RustAlgebloat
Do you love template bloat? Do you think waiting for your program to compile is awesome? Do you think complicated error messages are the best thing since sliced bread? If so, this linear algebra library is just for you!
Packages
- algebloat - The main library.
- algebloat_macros - Useful macros.
Documentation
See here.
Example
Some basic operations and row access (no allocations except during the initial matrix creation!):
let m = &mat![1.0, 2.0, 3.0;
4.0, 5.0, 6.0;
7.0, 8.0, 9.0];
println!("m =\n{}", m);
let t1 = m.t();
println!("t1 =\n{}", t1);
let r = m.row(0) + t1.row(0);
println!("r =\n{}", r);
let m2 = stack![m.view( ..2, ..2), m.view( ..2, 1..);
m.view(1.., ..2), m.view(1.., 1..)];
println!("m2 =\n{}", m2);
Output:
m =
âŽ¡1 2 3âŽ¤
âŽ¢4 5 6âŽ¥
âŽ£7 8 9âŽ¦
t1 =
âŽ¡1 4 7âŽ¤
âŽ¢2 5 8âŽ¥
âŽ£3 6 9âŽ¦
r =
[2 6 10]
m2 =
âŽ¡1 2 2 3âŽ¤
âŽ¢4 5 5 6âŽ¥
âŽ¢4 5 5 6âŽ¥
âŽ£7 8 8 9âŽ¦
Features
- WIP! This is very much incomplete... stay tuned!
- Expression templates (well, more like expression traits since this is Rust) assure all the caveats above while in principle providing allocation-free speed (only available with optimizations turned on)!
- Matrices
- Elementwise operations (right-hand side can be a scalar)
- Binary operators (
*/-+
) - Unary negation
- Binary functions
pow
atan2
hypot
- Unary functions
- Trigonometry/exponential functions
ceil
floor
ln
log10
sqrt
- Reductions
min
max
- Binary operators (
- Row and column access
- Views
- Multiplication
- Reshaping
- Stacking
- Flat views (view matrix as a vector)
- Element access
- Slicing
- Elementwise operations (right-hand side can be a scalar)
Building
Via Cargo:
./cargo_util.py --build