Re: [eigen] Re: LU precision tuning

[ Thread Index | Date Index | More Archives ]

2009/5/8, Hauke Heibel <hauke.heibel@xxxxxxxxxxxxxx>:
> My findings indicate that the bound Benoit defined also holds for
> complex numbers. Just take a look at the attached log file.


> Since the 2048 x 2048 decomposition test for complex numbers did not
> finish after full 8 hours I decided to terminate the program and
> simply submit the results computed so far.

Same here! There's quite a bit of room for improvement in performance
with complex numbers, but much of that would require us to have our
own complex class :(


Mail converted by MHonArc 2.6.19+