For reasons which escape me, the pdsolve() command syntax for solving pdes analytically and numerically is slightly different. (I personally keep being caught out by this, and it is really annoying!).
The syntax for analytic solutions is that the pdes+bcs/ics are given as a single set/list as in
pdsolve( [PDEs, BCs/ICs], ...some options)
whereas the syntax for solving pdes numerically is that the pdes ans ics are given as two separate sets/lists as in
pdsolve( [PDEs], [BCs/ICs], numeric ...some options)
Note that the syntax for ODEs is "consistent", because for analytic problems, it is
dsolve( [ODEs, BCs/ICs], ...some options)
and for numeric solutions, it is
dsolve( [ODEs, BCs/ICs], numeric ...some options)
So the pdsove/numeric case is a bit of an "oddity".
However, even if you get the syntax correct, the Maple limitation described by rlopez kicks in - but at least you will get the somewhat "more informative" error message
Error, (in pdsolve/numeric) unable to handle elliptic PDEs