Build BatchLAS, link it into a project and call it.

Page	What you find there
Project README	Status, supported operations, install and tuning in one page.
Building, testing and benchmarking	Requirements, tested platforms, presets, CMake options, tests, benchmarks, Python bindings, install traps.
The C++ API	Data-layout and memory contract, queues and events, option structs, workspaces, the linalg layer, entry-point conventions.
Consuming BatchLAS from CMake	An outside project that finds the installed package.
Python examples	The pybind11 package by example.
Ritz values	Rayleigh-quotient eigenvalue estimates for dense, CSR and batched operators.