Infinite Recursive Logarithm Sequence

Calculus Level 1

A recursive sequence is defined as a 0 = π a_0=\pi and a n = log a n 1 27. a_n=\log_{a_{n-1}}{27}.

Given that lim n a n \displaystyle\lim_{n\to\infty}{a_n} exists, what is lim n a n ? \displaystyle\lim_{n\to\infty}{a_n}?


Bonus: For what values of a 0 a_0 does lim n a n \displaystyle\lim_{n\to\infty}{a_n} exist?


The answer is 3.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

7 solutions

Chan Lye Lee
Oct 30, 2017

It is a nice question. As a n = log a n 1 27 a_n=\log_{a_{n-1}}{27} implies a n 1 a n = 27 = 3 3 a_{n-1}^{a_n}=27=3^3 which means that if the limit exist (says the limit is lim n a n = x = lim n a n 1 \displaystyle\lim_{n \to \infty} a_n = x=\lim_{n \to \infty} a_{n-1} , then x x = 3 3 x^x=3^3 ), it must be 3. I have not found any algebraic proof. I used Geogebra and notice that the result will still the same if 1.3 a 0 12.73 1.3 \le a_0 \le 12.73 , and the limit does not exist if a 0 a_0 is out of the range. I wish to know why lim n a n \lim_{n \to \infty} a_n does not exist so such a 0 a_0 .

The actual range a 0 a_0 can exist is exclusive of those numbers, because if it is at the boundaries, it would oscillate forever between the 2 numbers, and thus it would never converge to a number. If it is out of the range, the resulting number would eventually become negative, which we cannot take the logarithm of, thus making is impossible.

Kevin Tong - 3 years, 6 months ago
Kevin Tong
Nov 14, 2017

Start off by using change of base for logarithms to simplify the problem. a n = log a n 1 ( 27 ) = log ( 27 ) log ( a n 1 ) a_n=\log_{a_{n-1}}(27)=\frac{\log(27)}{\log(a_{n-1})} Now, let's start calculating the first few numbers in the sequence: a 0 = π a 1 = log π ( 27 ) = log ( 27 ) log ( π ) a 2 = log log π ( 27 ) ( 27 ) = log ( 27 ) log ( log π ( 27 ) = log ( 27 ) log ( log ( 27 ) log ( π ) ) a_0=\pi \\ a_1 = \log_\pi(27)=\frac{\log(27)}{\log(\pi)} \\ a_2 = \log_{\log_\pi(27)}(27)=\frac{\log(27)}{\log(\log_{\pi}(27)}=\frac{\log(27)}{\log(\frac{\log(27)}{\log(\pi)})} In this way, we see that the each consecutive term in the sequence embeds the previous term in itself. Therefore, we can redefine the recursive sequence as: a 0 = π , a n = log ( 27 ) log ( a n 1 ) a_0=\pi, a_n=\frac{\log(27)}{\log(a_{n-1})} Since infinity is not a number, but a concept, then we have lim n a n = log ( 27 ) log ( a n ) \lim_{n \to \infty} a_n = \frac{\log(27)}{\log(a_n)} Now we can solve for the limit: a n = log ( 27 ) log ( a n ) log ( a n ) a n = log ( 27 ) log ( ( a n ) a n ) ) = log ( 27 ) ( a n ) a n = 27 = 3 3 a n = 3 a_n = \frac{\log(27)}{\log(a_n)} \\ \log(a_n)\cdot a_n = \log(27) \\ \log((a_n)^{a_n}))=\log(27) \\ (a_n)^{a_n}=27=3^3 \implies a_n = 3 Bonus Because of the function we just made, we have proven that there is no initial value such that the limit as n approaches infinity is infinity. Therefore, the only way the limit could not exist is if the number eventually gets negative or when the value oscillates between 2 fixed values. So in other words a 1 = a 3 = log ( 27 ) log ( n ) = log ( 27 ) log ( log ( 27 ) log ( log ( 27 ) log ( n ) ) ) a_1=a_3=\frac{\log(27)}{\log(n)}=\frac{\log(27)}{\log(\frac{\log(27)}{\log(\frac{\log(27)}{\log(n)})})} and a 0 = a 2 = n = log ( 27 ) log ( log ( 27 ) log ( n ) ) a_0=a_2=n=\frac{\log(27)}{\log(\frac{\log(27)}{\log(n)})} . Let's just solve for one equation and make sure that the other is not equal. So, to make it simple, let's solve for the second case: n = log ( 27 ) log ( log ( 27 ) log ( n ) ) log ( log ( 27 ) log ( n ) ) n = log ( 27 ) log ( ( log ( 27 ) log ( n ) ) n ) = log ( 27 ) ( log ( 27 ) log ( n ) ) n = 27 1 ( log ( n ) ) n = 27 ( log ( 27 ) ) n ( log ( n ) ) n = ( log ( 27 ) ) n 27 n 1.29543 or 12.73327 n=\frac{\log(27)}{\log(\frac{\log(27)}{\log(n)})}\\ \log(\frac{\log(27)}{\log(n)})\cdot n = \log(27) \\ \log((\frac{\log(27)}{\log(n)})^n)=\log(27) \\ (\frac{\log(27)}{\log(n)})^n= 27 \\ \frac{1}{(\log(n))^n}=\frac{27}{(\log(27))^n} \\ (\log(n))^n=\frac{(\log(27))^n}{27} \\ n \approx 1.29543 \textrm{ or } 12.73327 Upon inspection, it is clear that the number becomes negative if it is not between these two numbers, so this means that the interval on which a 0 a_0 can exist is a 0 ( 1.29543 , 12.73327 ) a_0 \in (1.29543, 12.73327)

