Well, think of "embedding" an element as a subset. Use a small set like S={1,2,3} , construct P(S) and I think you will see how you can "embed" S in P(S). Then you
can extend this to any set, including infinite sets.
For the set of points of continuity, the argument I know uses the oscillation of a
function over an interval : Osc(f,p) is the oscillation of f at the point p, defined by:
Osc(f,p)=sup|f(x)-f(y)| over all points in neighborhoods containing p. Then f is
continuous at p if Osc(f,p)<1/n for every natural n. Try using this to construct/define
an infinite sequence of open sets (indexed by n , as in 1/n), such that p must
belong to each of them (which means p belongs to their intersection...) for f
to be continuous at p
Note that this oscillation is just a reformulation of the delta-epsilon def. of continuity.
For #3, I have not tried this yet, but try using the definition of Caratheodory measurability and the def. of symmetric difference of sets (which is used in the
first problem.).
For the Cantor set , notice that at each step you are removing an open interval
you can tell what the measure of each removed interval is. You end up with a disjoint
collection of open intervals
Also:what can you say (re basic topology)about the complement of an open interval..
Maybe expressing terms in Cantor set in base 3 will help show that the complement is
dense in the Cantor set C. Take a point p in the complement and consider any ball about
p, show it must contain points of C.
Sorry, got to go. I will be back tomorrow or thursday.