The classical jeep problem (see [Fin], [Phi], [Gal]) is to compute how far a jeep can go when starting at a dump with n cans of gasoline, provided that it needs 1 canful to drive 1 unit distance and it is allowed to carry 2 canfuls of gasoline (including the contents of its tank). The optimal solution algorithm (see [Gal]) assumes that arbitrary fractions of a canful can be put into the tank. Since this seems to be a bit unrealistic, a discrete variant was considered by D. Wood in [Woo]: It is only allowed to refill the tank, when it is empty and then 1 canful has to be filled in (thus 1 can can be transported). The following solution was proposed in [Woo]: With one canful transport n − 1 cans to the next dump and go on recursively. This gives the distance function f w (n) = 1 + 1 + 1/3 + 1/5 + ... + 1/(2n − 3). It seemed to be the natural adaptation of the classical optimal solution.
Unable to display preview. Download preview PDF.