reference:
an answer on reddit.com
a lecture note, found with google search
Some proofs are done by myself and might be faulty.
Discussion on Uniqueness of Decimal Expansion
Assume a real number has two different decimal expansion
and , with
if , then even with and
the difference between the two
decimal expansion would still be and they can't
possibly represent the same real number .
if , the only way to make the two expansions
equal is to have and
. This implies is rational and the
denominator is a divisor of .
Thus, we get to the conclusion that has 2 different decimal
presentations iff is rational and the only primes that divide the
denominator are 2 or 5.
Construct the bijection between and
One can see that (consider the isomorphism
). Then . This
means if we can find some bijection between and then we are
done.
Consider the decimal expansion for where
. A naive construction of is
However, this mapping is not well-defined, as it maps
to different numbers although they are
essentially the same point. The problem arises from the 2 different decimal
presentations of the same number. From previous discussion, we know
infinitely repeating 9 is the only case that this construction fails.
What if we add a constraint that, if there are multiple representations possible,
we only use the version to calculate the image of ? (You can also
choose version, and use as "marker" in the next paragraph. Here I
take just for illustration) This will make well-defined and injective,
but still not surjective, because numbers such as will have
no preimage.
Note that the problem with is essentially that, it could map a decimal
expansion without repeating 9 to an expansion containing repeating 9. What
about, instead of working with discrete digits, we work with digit blocks that
never end with 9? For example, would be seen as
, and
. In this way the
image contains infinitely repeating 9 iff its preimage does and vice versa. Then
with the limitation that we only use (instead of ) version of
decimal presentation, turns out a well-defined bijection and we're done.
