Joshua Pulsipher
Joshua Pulsipher
This came up with updating https://github.com/zavalab/JuliaBox/blob/master/InfiniteDimensionalCases/CaseStudy1/stochastic_optimal_powerflow.jl. Since expectations look for using supports with the `WeightedSample` label for use with expectations with respect to random parameters, any user defined supports are...
JuMP enables users to define set constraints the following two ways: ```julia julia> model = InfiniteModel(); julia> @infinite_parameter(model, t in [0, 1]); julia> @variable(model, x[1:2], Infinite(t)); julia> y = [0.5,...
Currently, we allow multi-dimensional infinite parameters to be arbitrary JuMP containers. This adds quite a bit of complexity (e.g., with the use of `VectorTuple`s) and leads to some ambiguous typing....
JuMP has started making changes to allow expressions with complex numbers. We should follow suit and enable this in InfiniteOpt. There are a number of places where `Float64` is hardcoded...
A useful capability would be to save/load `InfiniteModel`s via a file format.
In like manner to JuMP, we should define `Base.copy(model::InfiniteModel)`.
Implement JuMP's direct mode in combination with optimizer models to achieve performance enhancements. This may help with #18.
A modeling capability we should add are functions that have stochastic outputs, but not necessarily stochastic inputs (e.g., a Gaussian process). A motivating example pertains to subproblems in Bayesian optimization...
In connection with #105, as we better decouple the transformation/solution schemes from JuMP we should enable `InfiniteModel`s to handle vector-valued nonlinear functions. This should equate to an extension of our...
Following #213, we should had a FAQ page in the documentation that addresses a number of common questions that common in domain specific areas (e.g., optimal control). It might also...