Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/flaport/umfpack-rs
Some UMFPACK bindings for rust.
https://github.com/flaport/umfpack-rs
linear lu rust solve sparse suitesparse system umfpack
Last synced: 3 days ago
JSON representation
Some UMFPACK bindings for rust.
- Host: GitHub
- URL: https://github.com/flaport/umfpack-rs
- Owner: flaport
- License: lgpl-2.1
- Created: 2023-02-26T03:36:27.000Z (almost 2 years ago)
- Default Branch: master
- Last Pushed: 2023-11-03T21:52:03.000Z (about 1 year ago)
- Last Synced: 2025-01-04T01:51:41.580Z (5 days ago)
- Topics: linear, lu, rust, solve, sparse, suitesparse, system, umfpack
- Language: Rust
- Homepage: https://crates.io/crates/umfpack-rs
- Size: 769 KB
- Stars: 2
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# UMFPACK-rs
Some [UMFPACK](https://github.com/DrTimothyAldenDavis/SuiteSparse/tree/dev/UMFPACK) bindings for rust.
The `umfpack-rs` library provides unsafe bindings and safe wrappers to some of the
SuiteSparse UMFPACK routines. You can use the wrappers to solve a sparse linear system
that is either real-values (`f64`) with `umfpack_di_{symbolic,numeric,solve}` or complex
valued (`Complex64`) using `umfpack_zi_{symbolic,numeric,solve}`.
## Example
> Note that the rust wrappers attempt to be as close as possible to the SuiteSparse C-code
> while hiding unsafe operations. This means the exposed API might not be as clean as one
> might expect.
```rust
#[allow(non_snake_case)]
fn main() {
use umfpack::prelude::*;
let n = 5;
let Ap = vec![0, 2, 5, 9, 10, 12]; // column pointers of CSC sparse nxn matrix
let Ai = vec![0, 1, 0, 2, 4, 1, 2, 3, 4, 2, 1, 4]; // row indices of CSC sparse matrix
let Ax = vec![2.0, 3.0, 3.0, -1.0, 4.0, 4.0, -3.0, 1.0, 2.0, 2.0, 6.0, 1.0]; // values of CSC sparse matrix
let b = vec![8.0, 45.0, -3.0, 3.0, 19.0]; // dense target
let mut x = vec![0.0, 0.0, 0.0, 0.0, 0.0]; // initial value for unknown x
let control = Control::new(); // default control parameters
let mut info = Info::new(); // empty info buffer
// solving the system Ax=b happens in three steps:
// 1. Symbolic Analyzation of the sparse system
let mut symbolic = Symbolic::new();
umfpack_di_symbolic(
n, // m
n, // n
&Ap,
&Ai,
&Ax,
&mut symbolic,
Some(&control),
Some(&mut info),
);
// 2. Numeric Analyzation of the sparse system
let mut numeric = Numeric::new();
umfpack_di_numeric(
&Ap,
&Ai,
&Ax,
&symbolic,
&mut numeric,
Some(&control),
Some(&mut info),
);
// 3. Solving of the sparse system
umfpack_di_solve(
UMFPACK::A, // solve the system Ax=b
&Ap,
&Ai,
&Ax,
&mut x,
&b,
&numeric,
Some(&control),
Some(&mut info),
);
println!("symbolic walltime: {}", info.umfpack_symbolic_walltime());
println!("numeric walltime: {}", info.umfpack_numeric_walltime());
println!("solve walltime: {}", info.umfpack_solve_walltime());
for i in 0..(n as usize) {
println!("x [{}] = {:.1}", i, x[i]);
}
}
```
```
symbolic walltime: 0.000018095000086759683
numeric walltime: 0.0004187900001397793
solve walltime: 0.000003099999958067201
x [0] = 1.0
x [1] = 2.0
x [2] = 3.0
x [3] = 4.0
x [4] = 5.0
```
You can find more examples in the `examples` folder (sometimes alongside with the C++
equivalent). To learn more on how to use UMFPACK, read the [user guide](./assets/UMFPACK_UserGuide.pdf)
or [quick start pdf](./assets/UMFPACK_QuickStart.pdf).
## Installation
```bash
cargo add umfpack-rs
```
## License & Credits
© Floris Laporte 2023, LGPL-2.1
This library vendors, wraps and statically links to [SuiteSparse](https://github.com/DrTimothyAldenDavis/SuiteSparse), LGPL-2.1.