Re: LAPACK bindings error

"Thomas Klimpel" <[email protected]>
Newsgroups gmane.comp.lib.boost.ublas
Message-ID <trinity-056d3500-5f8e-496c-a23c-e2835e8317ca-1395682088503@3capp-gmx-bs02>
Hi,


> Is this the latest version of boost/numeric/bindings?
> https://svn.boost.org/svn/boost/sandbox/numeric_bindings/boost/numeric/bindings/

Yes, at least that's the version I use. (Maybe I should try to get up to speed with "git", so I could answer such questions with more confidence.)


I think the errors that you showed now should be fixable. The first error I see is
 
/home/dmitry/zfilter/boost_1_55_0/boost/numeric/bindings/lapack/driver/gesv.hpp:
191:9: error: invalid application of 'sizeof' to incomplete type 'boost::STATIC_ASSERTION_FAILURE<false>'
         BOOST_STATIC_ASSERT( (bindings::is_column_major< MatrixB >::value) );
 
One issue here is that the lapack routine xGESV expects two matrices: A of size (n,n), and B of size(n,nrhs). The bindings are generated automatically, and we didn't generate a version to also accept a vector for B. (Not sure whether this would be desirable.)  
In case you already tried to use a matrix for B, you might have forgotten to add "blas::column_major", i.e. "ublas::matrix<double, ublas::column_major>" instead of just "ublas::matrix<double>", because lapack always expects column_major matrices.


Another error I see is
 
/home/dmitry/zfilter/boost_1_55_0/boost/numeric/bindings/detail/property_map.hpp
:30:85: error: no type named 'property_map' in 'struct boost::numeric::bindings::detail::adaptor_access<const boost::numeric::ublas::permutation_matrix<long unsigned int>, void>'
     typedef typename mpl::at< typename adaptor_access<T>::property_map, Key >::type type;

The type "ublas::permutation_matrix" from "numeric/ublas/lu.hpp" is totally unrelated to lapack and the bindings. You just need a vector of integers here (for example ublas::vector<fortran_int_t> or std::vector<fortran_int_t> would work, if you include numeric/bindings/std/vector.hpp), lapack will know how to interpret this as a permutation matrix.

Regards,
 
Thomas Klimpel
_______________________________________________
ublas mailing list
[email protected]
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.