If you created it like:

boost::numeric:ublas::vector<double> my_vec;
/// whatever

it will deallocate itself. This is the preferred way.

if you need a pointer ( generally try to avoid this unless you need some sort of dynamic allocation):

auto my_vec = new boost::numeric:ublas::vector<double>();

you better use a a shared_ptr or unique_ptr depending on what you are trying to do, like:

auto my_vec = std::make_unique<  boost::numeric:ublas::vector<double> >();

or 

auto my_vec = std::make_shared< boost::numeric:ublas::vector<double> >();

// now my_vec will again deallocate itself when it goes out of scope. Typical use case for the unique pointer is when you want to use it in only in one place. The second one (shared) is useful when you want to pass the vector around (in other objects or functions).

Please excuse this probably simple question, but I couldn't find the answer anywhere. The question is: do uBLAS arrays (e.g. vector, matrix) deallocate themselves when going out of scope, or should one use shared_ptr s to them to make sure they are deallocated?

Thanks, and sorry if this is trivial; I'm new to boost, and also to c++ ;-)