Hmm, we'll need a little more info about what exactly you're trying to do to help you -  I don't understand what you mean by "1 attracted and 1 unattracted."  You mean as in one end of an object is positively charged and one is negative?  That's a dipole, and http://en.wikipedia.org/wiki/Dipole is as good a place to start as any.

For collision detection, you've just stepped into one of the most irritating areas of physical simulation.  Ellipse-ellipse collisions are notoriously tricky to calculate, simply because the equations are complicated.  I'm not entirely sure if there's a simple formula that will give you the answer - I've never used one, at least.  For circle-ellipse collisions, http://mathworld.wolfram.com/Circle-EllipseIntersection.html has enough for you to probably get it eventually.  Honestly, though, if I were you, I would approximate each ellipse as a series of line segments in order to solve that type of collision, as long as you have the CPU cycles to spare in your loop.  If you really want to do it, head to http://www.kevlindev.com/blog/?p=15 to see an example in Javascript; the meat is in http://www.kevlindev.com/gui/2D.js and to me it's absolutely unreadable because each function is collapsed onto a single line.  But there you go...