From equations to simulations
Open-source computational physics
Write the equations as you would on paper, each on the region where it holds, and join regions to couple different physics. Choose how to solve them, and differentiate the results. Eqiora checks units before anything runs and records how every result was produced.
Release 0.1.5 · Read the release notes.
At a glance
What Eqiora gives you
Equations as on paper
Write ODEs, DAEs, and PDEs with units. Dimension errors are caught before anything runs.
Batteries included
Import common equations and boundary conditions for flow, solids, circuits, and controls, or write your own.
More areas and one consolidated library in development
Multiphysics by region
Write each equation on the region where it holds and connect the regions to couple them.
Method as a separate choice
Pick finite elements or finite volumes, a time integrator, and a solver without editing the model.
Differentiable
Take derivatives of results with respect to model parameters, from JAX or PyTorch.
Fast, on a Rust core
Runs on CPU threads, with MPI and GPU for selected problems.
Broader MPI and GPU coverage in development
Install and run
uv add eqiora, no compiler needed. Linux wheels today.macOS and Windows wheels in development
Traceable results
Every result records the equations, mesh, and solver settings that produced it.
How it works
Equations, method, result
Write the equations
Write quantities, equations, geometry, and boundary conditions, or import common ones from the standard library. Write each equation on its region and connect regions where they meet.
Choose the method
Pick finite elements or finite volumes, a time integrator, and a solver separately, so the same model can be solved in different ways.
Run and differentiate
Read fields and diagnostics, trace each result to its settings, and compute derivatives with respect to model parameters.
First example
Predict, run, and change a decay model
For dx/dt + kx = 0 with x(0) = 1 and a decay rate of one per second, the model predicts x(1) = exp(-1) ≈ 0.367879. Install Eqiora, run the model, then double the rate and check the new value.
Featured walkthrough
From channel geometry to a vortex street
The Kármán vortex-street example starts with the channel geometry and named boundaries and ends with an evolving vorticity field. The source, numerical setup, and results are on one page.
Explore the flow exampleDocumentation
Where to go next
Get started
Install with uv, run a small model, and change one parameter.
Learn
Work through derivations, physical examples, and exercises.
Guides
Author a model, choose numerics, inspect results, or diagnose a failure.
Gallery
Follow a complete investigation from physical setup to result.
Standard library
Reuse electrical components, sampled controls, fluid and solid laws, and boundary conditions.
Browse the reference
Find language examples, physical components, Python and Rust APIs, and CLI usage.
Contribute
Improve an example, add a numerical method, or work on the tools.
