Assume that and none of the is a nonpositive integer. Then
has at most finitely many zeros if and
only if the can be re-indexed for in such a way that
is a nonnegative integer.
Next, assume that and that the and the quotients
are all real.
Then has at most finitely many real
zeros.
These results are proved in Ki and Kim (2000). For further information on
zeros see Hille (1929).