-
Join us for a Dyad Modeling Livestream today - this time at 1pm ET / 10 am PT! Michael Tiller will joining us today to model a hybrid-EV powertrain!
Tune in on YouTube and send us your thoughts in the chat!
-
New fastest explicit non-stiff ODE solver? That's right, we now have something beating the pants off of the high order explicit RK methods! Check out the new symbolic-numeric optimized Taylor methods available in DifferentialEquations.jl! It uses a mix of Taylor-Mode AD, a symbolic post-processing trick, and a new order adaptivity algorithm to give a new level of performance.
See the paper: https://arxiv.org/abs/2602.04086
-
Your college professor teaches you "A-stable methods are required for stiff ODEs". But PSA, the most commonly used stiff ODE solvers (adaptive order BDF methods) are not A-stable. #sciml #numericalanalysis #diffeq
-
Physics-Informed Neural Surrogates for Mesh-Invariant Modeling of High-Speed Flows at #AIAA #SciTech!
High-speed flight simulation is computationally brutal. A single CFD run can take hours on a cluster. That's fine for final validation, but not for early design exploration or real-time decision-making.
Neural surrogate that predicts aerodynamic behavior 595x faster than CFD while maintaining ~1% relative error.
Paper: https://lnkd.in/efe2Q_T9
-
Scientific machine learning (#SciML) is not just about adding scientific information to machine learning, but it's also about making machine learning accessible to heterogeneous data.
-
New livestream, #Dyad Modeling Live! In this stream we built up a thermal model of a room using #AgenticAI and added a heat pump with different control strategies and analyzed the power efficiency. Join the fun live next week! #julialang #sciml
-
How to properly cook your turkey, using agentic AI and #dyad #julialang! Happy Thanksgiving from JuliaHub, hope we can help you with acausal modeling to change your family's holiday!
-
ANSYS /Synopsys, one of the largest simulation companies in the world, is partnering with JuliaHub in order to bring #Dyad, #Julialang, and #SciML to next level of adoption. We have many things planned. This is how research becomes reality.
-
#SciML fact of the day: automatic differentiation fails to give the correct derivative on a lot of very simple functions 😱 😱 😱 . #julialang #automaticdifferentiation
-
Can Agentic AI turn single purpose code into reusable modular code? Dyad's specialized AI can!
Watch our latest video on AI-assisted model restructuring and physics enhancement:
https://www.youtube.com/watch?v=0RdA-t9_VocLearn more: https://help.juliahub.com/dyad/stable/
#ModelingAndSimulation #AIAgent #JuliaLang #SciML #Dyad #SystemsEngineering #Modelica
-
Check out the latest #Julialang Dispatch Podcast! This episode goes over the changes to the Julia language, its standard libraries, and the main changes in the package ecosystem (#sciml, JuMP, etc.) that occurred over the summer. Most posts coming soon as well!
-
Watch Dyad's AI agent build a complete thermal model from just an image! Picture -> validated DAEs in minutes.
Features: Auto parameter generation, model optimization, custom animations. All with production-ready Julia code.
-
What is #acausal modeling and how does it lead to better reproducibility and modularity in modeling and simulation? Check out this video which goes step-by-step into building acausal models using the RC circuit and RLC circuit
-
#LLMs make mistakes. Modeling languages like #Dyad have static analysis to compile-time check whether models are physically possible. What happens when you mix the two in an #agentic workflow? Automated construction of accurate models!
-
SciML Developer Chat Episode 1: Base Splits and Symbolics Precompilation
Welcome to the first episode of the SciML Dev Chat! We discuss the latest developments in the #SciML (Scientific Machine Learning) ecosystem for #julialang!
-
MIT Julia Lab: looking for postdoctoral researchers in #julialang open source software development, scientific machine learning (#SciML), and systems biological / pharmacological modeling (#QSP) for research in equation discovery for large stiff systems.
-
New workshop materials on High-Performance Scientific Modeling with Julia & SciML:
• #Julialang for scientific computing
• ODEs/PDEs & numerical methods
• Symbolic-numeric modeling
• Biological systems (Catalyst.jl)
• Parameter estimation
• #SciML & UDEs -
DifferentialEquations.jl is many things, and lots of people only use a small portion of it. Check out the JuliaCon 2025 workshop: introduces many aspects of the packages that the developers feel are underutilized and under-understood!
-
Check out the JuliaCon workshop SciML in Fluid Dynamics (CFD): Surrogates of Weather Models.
Go into depth on all of the major surrogate architectures and give them a try on the weather model challenge problem! -
Sundials.jl v5.0: Update to SUNDIALS v7 and Improved DAE Initialization
A major update that brings significant improvements to differential-algebraic equation (DAE) solving and upgrades to the latest Sundials C library
-
SciML Fellowship development 2025 - JumpProcesses.jl: Introducing vr_aggregator for VariableRateJumps, GPU-enhanced SimpleTauLeaping and Extending with τ-Leap Algorithms for Jump Process Simulation
#julialang #ssa #gillespie #sciml #tauleap
https://sciml.ai/news/2025/09/13/jumpprocesses_sciml_fellowship/
-
Introducing SymbolicSMT.jl for symbolic constraint solving and theorem proving! Built on Z3, test the feasibility of symbolic expressions built using Symbolics.jl. Given Constraints([x > 0, y > 0, x^2 + y^2 <= 1]), ask issatisfiable? isprovable?
-
Mixed precision and automatic GPU offloading for Newton solves and stiff ODE solvers has now landed thanks to improvements in LinearSolve.jl. Run LinearSolveAutotune.jl so that things like usage of GPUs in #julialang #sciml becomes automatic!
-
New Symbolic-Numeric algorithm which uses rational polynomial interpolation mixed with differential algebra in order to give a highly robust method for parameter estimation and solving inverse problems on ODEs.
#julialang #sciml #symbolicnumeric #ode
https://www.sciencedirect.com/science/article/pii/S0096300325003649?dgcid=coauthor
-
New blog post: Implicit ODE Solvers Are Not Universally More Robust than Explicit ODE Solvers, Or Why No ODE Solver is Best
Talks about how "robust" methods can be less robust in practice. Justifies hundreds of methods in #julialang #sciml
-
LinearSolve.jl just got autotuning 🔥 The problem: picking linear solvers is hardware-dependent voodoo. Solution: automatic benchmarking that tells you EXACTLY what to use. Share the results to improve the system!
Details: sciml.ai/news/2025/08/16/linearsolve_autotuning/
-
What is the #julialang programming language? This is stitched from the interviews of a lot of developers telling their stories as to why they are committed to the mission of #Julia. Performance, scaling, easy, fun?
-
New paper on fitting #neuroscience models to neuroimaging and electrophysiological data! We show that the previous techniques within the field (Spectral DCM) can be greatly improved using automatic differentiation, #julialang #sciml, ModelingToolkit, and more.
This demonstrates how the optimization and Bayesian estimation of the Julia SciML tooling could be the tool analyzing your fMRIs in the near future!
See the link for the full paper: https://direct.mit.edu/imag/article/doi/10.1162/IMAG.a.88/131630/Increasing-spectral-DCM-flexibility-and-speed-by
-
New blog post: Machine learning with hard constraints: Neural Differential-Algebraic Equations (DAEs) as a general formalism.
#sciml #ai4science #hardconstraints #neuralnetworks #dae #acausal #modelingtoolkit #julialang #modelica
-
Earn money working on open source software #oss! New project just posted: help make wrappers to connect Symbolics.jl to SymPy. $300 bounty. Information for signing up for the #SciML small grants program are contained in the link:
https://sciml.ai/small_grants/#create_wrapper_functions_to_sympy_for_symbolicsjl_300
-
Is your software stack #quantum ready? The #julialang #sciml differential equation solvers are able to to not only target CPUs, GPUs, and IPUs with good performance, but quantum computers as well through the QuDiffEq.jl backend without changing your code. Check out this work where a group of researchers tested its accuracy for modeling power systems dynamics, showing its correctness and readiness for real-world DAEs!