Should line four read ( log ( 27 ) log ( n ) ) n = 27 \left(\frac{\log(27)}{\log(n)}\right)^n = 27 ?

Daniel Ford - 3 years, 6 months ago

Log in to reply

Yes, I fixed it

Kevin Tong - 3 years, 6 months ago

Log in to reply

Yes, but then how can you divide? Don't you have 1 ( log ( n ) ) n = 27 ( log ( 27 ) ) n \frac{1}{(\log(n))^n} = \frac{27}{(\log(27))^n}

Daniel Ford - 3 years, 6 months ago

Log in to reply

@Daniel Ford Thanks, I fixed it.

Kevin Tong - 3 years, 6 months ago

I don't understand.

Michael Wang - 3 years, 6 months ago

Log in to reply

Kill yourself better...

Michael Wang - 3 years, 6 months ago

Pump you...

Michael Wang - 3 years, 6 months ago

jUST jOKING....

Michael Wang - 3 years, 6 months ago

I have a question about the line in which you defined f ( x ) f(x) . You defined f ( x ) = log x log π f(x) = \frac{\log{x}}{\log{\pi}} , and stated lim n a n = f ( 27 ) \lim_{n \rightarrow \infty} a_n = f(27) . Why should the limit be f ( 27 ) f(27) ? Also, this implies that lim n a n = log 27 log π 2.879 \lim_{n \rightarrow \infty} a_n = \frac{\log{27}}{\log{\pi}} \approx 2.879 , which is not 3, the actual limit. Can you explain this part, please?

Marian Minar - 3 years, 6 months ago

Log in to reply

Sorry for the bad wording. f ( x ) f(x) is not supposed to describe a n a_n , but rather, it is something embedded inside of itself and infinite amount of times to get a n a_n . In the a 2 a_2 case, we see that it is embedded in itself once. If we observe what happens for the subsequent terms in the sequence, we notice that as n n\to \infty , the function f ( x ) f(x) becomes embedded in itself an infinite amount of times, and therefore, I got the equation f ( 27 ) = log ( 27 ) log ( f ( 27 ) ) f(27)=\frac{\log(27)}{\log(f(27))} . I couldn't find the right wording to describe it, so I just called it a function during the time, which now that I look at it, is pretty confusing.

