Chaotic functions are ones that are incredibly sensitive to initial condtions. So even a very small change in the initial condtions will cause massive changes in the results.
Also check out Chaos by James Gleick.
I had a few lectures on chaos theory last year as part of the wave phenomena course. There are certain types of equations called non-linear partial differential equations, whereas linear equations can be solved "nicely" non-linear equations can get very complicated and are chaotic.
The textbook example of a chaotic function is:
f(n+1) = f(n)*(1-f(n))*k
This is a recursive function, to find the next value sequentially you multiply the previous value by 1 minus the previous value, and multiply by a constant k.
The idea is to take a value of k, and plug in values of f(1) = x and see if the function converges on a specific value as n increases.
Take k = 2.1 and any value as your f(1), i'm going to start with f(1) = 0.2. On my calculator I type in 3, hit enter, and then type in Ans*(1-Ans)*2.1 and because Ans is the previous value, hitting return emulates the recursive function. The function is only well behaved when you use numbers less than 1, convince yourself that a number greater than 1 will cause the function to go to -infinity, and a number less than 1 will have the same effect, so only start with numbers between 0 and 1
so lets see what happens
0.2
0.336
0.4685184
0.5229187068
0.5238969389
...
and after a few more iterations it tends to a number
0.5238095238
That number is called an attractor.
Now try any other value of f(1) between 1 and 0 and you will fall into this 'attractor'.
Lets change k, try k = 2.6, this takes a lot longer to converge, but it does converge to 0.6153846154
Those two examples were not that interesting, but what about k = 3.2? This doesn't converge to a single number, but instead oscillates between 0.5130445095 and 0.7994554905.
But we still have not reached chaos.
Let us now try k = 3.8 and f(1) = 0.5, after 50 iterations we are at 0.5371401901 with no sign of the number converging, now try 50 iterations with f(1)=0.5001, we get 0.6305575200.
That is chaos, a tiny change in the initial conditions, just 1 ten thousandth and after 50 iterations the values were completely different.
If you really want to learn more then do check out the James Gleick book, it is excellent.