Just double the other

Algebra Level 1

A = 1 + 1 1 + 1 1 + 1 and B = 2 + 1 2 + 1 2 + 1 \large A = 1 + \frac{1}{ 1 + \frac{ 1}{ 1 + \frac{ 1} { \ddots } } } \qquad \text{and}\qquad B = 2 + \frac{1}{ 2 + \frac{ 1}{ 2 + \frac{ 1} { \ddots } } }

Given the two infinite continued fractions A A and B B above, which is larger, 2 A 2A or B ? B?

2 A 2A B B They are equal

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.

9 solutions

Chris Lewis
Dec 7, 2018

We can get some loose bounds by truncating the fractions.

A = 1 + 1 1 + [ 1 1 + 1 ] \begin{aligned} A = 1 + \frac{1}{ 1 + \left [ {\frac{ 1}{ 1 + \frac{ 1} { \ddots } } } \right ] } \end{aligned}

The quantity in the square brackets is clearly smaller than 1 1 (we have 1 1 divided by something larger than 1 1 ); it follows that A > 1 + 1 1 + 1 = 3 2 A>1+\frac{1}{1+1}=\frac{3}{2}

Similarly, we have

B = 2 + 1 2 + [ 1 2 + 1 ] \begin{aligned} B = 2 + \frac{1}{ 2 + \left [ {\frac{ 1}{ 2 + \frac{ 1} { \ddots } } } \right ] } \end{aligned}

Here, we use the fact that the bracketed quantity is greater than zero to get B < 2 + 1 2 = 5 2 B<2+\frac{1}{2} = \frac{5}{2}

Hence 2 A > B \boxed{2A>B}

Very nicely done!

This approach echoes a standard way of getting good rational approximations of the original value, and we can take upper and lower bounds as shown. Thus, if we wanted to compare 2 continued fractions, we could take sufficiently many rational approximations till we can distinguish between them.

Note: More commonly, the upper bound truncates after an odd number of terms like in B, and the lower bound truncates after an even number of terms, like in A via 1 + 1 1 + 1 1 + [ ] . 1 + \frac{1}{ 1 + \frac{ 1}{ 1 + [ ] } } . This follows from the reasoning laid out by Chris.

Calvin Lin Staff - 2 years, 6 months ago

Log in to reply

Hi Calvin. I have paid for a premium account on brilliant but the activation link is not working. I wrote to the support staff but they seem to be out of office for 48 hours. Since you are a staff member can you please help me. The membership was gifted by me to myself.

Apoorva Singal - 1 year, 7 months ago

Log in to reply

Hi Apoorva,

I see that you bought the gift subscription and gifted it to yourself. We will look into why you're having issues. In the meantime, I've given your account a 7-day free trial (and will extend it as needed), so you should have access to Brilliant premium.

I see your email in our system. I'm sorry that our support staff hasn't gotten back to you as yet. Our support staff doesn't work on the weekends and prioritizes what to respond to on Monday. (It's Tuesday morning 8am right now).

Calvin Lin Staff - 1 year, 7 months ago

How is " \ddots " necessarily positive?

There must be a formal definition of " \ddots ": for a sequence x n + 1 = 1 + 1 x n x_{n+1}=1+\frac{1}{x_n} , A = 1 + 1 1 + 1 1 + 1 = lim n x n A=1 + \frac{1}{ 1 + {\frac{ 1}{ 1 + \frac{ 1} { \ddots } } } }=\lim_{n\to\infty}{x_n} . Note that here the "initial value" x 1 x_1 is arbitrary. It can be any positive, negative, or even complex number, even if it is not equal to A A .

It is then easy to prove (I won't put the proof here) that for all x 1 x_1 complex, either x n 1 + 5 2 x_n \to \frac{1+\sqrt5}{2} or x n 1 5 2 x_n \to \frac{1-\sqrt5}{2} , for n n \to \infty . So A = 1 ± 5 2 3 2 A=\frac{1\pm\sqrt5}{2}\ngtr\frac32 .

