[nraynaud]

I agree, I have had to learn Newton's methode for differential equations (for simulation), I've had to use logic and an automatic prover (for system validation), I've had to do some complexity analysis (for everything), I've had to play with linear applications, matrices and all the stuff around 3D, I've had to dig into geographical projection equations for another job (in GIS). I'm currently toying with computational geometry. I think what is important is some kind of math to understand how those people write (which far from obvious), and then random stuff to learn to switch.