From 4f18c363233e1ce5acf71d65f2f9096d03d46de7 Mon Sep 17 00:00:00 2001 From: Ian Weaver Date: Tue, 13 Oct 2026 23:09:35 -0700 Subject: [PATCH] docs: Add Aether.jl --- .../stars_and_galaxies/doraemonho-aether.md | 19 +++++++++++++++++++ ecosystem.asdf | 10 ++++++++++ 2 files changed, 29 insertions(+) create mode 100644 docs/case_studies/stars_and_galaxies/doraemonho-aether.md diff --git a/docs/case_studies/stars_and_galaxies/doraemonho-aether.md b/docs/case_studies/stars_and_galaxies/doraemonho-aether.md new file mode 100644 index 0000000000..5516221d68 --- /dev/null +++ b/docs/case_studies/stars_and_galaxies/doraemonho-aether.md @@ -0,0 +1,19 @@ +--- +title: "`Aether.jl`: A High-Performance 3D MHD and Multifluid Dust Code Written in a Dynamic Language with an Interactive Human-AI Development Framework" +id: "doraemonho-aether" +description: "See how Julia runs production astrophysical MHD simulations on everything from an interactive notebook to thousands of GPUs on the Frontier exascale supercomputer." +author: "[Ka Wai Ho](https://github.com/doraemonho)" +date: 2026-08-14 +--- + +Developed at the Kavli Institute for Theoretical Physics (KITP), Aether is a finite-volume code for compressible hydrodynamics and magnetohydrodynamics (MHD), with any number of dust fluids coupled to the gas through drag. It is designed to solve the two-language problem in computational astrophysics: the same Julia code used to prototype in a notebook is the code that runs in production on a cluster. + +**Package: [MHDFlows/Aether.jl](https://github.com/MHDFlows/Aether.jl)** + +![Orszag-Tang Vortex](https://arxiv.org/html/2608.14048v3/fig_orszag_tang.png) + +Every performance-critical kernel is written once with [KernelAbstractions.jl](https://github.com/JuliaGPU/KernelAbstractions.jl) and runs on CPUs and on NVIDIA, AMD, Apple, and Intel GPUs. On a single GPU, Aether's throughput is comparable to or higher than C++ codes on the same hardware, and in weak scaling tests on Frontier it keeps above 93% parallel efficiency out to 4096 GPU compute dies (GCDs). Aether was also built from scratch in an interactive human–coding agent workflow, with every design decision and development session recorded in the repository. + +See the paper below on arXiv for more: + +> Ho, K. W. (2026). Aether.jl: A High-Performance 3D MHD and Multifluid Dust Code Written in a Dynamic Language with an Interactive Human-AI Development Framework. [arXiv:2608.14048](https://arxiv.org/abs/2608.14048). [[ADS]](https://ui.adsabs.harvard.edu/abs/2026arXiv260814048H/abstract) diff --git a/ecosystem.asdf b/ecosystem.asdf index 95a82cbdf9..126468dbfa 100644 --- a/ecosystem.asdf +++ b/ecosystem.asdf @@ -450,6 +450,16 @@ Techniques: descr: | - `VLBISkyModels` provides an interface and library for of models that can be used to describe the on-sky emission seen by VLBI interferometers astropy: [] + Simulation: + Aether.jl: + repo: "https://github.com/MHDFlows/Aether.jl" + doc: "https://mhdflows.github.io/Aether.jl/stable/" + tagline: "🌌 Fast and Flexible Julia code for simulating Astrophysical MHD on CPUs and GPUs" + descr: | + - Finite volume hydrodynamics and MHD with constrained transport in Cartesian, cylindrical, and spherical-polar coordinates + - Any number of dust fluids coupled to the gas through stiff drag + - Kernels written once with [KernelAbstractions.jl](https://github.com/JuliaGPU/KernelAbstractions.jl) run on CPUs and NVIDIA, AMD, Apple, and Intel GPUs, with optional GPU-aware MPI + astropy: [] Statistics: Uncertainty: Measurements.jl: