Continuing With Continued Fractions

Calculus Level 2

1 = 3 2 1 = 3 2 3 2 1 = 3 2 3 2 3 2 1 = \begin{array} {l l} 1 & = 3 - \frac{2}{1} \\ & = 3 - \frac{2}{ 3 - \frac{2}{1} } \\ & = 3 - \frac{2}{ 3 - \frac{2} { 3 - \frac{2}{1} } } \\ & = \ldots \\ \end{array}

Observe the construction of the continued fraction above.

What is the value of

3 2 3 2 3 2 3 = ? 3 - \dfrac{2}{ 3 - \dfrac{2} { 3 - \dfrac{2} { 3 - \ddots } } }= \ ?


Inspiration, see solution comments

1 2 0 Does not exist

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.

3 solutions

Joey Hunt
Feb 7, 2016

The continued fraction is defined as the limit of the sequence

a 0 = 3 a_0=3 ,

a 1 = 3 2 3 a_1=3-\frac{2}{3} ,

a 2 = 3 2 3 2 3 a_2=3-\frac{2}{3-\frac{2}{3}} , ...

a n = 3 2 a n 1 a_n=3-\frac{2}{a_{n-1}} .

If the sequence converges, say to a a , then a a satisfies the equation a = 3 2 a a=3-\frac{2}{a} . Multiplying by a a on both sides and solving the resulting quadratic equation shows that if the limit exists, it is either a = 1 a=1 or a = 2 a=2 .

We see that a 0 > 2 a_0>2 . We proceed by induction to show that a n > 2 a_n>2 for all n > 0 n>0 , which would then imply that the limit would have to be a = 2 a=2 if it exist.

Assume (by hypothesis) that a n 1 > 2 a_{n-1}>2 . Then

1 a n 1 < 1 2 \frac{1}{a_{n-1}}<\frac{1}{2}

2 a n 1 > 2 2 -\frac{2}{a_{n-1}}>-\frac{2}{2}

a n = 3 2 a n 1 > 3 1 = 2 a_n=3-\frac{2}{a_{n-1}}>3-1=2

So if the limit exists, it must be 2. To show that the limit exists we show that the sequence is decreasing and bounded below (by 2). To show that the sequence is decreasing, assume that it isn't (and hope for a contradiction). Then there is some n > 0 n>0 so that

a n > a n 1 a_n>a_{n-1}

3 2 a n 1 > a n 1 3-\frac{2}{a_{n-1}}>a_{n-1}

0 > a n 1 2 3 a n 1 + 2 0>a_{n-1}^2-3a_{n-1}+2

which is true only if 1 < a n 1 < 2 1<a_{n-1}<2 . This is a contradiction since a n > 2 a_n>2 for all n 0 n\geq 0 . Therefore { a n } \{a_n\} is a decreasing sequence bounded below and must converge. We saw above that the limit must be 2.

Mathematics is ideology.

No matter how a person answers this, it is wrong because there are not enough premises given to make a reasonable deduction. The poster never said whether the sequence is a finite or infinite sequence.

In any finite sequence, the answer will be 1. In an infinite sequence, the answer will approach 2.

The answer you give also depends heavily on the order of operations in which you calculate. If you move from left to right.. you might see that it is 3-2/3... but the "2/3" part is actually approaching 2/2(ie 1)... so the answer would be approaching 3-1=2 in that sequence.

But in any finite/rational sequence, the order of operations would dictate that the go out from the most nested point in computation.... so the answer will be 1 in a finite sequence.

And so, the answer you give depends on what you are giving weight to... Are you giving priority/weight to finitude and the order of operations (PEMDAS)? Or are you giving priority/weight to the concept of infinite regress and ignoring the order of operations?

It gets more even complex if you don't know what the denominator will be in a finite sequence. If the denominator is very small and close to 0, you will end up subtracting a large number, and end up with a sequence that goes from positive to negative constantly. 3-2/.00000000000000000000000000000000000000000000000(infinite zeros)00001= -infinitely large number. but 3-2/(-infinitely large number)=infinitely large positive number...

