Abstract
In the present study the assumptions used and the analytical results derived by Qin and Murch (1992, Phil. Mag. A, 66, 957) are tested by computer simulation using the binary random alloy model. The simulation focuses on the calculation of collective cosines which are the average cosines of the angles between a jump of some atom and the successive jumps of any atom of the same or a different species. It is found that the results of the calculations of the <cosθ(1)>, the first term of the collective cosine series, are in reasonable agreement with the analytical results which employ the six-vacancy-jump approximation when the correlation effect is weak. Discrepancies become pronounced, however, when the correlation effect is strong. Calculation of the collective correlation factors by summing up the appropriate <cosθ(j)> gives very good agreement with the results obtained by the ‘displacement method’. Calculation using the approximate equations based only on <cosθ(1)> are only roughly correct implying that the necessary symmetry condition is not well satisfied.