Kevin Tong - 3 years, 6 months ago

I just edited my answer. Hopefully it makes more sense now.

Kevin Tong - 3 years, 6 months ago

Is there any ways to nicely compute the last given values by hand, assuming you used wolfram alfa or some.

Matthijs Lau - 3 years, 6 months ago

Log in to reply

I couldn't think of a nice way to calculate it by hand, which I always prefer instead of calculator. I tried to find a way, but in the end, I got to n 2 7 1 n = 27 n^{27^{\frac{1}{n}}}=27 which I don't think is solvable. I then tried using calculus by taking the limit as k approaches 0 of the integrals of both sides from n to n+1/k, which did not turn out well (I ended up getting the exponential integral, which I do not want to deal with).

Kevin Tong - 3 years, 6 months ago
Chew-Seong Cheong
Nov 12, 2017

The plot of a n a_n against n n shows the convergence clearly.

As mentioned by @Chan Lye Lee , a n = log a n 1 27 a_n = \log_{a_{n-1}} 27 , a n 1 a n = 27 = 3 3 \implies a_{n-1}^{a_n} = 27 = 3^3 . If lim n a n \lim_{n \to \infty} a_n exists, then lim n a n \lim_{n \to \infty} a_n = lim n a n 1 = 3 \lim_{n \to \infty} a_{n-1} = \boxed{3}

Bonus: We note from the graph that the values of a n a_n oscillate up and down. The limit does not exist when a n a_n oscillates between two values a 0 a_0 and a 1 a_1 . This happens when a 0 a 1 = a 1 a 0 = 27 a_0^{a_1} = a_1^{a_0} = 27 . Let these two values be α \alpha and β \beta , then the limit exist if α < a 0 < β \alpha < a_0 < \beta and it is found that α 1.29543 \alpha \approx 1.29543 and β 12.73327 \beta \approx 12.73327 . Note that α β = β α = 27 \alpha^\beta = \beta^\alpha = 27 .

this might be right

ONKE banny - 3 years, 6 months ago
Eric Kim
Nov 18, 2017

a = log log log . . . 27 27 27 a = \log _{\log _{\log _{...} 27} 27} 27

a = log a 27 a = \log _{a} 27

a a = 27 a^a = 27

a = 3 a = 3

Leonel Castillo
Nov 18, 2017

The first question one should ask is: What about π \pi is important in this problem? In this case, it is a property that π \pi has been known to have for thousands of years. A fundamental theorem everyone should know, that π > 3 \pi > 3 . So how can we apply this? Well, notice that 27 = 3 3 27 = 3^3 so log 3 27 = 3 \log_3 27 = 3 and therefore a 1 = log π 27 < 3 a_1 = \log_{\pi} 27 < 3 . But now that a 1 < 3 a_1 < 3 this implies that a 2 = log a 1 27 > 3 a_2 = \log_{a_1} 27 > 3 . So we have the general property that a 2 n > 3 a_{2n} > 3 and a 2 n + 1 < 3 a_{2n+1} < 3 . Now, we know that the limit exists so let's try to conjecture where that limit must be. If the limit is bigger than 3 then that would be contradictory as the a 2 n + 1 a_{2n+1} would never converge to a number bigger than 3. If the limit was smaller than 3 then that would be contradictory as the a 2 n a_{2n} would never converge to something smaller than 3. But we know the limit exists so it must be exactly 3 3 .

Jason Balarusan
Nov 19, 2017

My solution was: When you're lazy on a Sunday night, type '1, 2, and 3' into the answer box. Looks like I was correct.

Samuel Walker
Nov 16, 2017

Easy, just use an iterative method starting with π as the base of the logarithm

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...