Depending on which mathematical principle you give priority to and what you presume is the denominator part of that sequence and if you presume that the sequence is finite or infinite your answer will be radically different.

Charles Allison - 5 years, 3 months ago

Log in to reply

The comment of

In any finite sequence, the answer will be 1. In an infinite sequence, the answer will approach 2.

is false.

As calculated above, a 0 = 3 , a 1 = 2 1 3 , 15 7 a_0 = 3, a_1 = 2 \frac{1}{3}, \frac{15}{7} .


The issue with your truncation, is that you are "replacing it with 1", which isn't justified. Furthermore, that is not the standardized process in which continued fractions are calculated.

Calvin Lin Staff - 5 years, 3 months ago

Log in to reply

False. You truncated my response(which is intellectually dishonest).

Given that the denominator is 1 and given that it is a finite sequence, and all calculations will start out from the most nested point(because of the order of operations), the answer will always be 1.

3-2/1=1

3-2/(3-2/1)=1

3-2/(3-2/(3-2/1))=1

3-2/(3-2/(3-2/(3-2/1)))=1

3-2/(3-2/(3-2/(3-2/(3-2/1))))=1

3-2/(3-2/(3-2/(3-2/(3-2/(3-2/1)))))=1

3-2/(3-2/(3-2/(3-2/(3-2/(3-2/(3-2/1))))))=1

3-2/(3-2/(3-2/(3-2/(3-2/(3-2/(3-2/(3-2/1)))))))=1

You are presuming two things(which you never clarified in your post):

1) That this is an infinite sequence.

2) That the denominator is irrelevant and that we should ignore the order of operations.

The order of operations demands that multiplication and division should be done first.... not subtraction.

What is NOT justified is presuming that a0=3.

If it was defined as such, you should have clarified that in the question.

Miscommunication on your part..... presuming on your part.

Charles Allison - 5 years, 3 months ago

Log in to reply

@Charles Allison I read your entire comment. I said that you had "not the standardized process in which continued fractions are calculated". I should have been more explicit and clarified that the way that the continued fraction

3 2 3 2 3 2 3 3 - \dfrac{2}{ 3 - \dfrac{2} { 3 - \dfrac{2} { 3 - \ddots } } }

is evaluated, is as the limit of the sequence

3 , 3 2 3 , 3 2 3 2 3 , 3 2 3 2 3 2 3 3, 3 - \dfrac{2}{3}, 3 - \dfrac{2}{ 3 - \dfrac{2} { 3} } , 3 - \dfrac{2}{ 3 - \dfrac{2} { 3 - \dfrac{2} { 3 }}} \ldots

This removes all ambiguity about "what you are giving weight to".

So yes, I am presuming that you understand how continued fractions are evaluated, in order to answer this problem properly. In a similar manner, not everyone knows how to integrate / differentiate, and so if they want to answer a calculus problem properly, they will have to learn about it. That's the point of this problem, in indicate various misconceptions that people might have about continued fractions, especially those who do not know the mathematically correct way of dealing with it.

(Note: In particular, the approach that @Yasir Soltani used in his solution is incorrect. As mentioned, it involves an arbitrary choice to select 2 instead of 1.)

Calvin Lin Staff - 5 years, 3 months ago

Log in to reply

@Calvin Lin Yeah, i didn't give it much thought, it looked like it was approaching 2! :(

Yasir Soltani - 5 years, 3 months ago

@Calvin Lin I don't think his select of 2 instead of 1 or 3 was arbitrary at all...

He selected that because he presumed a different starting point for the iterative deduction that he made.

Whereas you are presuming that the most nested denominator will be "3", he presumed it will be "2" (because of the nested numerator) and I am inducing that the nested denominator would be 1 because of the previous sequence that you provided.

You are saying that we should start from a0 and move towards an infinite point...

I am saying that we should presume the infinitely large point's denominator and do an iterative deduction backwards.

