scipy.optimize.fmin

Abstraction: Gradient-free function minimization using Nelder-Mead simplex algorithm

Key points:

• Minimizes a scalar function using the downhill simplex (Nelder-Mead) algorithm; requires no gradient information
• Key parameters: func, x0 (initial guess), xtol (relative error in xopt), ftol (relative error in func value), maxiter, maxfun
• Returns xopt (optimal parameters) and optionally fopt, iteration count, function evaluation count, warnflag
• retall=True returns the solution at every iteration; disp=True prints convergence messages
• No gradient needed, but convergence can be slow for high-dimensional problems

Connections: Scipy · Numerical Optimization · Nelder Mead · Mathematical Optimization

Source: http://docs.scipy.org/doc/scipy-0.7.x/reference/generated/scipy.optimize.fmin.html