This line on page 6 confuses me a bit --For example, for applications to rotations, quaternions are demonstrably more efficient than the vectorial and matrix m

This line on page 6 confuses me a bit --For example, for applications to rotations, quaternions are demonstrably more efficient than the vectorial and matrix methods taught in standard physics courses And then claims that aerospace engineers mostly use quaternions. It's from 2002, so it's sense of what's in vogue might be dated, but I'd understood at least in the graphics manipulation world, the vector/matrix methods were dominant. Those are separate industries, so both statements could be true, anybody have opinions?