A polychord indicates an explicit superimposition of two or more chords, with full voicing, disallowing the omission of any notes. It is usually used to simplify the notation of an otherwise awkward chord symbol [think of Ebm7 over G, which could result in a bulky GMaj7(#9#11b13)], but can also make clear that the composer was thinking specifically about chord superimpositions, instead of chord alterations.
It is worth noting that a polychord can be also used to avoid the forced attribution of any harmonic function to a chord when there is no strict tonal context (for instance, a cluster, a non-tertian chord or a passing chord).
A polychord can be voiced in any way you want (open, close, interlocked or not), depending on any specific texture you intend to create.
Finally, it is rarely necessary to try seeing chords as inversions of other chords. One should trust the composer/arranger's initiative to notate a chord in such deterministic way, as to give an exact character, to influence the voicing of the arrangement, the conduction of the bass voice, and many other traits of the work being played/analyzed.
C# E G B
, I could call that chord Em6 or C#m7b5 (or even C#dim#7). For most series of notes there will be multiple ways you can notate it. It's the same thing with polychords - polychords can be notated as extended voiced chords (e.g. Db/G = G7#11b9) if you want, but sometimes the polychord notation is just easier to read at a glance. – absinthe Sep 3 '14 at 7:04