Nihat Yildiz

Determination of structural, spectrometric and nonlinear optical features of 2-(4-hydroxyphenylazo)benzoic acid by experimental techniques and quantum chemical calculations

The optimized geometrical structure, vibrational and electronic transitions, chemical shifts and nonlinear optical properties of 2-(4-hydroxyphenylazo)benzoic acid (HABA) compound were presented in this study. The ground state geometrical structure and vibrational wavenumbers were carried out by using density functional (DFT/B3LYP) method with 6-311++G(d,p) basis set. The vibrational spectra of title compound were recorded in solid state with FT-IR and FT-Raman spectrometry in the range of 4000-400 cm(-1) and 4000-10 cm(-1); respectively. The fundamental assignments were done on the basis of the recorded spectra and total energy distribution (TED) of the vibrational modes. The (1)H and (13)C NMR spectra were recorded in deuterated DMSO solution, and gauge-invariant atomic orbitals (GIAOs) method was used to predict the isotropic chemical shifts. The UV-Vis absorption spectra of the compound were observed in the range of 200-800 nm in ethanol, methanol and water solvents. To investigate the nonlinear optical properties, the polarizability, anisotropy of polarizability and molecular first hyperpolarizability were computed. A detailed description of spectroscopic behaviors of compound was given based on the comparison of experimental measurements and theoretical computations.

Correlation structure of training data and the fitting ability of back propagation networks: Some experimental results

White [6–8] has theoretically shown that learning procedures used in network training are inherently statistical in nature. This paper takes a small but pioneering experimental step towards learning about this statistical behaviour by showing that the results obtained are completely in line with Whites theory. We show that, given two random vectorsX (input) andY (target), which follow a two-dimensional standard normal distribution, and fixed network complexity, the networks fitting ability definitely improves with increasing correlation coefficient rXY (0≤rXY≤1) betweenX andY. We also provide numerical examples which support that both increasing the network complexity and training for much longer do improve the networks performance. However, as we clearly demonstrate, these improvements are far from dramatic, except in the case rXY=+ 1. This is mainly due to the existence of a theoretical lower bound to the inherent conditional variance, as we both analytically and numerically show. Finally, the fitting ability of the network for a test set is illustrated with an example.

CONSISTENT EMPIRICAL PHYSICAL FORMULAS FOR POTENTIAL ENERGY CURVES OF 38−66 Ti ISOTOPES BY USING NEURAL NETWORKS

Nuclear shape transition has been actively studied in the past decade. In particular, the understanding of this phenomenon from a microscopic point of view is of great importance. Because of this reason, many works have been employed to investigate shape phase transition in nuclei within the relativistic and nonrelativistic mean field models by examining potential energy curves (PECs). In this paper, by using layered feed-forward neural networks (LFNNs), we have constructed consistent empirical physical formulas (EPFs) for the PECs of 38–66Ti calculated by the Hartree-Fock-Bogoliubov (HFB) method with SLy4 Skyrme forces. It has been seen that the PECs obtained by neural network method are compatible with those of HFB calculations.

Neural network consistent empirical physical formula construction for density functional theory based nonlinear vibrational absorbance and intensity of 6-choloronicotinic acid molecule.

Being directly related to the electric charge distributions in a molecule, the vibrational spectra intensities are both experimentally and theoretically important physical quantities. However, these intensities are inherently highly nonlinear and of complex pattern. Therefore, in particular for unknown detailed spatial molecular structures, it is difficult to make ab initio intensity calculations to compare with new experimental data. In this respect, we very recently initiated entirely novel layered feedforward neural network (LFNN) approach to construct empirical physical formulas (EPFs) for density functional theory (DFT) vibrational spectra of some molecules. In this paper, as a new and far improved contribution to our novel molecular vibrational spectra LFNN-EPF approach, we constructed LFFN-EPFs for absorbances and intensities of 6-choloronicotinic acid (6-CNA) molecule. The 6-CNA data, borrowed from our previous study, was entirely different and much larger than the vibrational intensity data of our formerly used LFNN-EPF molecules. In line with our another previous work which theoretically proved the LFNN relevance to EPFs, although the 6-CNA DFT absorbance and intensity were inherently highly nonlinear and sharply fluctuating in character, still the optimally constructed train set LFFN-EPFs very successfully fitted the absorbances and intensities. Moreover, test set (i.e. yet-to-be measured experimental data) LFNN-EPFs consistently and successfully predicted the absorbance and intensity data. This simply means that the physical law embedded in the 6-CNA vibrational data was successfully extracted by the LFNN-EPFs. In conclusion, these vibrational LFNN-EPFs are of explicit form. Therefore, by various suitable operations of mathematical analysis, they can be used to estimate the electronic charge distributions of the unknown molecule of the significant complexity. Additionally, these estimations can be combined with those of theoretical DFT atomic polar tensor calculations to contribute to the identification of the molecule.