in reply to Coming soon: Algorithm::Loops

Actually, when a bit of analysis, it's really easy to come up with those numbers that have unique 6 terms solutions.

A 6 term solution is of the form:

xyz + xzy + yxz + yzx + zxy + zyx

But this can be rewritten as:

(x + y + z) * 222

But this is equivalent with:

((x - k) + (y - l) + (z + k + l)) * 222

So, all you need to find are (x, y, z) such that there is no (k, l) for which ((x - k), (y - l), (z + k + l)) has no duplicates, are less than 10, and 0 or more, and the sum is large enough that there are no solutions with less terms available.

This leads to (9, 8, 7) and (9, 8, 6) as the solutions, and hence to 5328 and 5106 and the only numbers with unique, and six term, solutions.

A simular argument shows that for the four digit problem, only 193314 and 199980 have unique, 24 term solutions. And for five digits, we have unique 120 term solutions for 9066576 and 9333240.

Abigail

Replies are listed 'Best First'.
Re^2: Coming soon: Algorithm::Loops (another analysis of the puzzle)
by tye (Sage) on Apr 13, 2003 at 17:40 UTC
    Update:
    sum is large enough that there are no solutions with less terms available.
    Missed that part.

    Never mind.   - Emila Latella

    Not quite. You've come up with a way to find numbers that have exactly one six-term solution. The problem was to come up with numbers that have exactly one solution and that one solution has six terms (or as many terms as possible while still only having one solution). This analysis can find you good candidates, but you'd have to vet them by demonstrating that they also have no solutions involving fewer terms.

                    - tye