Re: [eigen] Fast QR for 2x2, 3x3

[ Thread Index | Date Index | More Archives ]

2009/2/23 Andrea Arteaga <yo.eres@xxxxxxxxx>:
> Perhaps for 4x4 or more Gram-Schmidt is faster than Givens, but for 2x2 and
> 3x3 I think that Givens are the best way. The only way to test it is to
> implement both algorithms and compare the results.

Sure, that's true!

> Another question: Which product algorithm is used in Eigen?

For such small sizes, it is the brutal direct "algorithm". Except that
it is vectorized for e.g. Matrix2d,4f,4d.
There are other implementations but they are not used for such small
sizes. Even then, they also derive from the brutal "algorithm", we
don't use clever less-than-cubic-complexity algorithms (neither seem
to do other high performance packages).


Mail converted by MHonArc 2.6.19+