In this case this proof is incorrect since 1 + 1 1+\frac{1}{\ddots} is not necessarily greater than 1 1 . The same holds true for B B .

icreiuhe Zhu - 2 years, 5 months ago

Log in to reply

How nice. It approaches the golden ratio.

Tim Popely - 2 years, 3 months ago

Both infinite continued fractions above can be represented by the recursive sequences:

Let x 1 = 1 x_{1} = 1 and x n + 1 = 1 + 1 x n x_{n + 1} = 1 + \dfrac{1}{x_{n}}

and

y 1 = 2 y_{1} = 2 and y n + 1 = 2 + 1 y n y_{n + 1} = 2 + \dfrac{1}{y_{n}} .

You can show both sequences x n x_{n} and y n y_{n} are convergent.

Assuming convergence A = lim n x n = lim n x n + 1 = 1 + 1 A A 2 A 1 = 0 A = 1 + 5 2 A = \lim_{n \rightarrow \infty} x_{n} = \lim_{n \rightarrow \infty} x_{n + 1} = 1 + \dfrac{1}{A} \implies A^2 - A - 1 = 0 \implies A = \dfrac{1 + \sqrt{5}}{2}

Similarly, B = lim n y n = lim n y n + 1 = 2 + 1 B B 2 2 B 1 = 0 B = 1 + 2 B = \lim_{n \rightarrow \infty} y_{n} = \lim_{n \rightarrow \infty} y_{n + 1} = 2 + \dfrac{1}{B} \implies B^2 - 2B - 1 = 0 \implies B = 1 +\sqrt{2}

2 A = 1 + 5 > 1 + 2 = B \implies 2A = 1 + \sqrt{5} > 1 + \sqrt{2} = B

Rocco Dalto - 2 years, 5 months ago

2A = 2 + 2 divided by “stuff”. B = 2 + 1 divided by stuff. The stuff in A is smaller than the stuff in B...therefore A is larger.

Jason Pound - 2 years, 5 months ago

I have another doubt... If we just substitute the value of A in it self we get: A = 1+(1÷A)

We can simplify it into a quadratic equation and interestingly we get two solutions!: (1+√5)÷2 and (1-√5)÷2..... The first solution solves the question but the second one is negative. Why we won't consider it?

Anirudh Sharma - 2 years, 4 months ago

Log in to reply

by definition, if you add a positive number to a positive number, the sum would be positive

Tri Bui - 2 years, 1 month ago

x=n+1/x x=(n+√(n²+4))/2

Stream Ho - 2 years, 3 months ago

We can solve this by using method of convergence. The first term converges as 1 + (1/1.5) which equals 1.67. Second term converges to 2 + (1/2) which equals 2.5. Thus 2* 1.67 greater than 2.5.

canver m varghese - 1 year, 9 months ago
Dennis Rodman
Dec 17, 2018

This needs to be under "Basic."

The following are readily observable from inspecting the first line of the continued

fraction for A and the second line of the continued fraction for B:

A > 1.5

Then, 2A > 3

But, B < 2 + 1 = 3

Therefore, 2A is larger.

Great comparison. We just need the first denominator (similar to Chris' solution).

Calvin Lin Staff - 2 years, 5 months ago
Parth Sankhe
Dec 5, 2018

A = 1 + 1 A A=1+\frac {1}{A}

Do the same for B and compare.

A = 0.5 ( 1 + 5 ) A=0.5(1+√5) and B = 1 + 2 B=1+√2

Yep easiest method imo

Nikola Georgiev - 2 years, 5 months ago

Actually, 2A = 1±√5 while B = 1±√2. Shouldn't we consider the negative values?

Sumant Chopde - 2 years, 5 months ago

Log in to reply

Very good question, I'd like to know this too

Peter Pepper - 2 years, 5 months ago

No, since continued fractions are given by recurrence relations and we interpret A and B as limits of those sequences. Negative values are excluded since all members of the sequences are positive, and thus the limits are nonnegative.

