How can I simulate repulsion between multiple point charges (ball bearings) in Mathematica?

Question

I'm trying to write a program in Mathematica that will simulate the way charged ball bearings spread out when they are charged with like charges (they repel each other). My program so far keeps the ball bearings from moving off the screen, and counts the number of times they hit the side of the box. I have the ball bearings moving randomly around the box so far, but I need to know how to make them repel each other.

Here's my code so far:

Manipulate[
 (*If the number of points has been reduced, discard  points*)
 If[ballcount < Length[contents], 
   contents = Take[contents, ballcount]];

 (*If the number of points has been increased, generate some random points*)
 If[ballcount > Length[contents], 
  contents = 
   Join[contents, 
    Table[{RandomReal[{-size, size}, {2}], {Cos[#], Sin[#]} &[
       RandomReal[{0, 2 \[Pi]}]]}, {ballcount - Length[contents]}]]];

 Grid[{{Graphics[{PointSize[0.02],

  (*Draw the container*)
  Line[size {{-1, -1}, {1, -1}, {1, 1}, {-1, 1}, {-1, -1}}], 
  Blend[{Blue, Red}, charge/0.3],
  Point[

   (*Start the main dynamic actions*)
   Dynamic[

    (*Reset the collision counter*)
    collision = 0;

    (*Check for mouse interaction and add points if there has been one*)
    Refresh[
     If[pt =!= lastpt, If[ballcount =!= 50, ballcount++]; 
      AppendTo[
       contents, {pt, {Cos[#], Sin[#]} &[
         RandomReal[{0, 2 \[Pi]}]]}]; lastpt = pt], 
     TrackedSymbols -> {pt}];

    (*Update the position of the points using their velocity values*)
    contents = Map[{#[[1]] + #[[2]] charge, #[[2]]} &, contents];

    (*Check for and fix points that have exceeded the box in Y
      direction, incrementing the collision counter for each one*)
    contents = Map[
      If[Abs[#[[1, 2]]] > size, 
        collision++; {{#[[1, 1]], 
          2 size Sign[#[[1, 2]]] - #[[1, 2]]}, {1, -1} #[[
           2]]}, #] &,
      contents];


    (*Check for and fix points that have exceeded the box in X 
      direction, incrementing the collision counter for each one*)
    contents = Map[
      If[Abs[#[[1, 1]]] > size, 
        collision++; {{2 size Sign[#[[1, 1]]] - #[[1, 1]], #[[1, 
           2]]}, {-1, 1} #[[2]]}, #] &,
      contents];

    hits = Take[PadLeft[Append[hits, collision/size], 200], 200];
    Map[First, contents]]]},
 PlotRange -> {{-1.01, 1.01}, {-1.01, 1.01}}, 
 ImageSize -> {250, 250}],

(*Show the hits*)
Dynamic@Show
  [
   ListPlot
   [
   Take[MovingAverage[hits, smooth], -100
    ]
   ,
   Joined -> True, ImageSize -> {250, 250}, AspectRatio -> 1, 
   PlotLabel -> "number of hits", AxesLabel -> {"time", "hits"}, 
   PlotRange -> {0, Max[Max[hits], 1]}], Graphics[]
  ]
}}
  ]
 ,
 {{pt, {0, 1}}, {-1, -1}, {1, 1}, Locator, Appearance -> None},
 {{ballcount, 5, "number of ball bearings"}, 1, 50, 1},
 {{charge, 0.05, "charge"}, 0.002, 0.3},
 {smooth, 1, ControlType -> None, Appearance -> None},
 {size, 1, ControlType -> None, Appearance -> None},
 {hits, {{}}, ControlType -> None},
 {contents, {{}}, ControlType -> None},
 {lastpt, {{0, 0}}, ControlType -> None}
 ]

Answer 1

What you need for your simulation is a "collision detection algorithm". The field of those algorithms is widespread since it is as old as computer games (Pong) are and it is impossible to give a complete answer here.

Your simulation as it is now is very basic because you advance your charged balls every time step which makes them "jump" from position to position. If the movement is as simple as it is with the constant velocity and zero acceleration, you know the exact equation of the movement and could calculate all positions by simply putting the time into the equations. When a ball bounces off the wall, it gets a new equation.

With this, you could predict, when two balls will collide. You simply solve for two balls, whether they have at the same time the same position. This is called A Priori detection. When you take your simulation as it is now, you would have to check at every timestep, whether or not two balls are so close together, that they may collide.

The problem there is, that your simulation speed is not infinitely high and the faster your balls are, the bigger the jumps in your simulation. It is then not unlikely, that two balls over-jump each other and you miss a collision.

With this in mind, you could start by reading the Wikipedia article to that topic, to get an overview. Next, you could read some scientific articles about it or check out, how the cracks do it. The Chipmunk physics engine for instance is an amazing 2d-physics engine. To ensure that such stuff works, I'm pretty sure they had to put a lot thoughts in their collision detection.