Hi Bogdan,

Am 30.11.2011 um 00:22 schrieb Bogdan Burlacu:
I also compiled your example and boost::numeric::ublas computes the
correct result. Using gentoo as well and boost 1.47 (sorry the system
is not so up-to-date).

Thanks a lot, I have tested it again with Boost 1.47 on my two OSX systems (on Gentoo only the 1.48 version is installed),
but also with the 1.47 version the result with ublas::norm_2 I get the 1.46 value for the result.

Okay, I have build the Boost myself, I run boostrap.sh than b2 with:

b2 --with-exception --with-filesystem --with-math --with-random --with-regex --with-date_time --with-thread --with-system --with-program_options --with-serialization --with-iostreams --disable-filesystem2 threading=multi runtime-link=shared variant=release toolset=gcc|darwin install 
I think a hardware / system problem can be excluded, because on my 3 systems  the problem is shown.
The ublas structures are header-only, aren't they? So linking problems can be also exclude.
You and Thomas tested the example code without any problem.

Is there a compiler problem or compiler option problem on my system?
I'm very confused. My next test: I build the Boost again...

Phil