🚀 go-pugleaf

I am trying to typeset a linear algebra equation of the form:

( word )     ( 1 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 1 0 0 ) ( word )
( word )     ( 0 0 1 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 1 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )  =  ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 1 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 1 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )
( word )     ( 0 0 0 0 0 0 ) ( word )

using  eqn(1)  running on the UNIX(r) operating system, but have
found that

  1)  when set using the "matrix" construct  eqn(1)  uses up its
      string space before it is done. The transform matrix is
      16x7=102 entries and  eqn(1)  can only create 100 named
      strings.

  2)  when set using "piles"  eqn(1)  processes the equation just
      fine but  troff(1)  gives up due to "word overflow".

Please help me overcome these bizarre and arbitrary limits.

Thanks.

:Doug
(Doug "There can be no ersatz for substitution." Lewan)