That is not quite what the code does. What it does is it sorts the original
values lowest to highest. Negatives will obviously be the lowest values. When
it is doing the combinations it moves in the direction of adding the next
highest number. If the combination exceeds the target value then it abandons
moving to the following next highest value because it obviously is not a
possible solution. I am not sure that I explained that very well but sufice
it to say that it works and it speeds up the execution by potentially a few
orders of magnitude.
--
HTH...

Jim Thomlinson


"Dave O" wrote:

You're right, Jim, the OP could reduce his solution space by
disregarding the numbers greater than his "target" number. In an
accounting environment, however, debits and credits (positive as well
as negative) may need to be considered- the negative numbers may react
with the positive larger numbers to arrive at the correct solution.