For molecular design of freon alternatives, we attempted to extract bond parameter characterizing the infrared absorption intensity in the 1500-500 cm^-1 region from 44 kinds of fluorine-containing molecules: freon alternatives in gas phase by the analysis of sensitivity and differential coefficients of input parameter for the three layers perceptron type neural network. The analysis of differential coefficients of input parameter for the three layers perceptron type neural network was developed by Aoyama and Ichakawa[3, 4] and was newly equipped into a neural network simulator Neco [2, 5, 8 -13].
The importance of 8 bond types: C-C, C=C, C-O, C=O, C-H, C-F, C-Cl, O-H were examined using a well educated neural network, where the error of leave-one out test is less than 0.007. The error is acceptable because the value corresponds almost 5% error of intensity and 10% error is usually included in the observed values.
The numbers of C=O and O-H bonds increase the intensity whereas C-O and C-F have less effect. In Figure 2, the sensitivity analysis suggested that the number of C-C bonds is to be unimportant for the intensity. However, the result of the differential coefficients analysis suggested the importance of the number of C-C bonds as shown in Figure 3.
The results of the sensitivity analysis and the differential coefficients suggested that ether type freon alternatives have relatively small infrared absorption intensity in the 1500-500 cm^-1 region.

View full abstract

We developed a new conformational deformation algorithm (parallel translation method) for the Monte Carlo (MC)method to study the folding of biological macromolecules.
By comparing the structures obtained using the parallel translation algorithm with those obtained using the conventional algorithm (dihedral rotation method), we found that the structure forming tendency by MC simulation using the parallel translation algorithm was different from that using the dihedral rotation algorithm even under the same force field parameters. Dihedral rotation algorithm was effective for C-peptide(single α-helix structure, 13-mer) folding simulation. On the other hand, the parallel translation algorithm was more effective for 28-mer( β β α structure) folding simulation.
With better force field parameters, the parallel translation method will become useful for practical folding simulation of biomacromolecules.

View full abstract

A neural network was applied to the prediction of the carcinogenicity of 41 kinds of organic chlorine-containing compounds. Seven kinds of structural and quantum-chemical descriptors: molecular weight, log P, Gibbs free energy, ionization potential, LUMO energy, HOMO-LUMO energy gap, and Connolly volume were determined. These descriptors were entered into the input layer of a three-layered neural network, and carcinogenicity data from the NTP database were entered into the output layer as teaching data. The network was trained with an error-back-propagation method, and a leave-one-out test showed a correct classification rate of 93%.

View full abstract

Using two programs about summing up 0.1, which is the circulating decimal in binary numbers, 10⁹ times, the efficiency of parallel processing for performance and accuracy of calculation was demonstrated. One program uses sequential summing up (program1), and the other involves summing up using partial sum technique of 10⁴ times 10⁵ times (program2). These programs were parallelized with a message passing interface: MPI. They were executed on 4 parallel computers, Alta Technology AltaCluster, Hitachi SR8000, IBM RS/6000 SP and SGI Origin2000, up to 8 processors. The performance is proportionally improved with increasing number of processors, because the communication process is small compared with the computation process. The computing precision was quite similar for the 4 computers. In the case of program1 the precision was improved drastically by increasing the number of processors, but little improvement was observed in the case of program2. It was clearly shown that the numerical error accumulation, namely the loss of digits, was avoided by parallel processing.

View full abstract

An analyzer called "Q-chanG4" was developed with Visual Basic for photoluminescence quenching in a solid. This software can analyze the quenching mechanism in a solid matrix using emission decay curves and Stern-Volmer plots. Photoluminescence quenching of Tris(2, 2'-bipyridine)ruthenium(II) in a polyethylene glycol solid by methylviologen was analyzed with Q-chanG4. The quenching mechanism was determined as a combination of a dynamic quenching and a static quenching involving multi-step equilibria and dynamic quenching. The dynamic quenching rate constant in the polyethylene glycol was as large as that in an aqueous solution, suggesting that the diffusion coefficient of molecules in the polyethylene glycol is almost the same as that in an aqueous solution in spite of the solid state.

View full abstract