Create OCO problems and use them as benchmarks for OCO algorithms.
An OCO algorithm. Is instantiated with the number of dimensions and (optional) hyper-parameters. Must implement the
An OCO problem. Gives access to the function value, gradient and Hessian at any point in its domain. Some have built-in update functions that represent the time-varying element of the problem.