Comments on "Inner Product Computers"

same initial elements. For the case of m= 12 this results in a reduction by a factor of about 80 in the number ofcases examined. Samples at m = 13 and 14 indicate that this pruning method will yield an additional improvement factor of about 2.5 to 3 for each increase in m. Using a characteristic vector representation, the test for a repeated spacing is improved by a factor of m2/8 over the straight-forward approach. Fig. 1 summarizes the results of the search program. For m= 10, 11, and 12 the optimum length is isolated in that no Golomb ruler exists which is one longer in length. In particular at 12 the length must be increased by 5 beyond the optimum length before the second ruler is found. It is conjectured that as m be-