Fixed Routes

J. E. Beasley
1984 Journal of the Operational Research Society  
In the section headed "Discussion", a printing error resulted in a loss of sense which misrepresented the author's intended meaning. We apologise for any inconvenience this may have caused and reproduced below is a corrected version of the first two paragraphs of that section. DISCUSSION Looking at the results for the adapted 3-optimal and Christofides algorithms, we have that in nine of the 22 problems they give the same result; in five of the 22 problems the adapted 3-optimal procedure gives
more » ... al procedure gives a better result; and in eight problems the Christofides procedure gives a better result. These results are not particularly surprising in that the (modified) Christofides procedure can be viewed as a 3-optimal interchange on a solution formed by considering frequency of occurrence, whilst the adapted 3-optimal procedure is a 3-optimal interchange on a savings-derived solution. Hence both algorithms can be viewed as a 3-optimal interchange on an initial solution formed in some way. On average, the computer time taken for the modified Christofides procedure is some T times greater than the time taken for the adapted 3-optimal procedure (T large). Even if we had kept to the original Christofides procedure, this computer time difference would not be significantly altered and the results would certainly be worse overall. 276
doi:10.2307/2581930 fatcat:uy57bpvsw5derlltc7ohjn6ex4