Register to reply

Hard Sequence problem

by aznluster
Tags: sequence
Share this thread:
aznluster
#1
Dec20-10, 12:15 AM
P: 5
Find [tex]\sqrt{1+\sqrt{1+2\sqrt{1+3\sqrt{1+...}}}}[/tex]
Phys.Org News Partner Mathematics news on Phys.org
'Moral victories' might spare you from losing again
Fair cake cutting gets its own algorithm
Effort to model Facebook yields key to famous math problem (and a prize)
QuantumJG
#2
Dec20-10, 04:39 AM
P: 32
It comes out to be 3. Why? I'm trying to find out.
QuantumJG
#3
Dec20-10, 05:01 AM
P: 32
Ok so this is basically one of Ramanajan's identities:

[tex] x + n + a = \sqrt{ax + (n+a)^{2} + x \sqrt{a(x +n) + (n+a)^{2} + (x + n) \sqrt{...}}}}[/tex]

aznluster
#4
Dec20-10, 11:14 PM
P: 5
Hard Sequence problem

Wouldn't that add up to 2? Or maybe I'm doing something wrong. Is there a proof of that identity somewhere?
Dickfore
#5
Dec20-10, 11:45 PM
P: 3,014
Let

[tex]
x_{n} = \sqrt{1 + n \sqrt{1 + (n +1) \sqrt{1 + \ldots}}}
[/tex]

Then, we have the backwards recursive formula:

[tex]
x_{n} = \sqrt{1 + n x_{n + 1}}
[/tex]

We want to find [itex]x_{1}[/itex]. Let us see what is the limit [itex]x \equiv \lim_{n \rightarrow \infty}{x_{n}}[/itex]?



If we want a finite limit, we must have:

[tex]
x_{n} \sim -\frac{1}{n}, \; n \rightarrow \infty
[/tex]

but, this is impossible since all [itex]x_{n} > 0[/itex]. Therefore, the limit is infinite. Let us assume that is asymptotically a power law:

[tex]
x_{n} \sim A \, n^{\alpha}, \; n \rightarrow \infty, \; \alpha > 0
[/tex]

Then, we have:

[tex]
A n^{\alpha} \sim \sqrt{1 + n A (n + 1)^{\alpha}} \sim \sqrt{A} \, n^{\frac{1 + \alpha}{2}}
[/tex]

which means:

[tex]
\alpha = \frac{1 + \alpha}{2} \Rightarrow \alpha = 1
[/tex]

and

[tex]
A = \sqrt{A} \Rightarrow A = 0 \vee A = 1
[/tex]

So, we may conclude:

[tex]
x_{n} = n (1 + y_{n}), \; y_{n} \rightarrow 0, \; n \rightarrow \infty
[/tex]

The backwards recursive relation for [itex]y_{n}[/itex] is:

[tex]
n (1 + y_{n}) = \sqrt{1 + n \cdot n (1 + y_{n + 1})}
[/tex]

[tex]
y_{n} = \sqrt{1 + y_{n + 1} + \frac{1}{n}} - 1
[/tex]

[tex]
y_{n + 1} = y_{n} (2 + y_{n}) - \frac{1}{n}
[/tex]

Define a discrete z-transform:

[tex]
Y(z) \equiv \sum_{n = 1}^{\infty}{y_{n} z^{-n}}
[/tex]

As we can see, [itex]y_{1}[/itex] is the coefficient in the Laurent series of [itex]Y(z)[/itex] around [itex]z = 0[/itex]. According to the residue theorem:

[tex]
y_{1} = \mathrm{Res}(Y(z), z = 0) = \frac{1}{2 \pi i} \, \oint_{C_{z = 0}}{Y(z) \, dz}
[/tex]

We need to find a functional equation for [itex]Y(z)[/itex] from the above (non-linear) recursive relation.


Register to reply

Related Discussions
Sequence analysis of the Fibonacci sequence using matrices? Calculus & Beyond Homework 7
Energy of a sequence - absolutely summable sequence or square summable sequence Engineering, Comp Sci, & Technology Homework 0
Next number in sequence, or pattern of sequence. Calculus 2
Formula for the sequence c defined by Sequence b Calculus & Beyond Homework 17
Proving a given sequence is a delta sequence ~ Calculus & Beyond Homework 1