Denis Husadzic - 2 years, 5 months ago
Calvin Lin Staff
Dec 4, 2018

[This is not a complete solution.]

For those who know how to solve the continued fraction, we can show that A = 1 + 5 2 A = \frac{1 + \sqrt{5} } {2} and B = 1 + 2 B = 1 + \sqrt{2} .
From here, it directly follows that 2 A > B 2A > B .

How can we show this directly by just comparing fractions?
Cryptic Hint: What does 2 A 2A look like in the above format?


(More details in comments) Cryptic hint elaborated on:
If you naively tried to find 2 A 2A , you will say that it looks like

2 A = 2 + 1 1 2 + 1 2 + 1 2A = 2 + \frac{1}{ \frac{1}{2} + \frac{ 1}{ 2 + \frac{ 1} { \ddots } } }

However, this isn't a continued fraction, which requires that all of the terms in the fraction are positive integers. Furthermore, this doesn't quickly allow for an easy comparison with B B .

As such, this isn't easy to work with. What's the next step in trying to work out this problem?

@Calvin Lin Sir, is 2A = [2;1/2,2,1/2,2,1/2........] ????

Aaghaz Mahajan - 2 years, 6 months ago

Log in to reply

That's not a continued fraction, because the coefficients should be integers. Under these conditions, we get a unique representation, which then makes it easy to compare the values of continued fractions.

Note about unique representation of continued fractions: For rationals, we require that the final fractional part not be 1, because that can be combined with the previous term.


In fact, if we tried to directly express 2 A = 1 + 5 2A = 1 + \sqrt{5} as a continued fraction, we will get

2 A = 3 + 1 4 + 1 4 + 1 2A = 3 + \frac{1}{ 4 + \frac{ 1}{ 4 + \frac{ 1} { \ddots } } }

From here, it is obvious that 2 A > 3 > B 2A > 3 > B . However, getting to this form directly from the continued fraction seems impossible.

Alright, I'd admit that my hint isn't very direct, so I made it a "Cryptic hint". The point is for you to recognize that "2A isn't as easy to work with". So, what's the next step in trying to work out this problem?

Calvin Lin Staff - 2 years, 6 months ago

Log in to reply

Yeah Sir.........after replying to you, I figured out the continued fraction of 2A in the form which you have stated.........But I got there only by evaluating the value of A and then, forming the continued fraction of 2A..............Again, it is just a circular argument......I don't really get it that how multiplying by a number can help us in finding the continued fraction............

Aaghaz Mahajan - 2 years, 6 months ago

Log in to reply

@Aaghaz Mahajan Right, and I am agreeing with you that "getting to this form directly from the continued fraction seems impossible".

So, the important question now is "If 2 A 2A is hard to deal with, what else could we try?"

Meta: In problem solving, you're testing out different paths and trying to pick the most probable. Sometimes, when we explore a path, we come to a (presumed) dead end. If so, we can either figure out how to bash the dead end, or try another path.

Calvin Lin Staff - 2 years, 6 months ago

Wait, aren’t the solutions for both equation also include the value where A=(1-sqrt(5))/2 and B=1-sqrt(2) ? How can we be sure that A and B will take the positive value? *sqrt is square root, Sorry, typing this in handphone without math symbols

ahnaf amiral - 2 years, 5 months ago

Log in to reply

That's a good point. We know that the solution has to satisfy A = 1 + 1 A A = 1 + \frac{1}{A} , but there are 2 solutions to the quadratic equations.

So, there are 2 questions we have to answer:

1) How do we even know that one of the solutions is valid?

2) How do we know which solution is valid?

Hint: A continued fraction is defined by the limit of the truncated fractions.

Calvin Lin Staff - 2 years, 5 months ago

A is exactly the golden ratio!

