let's treat the simple case of only two big circles, each with one child -- you'll then have to figure out how to extend from there as fits your purpose.

you'll see that for each big circle's child to "line up" with the other big circle's child, their centers need to lie on the line connecting the centers of the two big circles.

therefore the problem boils down to finding the angle between a big circle and it's neighbor, which is just atan2(y2-y1,x2-x1), then rotate it's child to that angle relative to the big circle's center.  assuming the other big circle does the same thing with it's child, then both children will be pointing towards each other when done.

(if you're already working with vectors, then don't actually DO the atan2 and rotate, that's the concept, but not the implementation -- instead just normalize the delta vector, scale by the sum of the radii, then add to big circle's origin, and that's the new origin for the child)

hth :)

it's not working, and I have no clue why...i have tried switching to the builtin PVector class but it didn't ork either...can you see anytihng wrong?

void updateNeighbours(){
  aqPos = new PVector(0,0,0);
 
 for (int i = 0; i < z.animals.size(); i++) {
 
 Animal a = (Animal)z.animals.get(i);
  float distance = distance(this.position.x, this.position.y, a.position.x, a.position.y);
   if(distance<100){
     PVector neighbour = new PVector(a.position.x, a.position.y, 0);
      aqPos.add(neighbour);
   
     aqPos.normalize();
     
     aqPos.mult(this.radius+aqRadius);
}


 }

the delta is "other big circle's center" minus "this big circle's center"
(that vector then points from this, to other)
then normalize that
then scale by "this big circle's radius" plus "this child's radius"
then add to "this big circle's center"
..giving "this child's center"

averaging vectors is just as with scalers -- add them all up, divide by count.  however, the resulting vector will likely not point at anything "interesting".  imagine 3 in a rougly equilateral triangle arrangement, they'll each point to the midpoint between the other two.  but what did you expect?  that's what i meant earlier about you having to figure out how to extend it - with only one child, but multiple neighbors, it's not clear what result you would expect.

hth :)

i should also respond that yes given your example of three circles forming a triangle - I would want each one's vector that drives their child circle to be pointing to a common center...

here's my sketch if anyone is willing to comment - it must be something really obvious and small....look in the Animal tab for the area under question:

http://www.upsdn.ca/processing/animal_particles_07h/applet/Animal.pde

http://www.upsdn.ca/processing/animal_particles_07h/applet