# diffsol-la
Linear algebra foundation for [diffsol](https://github.com/martinjrobins/diffsol) —
the vector, matrix, and linear-solver traits used by the ODE/DAE solvers,
together with concrete backends.
Most users should depend on the [`diffsol`](https://crates.io/crates/diffsol) crate
directly. Use `diffsol-la` when you want to build on the linear algebra layer itself.
## Traits
- **Vectors**: `Vector`, `VectorView`, `VectorViewMut`, `VectorIndex`, `VectorHost`.
- **Matrices**: `Matrix`, `DenseMatrix`, and sparsity handling.
- **Linear operators and solvers**: the `LinearOp` trait describing a linear
operator `A`, and the `LinearSolver` trait for solving `Ax = b`.
- **Backends**: `NalgebraLU`, `FaerLU`, `FaerSparseLU`, `KLU` (suitesparse), `OxideLU` (cuda-oxide) and `CudaLU` (deprecated).
- **Support types**: `Context`, `Scalar`, `Scale`, and the `LaError` error type.
## Implementations
Each implementations is behind a feature flag:
- `nalgebra`: nalgebra-backed containers and solvers (enabled by default).
- `faer`: faer-backed containers and solvers (enabled by default).
- `cuda-oxide`: CUDA containers and solvers with kernels written in Rust via
[cuda-oxide](https://github.com/NVIDIA/cuda-rust), for `f32` and `f64` (disabled by default).
- `cuda`: in-built C CUDA containers and solvers, `f64` only (disabled by default).
Deprecated, use `cuda-oxide` instead.
- `suitesparse`: the KLU sparse direct solver from SuiteSparse (disabled by default).
## Links
- Main crate: <https://crates.io/crates/diffsol>
- Documentation: <https://docs.rs/diffsol-la>
- Book: <https://martinjrobins.github.io/diffsol/>
## License
Licensed under the MIT license. See the [repository](https://github.com/martinjrobins/diffsol) for details.