OSRI Simulation Lab
On this page
Find a simulation tool, learn from a published model, and keep enough information for someone else to repeat the work.
- Browse the tool catalog: 60 tools across robotics, engineering, finance, economics, operations, energy, natural sciences, and computing.
- Inspect the 21 candidate attempts: author-code runs, native outputs, comparisons, adaptations, and verified blockers.
- Read the queueing methods note on reproduced confidence intervals and duplicate random streams.
- Plan an example for any of the 60 tools: paper artifacts, construction steps, comparison targets, and unresolved access needs.
- Choose a literature example: paper locations, source code, setup routes, outputs to inspect, and execution status.
- Run the CPU examples: battery discharge, option pricing, household saving, and OpenUSD result exchange.
- Read the bibliography or download tools, papers, examples, and BibTeX.
Choose a question#
| What would you like to investigate? | Start here |
|---|---|
| How does a robot learn to move in a simulated world? | Isaac Sim, Isaac Lab, and OpenUSD |
| How do battery voltage and discharge time relate? | PyBaMM's published code example |
| How do order flow and execution change a simulated market? | ABIDES |
| How can simulated payoffs be checked against a pricing formula? | Black–Scholes with QuantLib |
| How does income risk affect household saving? | Carroll's model with Econ-ARK |
| How does a flexible structure respond to flow? | The Turek–Hron benchmark with preCICE |
| How do networks, dispatch, and capacity affect an energy system? | PyPSA-Eur |
| How do staffing and demand affect waiting? | A published SimPy and Ciw service model |
| How do routing and signals affect urban traffic? | SUMO's LuST scenario |
A browser is enough to read papers, inspect models, compare assumptions, and contribute a source. The CPU examples need a Python environment. The Isaac walkthrough needs a compatible NVIDIA GPU and its documented runtime. Larger models have their own requirements.
Understand the evidence#
The catalog is a growing inventory, not a claim to list every open-source simulator. It includes solvers, modeling frameworks, and supporting scene tools. A source-code license does not necessarily cover a complete runtime, an input dataset, or a model asset. Isaac Sim and CARLA particularly require reading the dependency notes. OpenUSD composes scenes and data; the numerical calculation remains in the selected solver.
Each example states its relationship to the paper. Executed adaptation means the included teaching code ran and passed the listed checks. Upstream instructions checked means the authors' or maintainers' code and run route were inspected but not executed here. Access pending means an artifact still needs retrieval. None of these labels alone establishes reproduction of a published result.
Sources were checked on 2026-11-15. See verification, source and access notes, and the contribution guide.
Start a study#
Keep a new study in its own research repository, with a defined question, responsible human, model version, inputs, assumptions, and intended comparison. This hub holds shared learning resources. Use the simulation workflow and run-record template when developing a study. Human review and acceptance remain separate from automated checks.