Hello Julia community,
Today we are introducing the SciML Open Source Software Organization for Scientific Machine Learning: high performance differential equation solving with automated model fitting and discovery plus neural network accelerated methods. For the full explanation, please see our introduction blog post.
Hello Julia community,
Awesome! Can I contribute a project? If so, how do I go about doing it?
Just get involved with any of our packages. Open an issue saying you’re interested and someone can show you the ropes. We do also transfer in some packages as well, but that requires a thorough review since, as mentioned in the blog post, we are committed to maintaining everything in our organization to a certain standard, so we have to make sure it meets that standard (autodiff compatibility, hardware support, connectivity with differential equation and deep learning libraries, performance, etc.).
Thanks. What it would be would be a wrapper of a project or two of the tools inside the organization, so nothing major but definitely useful! I’ll let you all know when the project approaches that state.
Note that it doesn’t need to be in the full state before moving in: there are projects still under heavy development in SciML like ReservoirComputing.jl. That said, it’s about knowing that there’s enough people in the organization such that, if someone opens an issue about performance or AD, that we are committed to answering that issue and solving it. We do not want to have repositories where we do not have that kind of commitment since that would ruin the trust that people build in us, but we understand that software is never perfect and it has to start somewhere.
Makes perfect sense to me. I appreciate the standard the core packages in that group have maintained thus far. How I first heard about Julia was “it has world class DiffEq libraries”. I presume many others are in the same boat. And it’s possible when this thing is all done it doesn’t fit well with the scope of the org, no hurt feelings there either.
Just happy to see this twist on ML becoming a reality because it is a breath of fresh air compared to the “Add more layers” modus operandi of the state of things today. Keep crushing it everyone and thank you all for your dedication and contributions!