Don't stop just yet!
Carbonated, I don't think there is a problem in 'bottling' chaos. Computers can create chaotic results from very simple formulas. The problem is in predicting chaotic behavior, not creating it.
Okay, Beestie, one chart demonstration of the butterfly effect. This is the granddaddy example from Lorenz himself who stumbled onto these results.
Lorenz had a weather simulator on his computer. It would spit out a row of numbers for each day. It would take maybe a minute to simulate each day. One day, Lorenz was looking at old data, and he decided he wanted to run that particular simulation over again. To save time, he decided he wanted to start the simulation 'from the middle' rather than from the beginning. This seemed easy enough, he input as his initial conditions the numbers from some day in the middle of the data he had on the printout. Now here's the catch - the data on the printout were printed with three decimal places. The computer really kept track of things up to six decimals. In other words, the numbers that Lorenz input were just a tiny tiny bit off from what they had been in with the simulation first time around.
But this tiny difference shouldn't make a difference, right?
But these were his results:
Time is the X axis, and the results from both simulations are plotted on the same graph. Notice how on the left, the beginning of the simulation, there is one solid line. That means both simulations were spitting out almost identical results. But now look as how the two lines diverge over time. By the end of the graph, the two lines are totally apart. That means that with the weather simulation Lorenz was using, tiny differences in initial conditions will result in totally different outcomes. Maybe not at first, but after some time.