Jason Horner - 2 years, 5 months ago
Amit Bansal
Dec 19, 2018

We can get these equations A=1+ 1 A \frac{1}{A} and B=2+ 1 B \frac{1}{B} by substituting the infinite fraction as it is a same repeating pattern. A 2 A 1 = 0 A^2-A-1=0 and B 2 2 B 1 = 0 B^2-2B-1=0 So the roots are A = ( 1 + 5 ) / 2 A=(1+\sqrt 5)/2 and B = 1 + 2 B=1+\sqrt 2 . Hence 2A>B ( A and B are greater than 0 hence we cannot have A = ( 1 5 ) / 2 A=(1-\sqrt 5)/2 and B = 1 2 B=1-\sqrt 2 )

Actually, 2A = 1±√5 while B = 1±√2. Shouldn't we consider the negative values?

Sumant Chopde - 2 years, 5 months ago

Log in to reply

We have to dismiss the -√2 and -√5 value, because in this case it woudn't be equal to the A and B they are giving us, because A is 1+ (something superior to 0) and B is 2+(something superior to 0), so we remove the 2 solutions that don't obey to this

Mouad Bachraoui - 2 years, 5 months ago
David Bennett
Dec 18, 2018

Let A = 1 + R , A = 1 + R, where R R is the continued fraction. Then,

R = 1 / ( 1 + R ) R = 1/(1+R) (the continued fraction equals its subfraction.)

R + R 2 = 1 R + R^2 = 1

R 2 + R 1 = 0 R^2 + R - 1 = 0

apply quadratic formula:

R = 1 ± s q r t ( 5 ) 2 R = \frac{-1 \pm sqrt (5)}{2}

R = 0.61803 R = 0.61803\dots (choosing the positive solution)

Let B = 2 + S , B = 2 + S, where S S is the continued fraction. Then,

S = 1 / ( 2 + S ) S = 1/(2+S)

2 S + S 2 = 1 2S + S^2 = 1

S 2 + 2 S 1 = 0 S^2 + 2S - 1 = 0

apply quadratic formula:

S = 2 ± s q r t ( 8 ) 2 S = \frac{-2 \pm sqrt(8)}{2}

S = 0.41421... S = 0.41421... (choosing the positive solution)

Then A = 1.61803 , A = 1.61803\dots, B = 2.41421 , B = 2.41421\dots, 2 A = 3.2361 2A = 3.2361\dots and 2 A > B 2A > B .

Actually, 2A = 1±√5 while B = 1±√2. Shouldn't we consider the negative values?

Sumant Chopde - 2 years, 5 months ago

Log in to reply

We could consider the negative values, as they are solutions for the infinite continued fraction (albeit unintuitive ones,) but the problem implies the answer to be unique, therefore both are positive.

David Bennett - 2 years, 5 months ago
J.S. Eenhoorn
Dec 19, 2018

A = 1 + 5 2 A = \frac{1+\sqrt{5}}{2} and B = 1 + 3 B = 1+\sqrt{3} . Then 2 A = 1 + 5 2A = 1+\sqrt{5} and B = 1 + 3 B = 1 + \sqrt{3} so 2 A > B 2A > B .

Continued fractions have no mathematical meaning, they are just a formal expression without a defined value.

So I find it annoying when they are thrown around like they were some well established and meaningful mathematical object.

Most people interpret them as being the limit of the sequence A n + 1 = 1 + 1 A n A_{n+1} = 1+\frac{1}{A_n} but this is totally different from the original continued fraction.

And even if we interpret them like that, it all depends on where we start from at A 0 A_0 . Surely enough there is only one stable fixed point, but sometimes it is impossible to continue the succession after a while (because we get A m = 0 A_m=0 ) and it is also possible that we start at the non stable fixed point and never go away from that.

The negative solution would still be a totally acceptable value for the limit of the sequence and hence as the value of the continued fraction: who said a continued fraction should be positive? Wasn't Ramanujan himself who suggested 1 12 -\frac{1}{12} as the sum of the harmonic series?

