tt_squash_tails
===============

.. py:function:: t3toolbox.backend.tt_operations.tt_squash_tails(tt_cores)

   .. code-block:: python

      def tt_squash_tails(
              tt_cores: typ.Union[
                  typ.Sequence[NDArray],  # ragged. len=d, elm_shape=stack_shape+(rLi,ni,rR(i+1))
                  NDArray,                # uniform. shape=(d,)+stack_shape+(r,n,r)
              ],
      ) -> typ.Union[
          typ.Tuple[NDArray, ...],  # ragged: r0=rd=1. len=d, elm_shape=stack_shape+(1,n0,r1),...,(r(d-1),n(d-1),1)
          NDArray,                  # uniform: same shape, boundary bond summed into slot 0, rest zeroed
      ]:


   Collapse the leading and trailing TT bonds to one without changing the represented tensor.

   Ragged cores shrink to boundary bond 1; a uniform supercore keeps its shape (the bond is
   summed into slot 0 and the remaining slots zeroed, so the boundary rank *mask* becomes 1).