# Nexus

> Differentiable energy-system optimisation for Python. Nexus solves capacity
> expansion, economic dispatch, unit commitment and optimal power flow at exact
> parity with PyPSA and GenX, and additionally differentiates through the
> optimisation to recover the techno-economic inputs behind an observed
> dispatch — answering "which of my inputs is wrong?" in a handful of solves
> instead of a sampling campaign.

Two packages. `nexus-energy` is the energy-system modelling layer and is what
most users want; it installs `nexus-opt`, the Rust solver core, automatically.

```bash
pip install nexus-energy
```

Every code example in the guides below has been executed against a clean
install; printed outputs are real. Where the library's scope stops, the guides
say so explicitly rather than omitting it.

## Guides

- [nexus-energy user guide](https://raw.githubusercontent.com/VishalRam24/nexus-energy/main/WIKI.md): Energy-system optimisation: capacity expansion, dispatch, unit commitment, power flow — plus differentiable inverse calibration that recovers the hidden inputs behind an observed dispatch. Markdown. Rendered at https://vishalram24.github.io/nexus/docs/?p=energy
- [nexus-opt user guide](https://raw.githubusercontent.com/VishalRam24/nexus-opt/main/WIKI.md): The Rust-core solver underneath: LP, MILP, QP, conic, nonlinear, black-box, multi-objective and stochastic programming behind one API. Markdown. Rendered at https://vishalram24.github.io/nexus/docs/?p=opt
- [Everything in one file](https://vishalram24.github.io/nexus/llms-full.txt): both guides concatenated as markdown — the complete documentation in a single fetch.

## Source

- [nexus-energy](https://github.com/VishalRam24/nexus-energy): source, tests and benchmarks. Released on PyPI as `nexus-energy` (https://pypi.org/project/nexus-energy/).
- [nexus-opt](https://github.com/VishalRam24/nexus-opt): source, tests and benchmarks. Released on PyPI as `nexus-opt` (https://pypi.org/project/nexus-opt/).
- [Project site](https://vishalram24.github.io/nexus/): overview, and an explicit statement of what the library does not claim.

## Optional

- [Benchmark scorecard](https://github.com/VishalRam24/nexus-energy/blob/main/COMPARISON_SCORECARD.md): every measured head-to-head, including the one open case where nexus under-counts OPEX by ~4.3% and reports it as a bug rather than a win.
- [Feature matrix](https://github.com/VishalRam24/nexus-energy/blob/main/FEATURE_MATRIX.md): parity table against PyPSA, GenX, Calliope, oemof, SpineOpt, Sienna, PowerModels and Tulipa.