> <\body> interface> completes by providing the user with special variants for matrices with numerical type entries: <\itemize> |analyziz/matrix_double.hpp>, |analyziz/matrix_floating.hpp>, |analyziz/matrix_ball.hpp>. Computations of eigen values and vectors are implemented in the following files: <\itemize> |analyziz/eigen.hpp> contains the generic implementation, |analyziz/eigen_floating.hpp> uses the latter solver with gradual precision doubling, |analyziz/eigen_ball.hpp> includes certification with ball coefficients. interface> <\session|mathemagix|default> <\input> <|input> use "analyziz"; type_mode? := true; <\input> <|input> rnd () == { \ \ x == uniform_deviate (0.0, 1.0); \ \ return ball (x, 0.0000001); }; <\unfolded-io> <|unfolded-io> M == [ rnd () \| i in 0..3 \|\| j in 0..3 ] <|unfolded-io> ||>>|]>>: |)>> <\unfolded-io> <|unfolded-io> significant_digits := 5; // print 5 digits only eigen_solve M <|unfolded-io> |-0.49790+0.052058*\|-0.65354+0.031053*\>||0.90368-0.094485*\|0.33724-0.016024*\>||-0.12633+0.013209*\|0.75461-0.035855*\>>>|]>>: |)>|)>> <\unfolded-io> <|unfolded-io> v == eigenvalues M <|unfolded-io> >: |)>|)>> <\unfolded-io> <|unfolded-io> radius v[0] <|unfolded-io> : . If you don't have this file, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.>