He is saying that we should presume the infinitely large point's nominator and do an iterative deduction backwards.

Charles Allison - 5 years, 3 months ago

Log in to reply

@Charles Allison I am saying that is the convention. You may insist on not using the convention, in which case you may get a different result. I am not here to argue about "Is the convention right?".

Similarly, if we do not use the same convention for words, then we can reach different conclusions from the same sentence.

Calvin Lin Staff - 5 years, 3 months ago

"as calculated above a0=3 "

Actually... it was never "calculated" above.... it was defined as such in the answer.

Which is pretty lame since you didn't have the foresight to communicate point properly.

Charles Allison - 5 years, 3 months ago

Thanks for such a clear solution!

John Frank - 5 years, 4 months ago
Yasir Soltani
Feb 4, 2016

I did it the way Calvin told me to do i.e OBSERVE \textbf{OBSERVE}

Please explain in detail.

Calvin Lin Staff - 5 years, 4 months ago

I kept building it up like 3 2 2 3 - \dfrac{2}{2} 3 2 3 2 2 3 - \dfrac{2}{ 3 - \dfrac{2}{2}} 3 2 3 2 3 2 2 3 - \dfrac{2}{ 3 - \dfrac{2} { 3 - \dfrac{2} { 2} } } 3 2 3 2 3 2 3 3 - \dfrac{2}{ 3 - \dfrac{2} { 3 - \dfrac{2} { 3 - \ddots } } }

Yasir Soltani - 5 years, 4 months ago

Log in to reply

Problem is both fractions have the same exact progression as a continued fraction BUT different values ...

Sam Lee - 5 years, 4 months ago

Why did you choose to use 2 2 \frac{2}{2} instead of 2 1 \frac{2}{1} ? That choice seems almost arbitrary.

Calvin Lin Staff - 5 years, 4 months ago

Log in to reply

Mathematics is ideology.

No matter how a person answers this, it is wrong because there are not enough premises given to make a reasonable deduction. The poster never said whether the sequence is a finite or infinite sequence.

In any finite sequence, the answer will be 1. In an infinite sequence, the answer will approach 2.

The answer you give also depends heavily on the order of operations in which you calculate. If you move from left to right.. you might see that it is 3-2/3... but the "2/3" part is actually approaching 2/2(ie 1)... so the answer would be approaching 3-1=2 in that sequence.

But in any finite/rational sequence, the order of operations would dictate that the go out from the most nested point in computation.... so the answer will be 1 in a finite sequence.

And so, the answer you give depends on what you are giving weight to... Are you giving priority/weight to finitude and the order of operations (PEMDAS)? Or are you giving priority/weight to the concept of infinite regress and ignoring the order of operations?

It gets more even complex if you don't know what the denominator will be in a finite sequence. If the denominator is very small and close to 0, you will end up subtracting a large number, and end up with a sequence that goes from positive to negative constantly. 3-2/.00000000000000000000000000000000000000000000000(infinite zeros)00001= -infinitely large number. but 3-2/(-infinitely large number)=infinitely large positive number...

Depending on which mathematical principle you give priority to and what you presume is the denominator part of that sequence and if you presume that the sequence is finite or infinite your answer will be radically different.

Charles Allison - 5 years, 3 months ago
Ramiel To-ong
Feb 3, 2016

QUADRATIC FORMULA.

May you please elaborate? Thank you!

John Frank - 5 years, 4 months ago

Please explain in detail.

Calvin Lin Staff - 5 years, 4 months ago

Log in to reply

Sir can you please explain why 'does not exist' cant be the answer as our x tends to give two possible values ( as in case of limits) also i read people saying that since the sequence is infinite answer cant be 1 but its not necessary that infinite sequences cant give 1 as final result.

Somesh Patil - 5 years, 4 months ago

Log in to reply

Review how to calculate the limits of sequences .

In particular, when does the limit not exist?

Calvin Lin Staff - 5 years, 3 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...