Using College to Decipher a MATHCOUNTS Classic
by djmathman, Mar 8, 2016, 5:34 PM
It's always fun when you can take the things you learn in college math class and apply them in unexpected situations 
Solution

Quote:
Let
be a fixed positive real number. Prove that
has a defined value. In other words, show that the infinite nested radical doesn't diverge.

![\[\sqrt{x+\sqrt{x+\sqrt{x+\cdots}}}\]](http://latex.artofproblemsolving.com/5/7/b/57b378c77e648c8154c3c13e99aa31f5a6b3df7d.png)
Solution
Let
be a sequence of real numbers satisfying
and
for all
. It's clear that this expression as
goes to infinity becomes the expression we wish to compute (e.g.
,
, etc.), so it suffices to show that the
converge to some real value as
.
I will now cite a powerful lemma that will help us save work.
LEMMA 1: Let
be a sequence of real numbers which is monotonic (meaning it's either always increasing or always decreasing) and bounded. Then
is also convergent.
Proof (hidden - contains some analysis concepts)
Before we continue, we introduce another lemma which ties in to the previous one.
LEMMA 2:
for all
.
Proof. We use induction. The base case,
, is easy. Now for the inductive step, assume that
for some
. Then
Hence by induction we're done.
Now back to the main problem. Note that by Lemma 2 the
form a monotonically increasing sequence. Hence by Lemma 1 it suffices to show that the
are bounded, i.e. there exists a number
such that
for all
.
To do this, we shall in fact show that
i.e. the solution in
to the equation
. The proof goes by induction. The base case is trivial. For the inductive step, assume that
for some
. Then
so by induction we're done.
Hence we have shown that the
are bounded above by
. Combining this with our previous work and the fact that this isn't that AIME #7 in the sense that all of our square roots are positive, we're done. 









I will now cite a powerful lemma that will help us save work.
LEMMA 1: Let


Proof (hidden - contains some analysis concepts)
Assume WLOG that the sequence
is non-decreasing; the case where the sequence is non-increasing is similar. Let
denote the set of all
(
). Note that since
contains
, we know that
is nonempty. Furthermore, it is given that the sequence
is bounded above, so
must be bounded above as well. It follows by the Supremum Property in
that
has a supremum. Denote
; it suffices to show that
as
.
Let
. Suppose for the sake of contradiction that the
do not converge to
. By the definition of convergence, this means that one cannot find
such that for all
we have
. Note that since
is non-decreasing,
for all
. Hence
implies
for all
. It follows that since this is in direct contradiction with our assumption, we have that
for all
- and hence it also holds true for all
(otherwise non-decreasing would be violated). This in particular means that
is an upper bound for the set
. However, we established that
- contradiction! Hence we can find a
such that
for all
, meaning that indeed
as
.














Let























Before we continue, we introduce another lemma which ties in to the previous one.
LEMMA 2:


Proof. We use induction. The base case,



![\[\sqrt{x+y_{k+1}}>\sqrt{x+y_k}\quad\implies y_{k+2}>y_{k+1}.\]](http://latex.artofproblemsolving.com/9/0/b/90b6422741ed4543c3e6cf21a00bf609d7e87d94.png)
Now back to the main problem. Note that by Lemma 2 the





To do this, we shall in fact show that
![\[y_k<L=\frac{1+\sqrt{1+4x}}2,\]](http://latex.artofproblemsolving.com/b/d/d/bdd1708ea12011eb3e185556417e9689dd916422.png)




![\[y_{k+1}=\sqrt{x+y_k}<\sqrt{x+L}=L,\]](http://latex.artofproblemsolving.com/1/d/9/1d981a999f9550c6298f060f97c1d2d64479d49d.png)
Hence we have shown that the



This post has been edited 4 times. Last edited by djmathman, Mar 8, 2016, 6:07 PM