>>11741649I work in Applied Mathematics. More specifically, numerical analysis. Here are some common parameters which people screw with. Many of them are not misunderstood like a true fudge factor, but still there is less rigor when these parameters are used.
>Floating point machine accuracy>Initial step size selection of line search in numerical optimization>CFL number in diffusion-free fluid dynamics solvers>Anything having to do with neural networks in Optimal Control algorithms>Quadratic cost matrices in an Linear-Quadratic Regulator>The choice of inner product or preconditioner in a CG or BCG linear system solver 
