of the nitro group of 3-nitrotoluene (3-NT), 4-nitrotoluene (4-NT), 3-nitrophenol (3-NP), 4-nitrophenol (4-NP), 3-nitroaniline (3-NA), and 4-nitroaniline (4-NA) were calculated by five different types of density functional theory (DFT) methods with three different levels of basis sets. Analyses of the torsional angles of the nitro, methyl, amino, and hydroxyl groups indicate that 3-NP, and 4-NP are planar molecules, but 3-NT, 4-NT, 3-NA, and 4-NA are not planar molecules. Internal rotational barriers of the nitro group were calculated as the V₂ barrier, and the NO₂ torsional potentials for each molecule were given. The heights of the V₂ barrier vary with the DFT methods, the basis sets, and the kinds and positions of substituents. The average values of the V₂ barriers for 3-NT, 4-NT, 3-NP, 4-NP, 3-NA, and 4-NA are 6.44, 6.92, 6.64, 7.93, 6.38, and 9.13 kcal/mol, respectively. Torsional potentials of the OH and NH₂ groups of nitrophenol and nitroaniline derivatives were also studied by a B3LYP/6-31G* approach. Except for the OH group in 2-NP, these derivatives have the V₂ barrier.