You can build almost anything in the functions, but you'll find most things covered in the red nodes already.
My knowledge of math is very low, but there are some usefull things to know when using them. Try and stay within 0 and 1 for your arguments, ie black to white. If you multiply a signal with 0 to 1, you are effectively blending it, ie 1 x 0 = 0 (white multiply by 0 = black). Conversly white multiply by 1 = 1 (white). Anything between are values of grey, or in number terms 0 to 1 with every possible decimal place in between like 0.0000001 to 0.9999999 for example. But remember not all signals are between 0 to 1. You have to get used to what pumps out what.
Take a power fractal, it creates values between -1 through 0 to +1, the displacement value in the PF multiplies that result.....
The thing is, is that what you do with the blue nodes can be really basic, like mixing two power fractals. eg PF1 - PF2 = a new noise shape. In photoshop and most art packages you have blending modes, which are essentialy just simple math equations.
You can use things like a 'conditional scaler' to say if a point in the scene is say black, then I want to do This, if it isn't then carry on. I know pure mathmaticians don't like using the conditional scaler, but to be honest, it's an easy way to look at your arguments, even if it isn't purely mathmaticly efficient.
Although the documentation isn't essentially a maths text book, it does give you a lead on what it does. If you want more detail, then have a look at wikipedia to help you some more.
