Croatica Chemica Acta, Vol. 66 No. 1, 1993.
Original scientific paper
Weighted Self-Returning Walks for Structure-Property Correlations
Danail Bonchev
; Department of Marine Sciences, Texas A&M University, Galveston, Texas 77553-1675, U. S. A.
Xiaoyu Liu
; Department of Marine Sciences, Texas A&M University, Galveston, Texas 77553-1675, U. S. A.
Douglas J. Klein
; Department of Marine Sciences, Texas A&M University, Galveston, Texas 77553-1675, U. S. A.
Abstract
Two algorithms are developed for the weighting of self-returning walks in molecular graphs. Algorithm I modifies the adjacency matrix of a graph by a parameter accounting for the Kekule structure’s Pauling bond orders of conjugated molecules. Algorithm II specifies a two-parameter function which is a weighted sum of the self-returning walks of different lengths. The two procedures are shown to improve the structure-property correlations for benzenoid hydrocarbons compared to the cases with non-weighted walks and with molecular connectivity indices. Five properties of benzenoids are studied: heats of atomization and formation, gas-chromatographic retention indices, n-octanol/water partition coefficients and melting points. The best result has been found for the heats of atomization (with statistical parameters r = .9995, o = 1.11, F = 9442, N = 12).
Keywords
Hrčak ID:
137100
URI
Publication date:
1.6.1993.
Visits: 878 *