07-10, 14:00–14:30 (Europe/Amsterdam), Else (1.3)
Julia's SciML represents nearly a hundred of some of the most widely used Julia packages. How's it doing? In this talk we will discuss many of the latest improvements, where we are lacking, and the near future roadmap. This is made to be a celebration of our wins and an appreciation of our lingering challenges, highlighting to the community what to be proud of but also what needs more time to cook.
SciML is huge. If I say "I am using a SciML package", that could mean DifferentialEquations.jl, but it could also mean NonlinearSolve.jl, or ExponentialUtilities.jl, MuladdMacro.jl, or anything in the long tail. Yet it is treated, maintained, and documented as a cohesive whole. But in terms of maturity, that is definitely not the case. This talk will highlight and put firm grading on the maturity of different parts of the project, where pieces like the ODE solvers are highly mature while other aspects like GPU-based optimizers or high index DAEs have a medium level of maturity, while other promising and popular libraries such as MethodOfLines.jl have a lot Discourse discussion but are knowingly at an immature stage. Part of this talk is to paint in broad strokes a picture of the current state of the ecosystem to help the general user base better understand the current state of the project.
But I think another major point is really, what's next? Some pieces that are immature are the main focus of the current development, especially aspects like boundary value problems, complementary problems, and parallelism in nonlinear optimization. Other areas such as uncertainty quantification schemes have been progressing but comparatively lack the team activity in comparison to some of the other focus areas. We hope to outline the lay of the land but also provide some perspective on the driving forces behind this progression to both highlight our near future goals but also indicate the road blocks that potential contributors can use as a starting point for helping the project themselves.
Dr. Chris Rackauckas is the VP of Modeling and Simulation at JuliaHub, the Director of Scientific Research at Pumas-AI, Co-PI of the Julia Lab at MIT, and the lead developer of the SciML Open Source Software Organization. For his work in mechanistic machine learning, his work is credited for the 15,000x acceleration of NASA Launch Services simulations and recently demonstrated a 60x-570x acceleration over Modelica tools in HVAC simulation, earning Chris the US Air Force Artificial Intelligence Accelerator Scientific Excellence Award. See more at https://chrisrackauckas.com/. He is the lead developer of the Pumas project and has received a top presentation award at every ACoP in the last 3 years for improving methods for uncertainty quantification, automated GPU acceleration of nonlinear mixed effects modeling (NLME), and machine learning assisted construction of NLME models with DeepNLME. For these achievements, Chris received the Emerging Scientist award from ISoP.