True, you need to assign weights to a vector of outputs as discussed in this article. After that, scipy.optimize is easy to use though.