3 ms·
What kind of case studies? The number one thing I care about in a scientific computing platform is library support: how does Julia perform on that? Do they hav
by philipov 8y ago
What kind of case studies? The number one thing I care about in a scientific computing platform is library support: how does Julia perform on that?
Do they have ML libraries to compete with PyTorch and Tensorflow? How about geometric calculations such as Voronoi and Convex Hull? Can it handle noneuclidean geometry? Is there a good library for network analysis? GIS? Do they have a symbolic computation library to stand up to SymPy?
If Julia can answer Yes to all of the above, then maybe it's worth checking out.
- acangiano 8y agoIt would seem to me that a priority of the project should be built-in integration with Python or R. A high-level FFI of the sort.
- ChrisRackauckas 8y agoLike RCall.jl and PyCall.jl? These have been around for awhile and are pretty simple to use. Just `@pyimport Package` and now `Package.function()` works, along with other things.
- acangiano 8y agoExcellent.
- ChrisRackauckas 8y agoYes, yes, yes, yes? Flux.jl and KNet.jl. LightGraphs.jl. Etc. You can find these packages on JuliaObserver.com. Some of them are handled by PyCall interfaces: SymPy.jl, SymEngine.jl, QHull.jl, etc.
- philipov 8y agoSweet, PyCall looks awesome and exactly what I was hoping for. Full interop with no copying; I look forward to learning how it was implemented. Thanks! Is there a mature option for calling Julia from Python?
- ChrisRackauckas 8y agoThere's pyjulia. We used it to build the diffeqpy package. It's not too mature yet.