Looping question

[Robocop]

Having just started using C again after some years off, i've been
stumped by a problem i think someone more experienced could probably
solve pretty easily. I have these 4 objects (vectors), and i want to
find a combination of these 4 objects into two pairs, where the sum of
a specific attribute of each vector in each pair comes closest to some
base value. So all i want to do is run through the permutations of
these 4 objects and find the combination that results in two pairs
looking as close as possible to some base value.

The only thing i could think of creating a temporary vector,
initiating it by just dumping in one permutation, then looping over my
vector with the 4 objects in it, doing different permutations,
comparing those to the values in the temporary vector, and if this
permutation is better, pushing back the current pair values into the
temporary vector. After all the looping is done, i would just use the
temporary vector as my best fits.

Could anyone think of a possible solution that is simpler? I'm
currently trying to implement this solution but it's not quite
working. Any comments would be greatly appreciated.

[Richard Heathfield]

Robocop said:

Having just started using C again after some years off, i've been
stumped by a problem i think someone more experienced could probably
solve pretty easily. I have these 4 objects (vectors), and i want to
find a combination of these 4 objects into two pairs, where the sum of
a specific attribute of each vector in each pair comes closest to some
base value. So all i want to do is run through the permutations of
these 4 objects and find the combination that results in two pairs
looking as close as possible to some base value.

The only thing i could think of creating a temporary vector,
initiating it by just dumping in one permutation, then looping over my
vector with the 4 objects in it, doing different permutations,
comparing those to the values in the temporary vector, and if this
permutation is better, pushing back the current pair values into the
temporary vector. After all the looping is done, i would just use the
temporary vector as my best fits.

Could anyone think of a possible solution that is simpler? I'm
currently trying to implement this solution but it's not quite
working. Any comments would be greatly appreciated.

If you absolutely must have a best fit, then brute force is the only way -
try everything, see what works best. Sounds like you're already doing
that.

If you can get away with a "good enough for rock n' roll" fit, however,
your options widen considerably. For example, you might want to consider a
straight Monte Carlo, or shotgun hillclimbing, or a genetic algorithm.

But you want simple, right? Well, I think that puts us right back to brute
force again.

If you have C code that isn't working, I suggest you throw out as much of
the code as you can without affecting the brokenness that you're
experiencing (but don't throw out error-checking, or we'll only tell you
to put it back in!). Very often, performing this reduction exercise will
allow you to find the bug yourself. If it doesn't, however, at least it
will leave you with a program suitable for posting to comp.lang.c so that
we can take a look and maybe spot the problem.

--
Richard Heathfield <http://www.cpax.org.uk>
Email: -http://www. +rjh@
Google users: <http://www.cpax.org.uk/prg/writings/googly.php>
"Usenet is a strange place" - dmr 29 July 1999