I intend to implement an optimization algorithm which requires the computation of the Lipschitz constant. My function is a multivariate function with more than 50 variables. I am wondering whether anyone knows of any methods to estimate the Lipschitz constant as the publications that I've found only deal with the univariate, bivariate, or trivariate case.
Suggestions are welcome!
If your function is differentiable (convex), the best global Lipschitz constant is the supremum of the norm of the derivative (subgradients).