gem - Optimization Library in Ruby -
i writing machine learning/classification algorithm in ruby, , optimize of parameters in algorithm. have 4 variables, p, x, y, , z, 0 <= x + y + z <= 1 , 0 <= p <= 1. clearly, optimization problem.
however, because thing trying maximize not mathematical function per se (i.e. cannot represented in matrix, no obvious notion of derivative, etc) rather call program, return value classification accuracy, cannot find right tool job. program takes 5 minutes per run, it's super expensive begin with. know of ruby libraries need? or need switch language? (in case, how optimize program written in ruby using script in language?)
things i've tried/looked at:
rglpk - has linear programming, objective function (clearly) non linear
sciruby - not developed (e.g. doesn't have equivalent scipy.optimize)
rbgsl - close. can specify own objective function (nonlinear). can specify step function , multiple variables optimized. can seem specify bounds on variables, despite c version of multimin having argument specify constraints. missing something?
Comments
Post a Comment