Writing an eigenvalue solver in Rust for WebAssembly

4 months ago

The author implemented an eigenvalue solver in Rust for WebAssembly to visualize the Gershgorin Circle Theorem.
The Gershgorin Circle Theorem helps bound the region of complex plane where matrix eigenvalues lie by using diagonal elements as centers and the sum of off-diagonal elements' norms as radii.
The QR algorithm was used for eigenvalue computation, involving iterative QR decomposition until convergence, with enhancements like Hessenberg reduction for better performance.
Complex number arithmetic was implemented in Rust, including operations like addition, multiplication, and norm calculations, due to the absence of a built-in complex type.
Matrix operations, including QR decomposition and Householder reflectors, were implemented to support the eigenvalue solver.
The project was compiled to WebAssembly using wasm-pack, with JavaScript bindings auto-generated by wasm-bindgen for seamless interoperation.
The interactive component was embedded in a blog post using Hugo, showcasing the solver's capabilities with examples like Hadamard matrices.

Hasty Briefsbeta