Nicola Gabbia - 2 years, 3 months ago
Pi Han Goh
Dec 9, 2018

Method 1: This question is essentially asking us to prove that A B > 1 2 \dfrac AB > \dfrac12 . I will prove a stronger bound, A B > 2 3 \dfrac AB > \dfrac23 .

Using the same idea as Chris Lewis' solution, we have

A > 1 + 1 1 + 1 = 3 2 A = 1 + 1 A < 5 3 A = 1 + 1 A > 8 5 A = 1 + 1 A < 13 8 A = 1 + 1 A > 21 13 . \begin{aligned} A\quad\quad\quad\quad &>& 1 + \dfrac1{1+1} = \dfrac32 \\ A = 1 + \dfrac1A &<& \dfrac53 \\ A = 1 + \dfrac1A &>& \dfrac85 \\ A = 1 + \dfrac1A &<& \dfrac{13}8 \\ A = 1 + \dfrac1A &>& \dfrac{21}{13}. \end{aligned}

Similarly, B < 2 + 1 2 = 5 2 B = 2 + 1 B > 12 5 B = 2 + 1 B < 29 12 . \begin{aligned} B \quad\quad\quad\quad &<& 2 + \dfrac1{2} = \dfrac52 \\ B = 2 + \dfrac1B &> & \dfrac{12}5 \\ B = 2 + \dfrac1B &<& \dfrac{29}{12}. \end{aligned}

Because A > 21 13 > 29 18 6 A > 29 3 , A> \dfrac{21}{13} > \dfrac{29}{18} \Rightarrow 6A > \dfrac{29}3 , and 4 B < 29 3 4B < \dfrac{29}3 , putting it all together gives 6 A > 29 3 > 4 B 6 A > 4 B A B > 2 3 . 6A > \dfrac{29}3 > 4B \qquad\Rightarrow \qquad 6A > 4B \qquad\Rightarrow \qquad \dfrac AB > \dfrac23.

Since A B > 2 3 > 1 2 \dfrac AB > \dfrac23 > \dfrac12 . It follows that 2 A > B 2A > B . Hence, the answer is 2 A \boxed{2A} .


Method 2: We want to prove that 2 A > B 2A > B , which is equivalent to proving that 10 A 5 B > 0 ( 10 A 2 ) ( 5 B 2 ) > 0. 10A - 5B > 0 \qquad \Leftrightarrow \qquad (10A - 2) - (5B - 2) > 0 .

It's obvious that 1 < A < 2 1 < A < 2 and 2 < B < 3 2 < B < 3 , so A < B 1 A > 1 B 1 A 1 B > 0 A < B \Rightarrow \frac 1A > \frac 1B \Rightarrow \frac1A - \frac1B > 0 .

With 10 A 2 = 10 ( 1 + 1 1 + 1 1 + 1 ) 2 = 8 + 10 1 + 1 1 + 1 , 10A - 2 = 10 \left( 1 + \frac{1}{ 1 + \frac{ 1}{ 1 + \frac{ 1} { \ddots } } } \right) - 2 = 8 + \frac{10}{ 1 + \frac{ 1}{ 1 + \frac{ 1} { \ddots } } }, and 5 B 2 = 5 ( 2 + 1 2 + 1 2 + 1 ) 2 = 8 + 5 2 + 1 2 + 1 . 5B - 2 = 5 \left( 2 + \frac{1}{ 2 + \frac{ 1}{ 2 + \frac{ 1} { \ddots } } } \right) - 2 = 8 + \frac{5}{ 2 + \frac{ 1}{ 2 + \frac{ 1} { \ddots } } } .

Putting them together gives ( 10 A 2 ) ( 5 B 2 ) = 5 ( 1 1 + 1 1 + 1 1 2 + 1 2 + 1 ) + 5 ( 1 + 1 1 + 1 1 + 1 ) = 5 ( 1 A 1 B ) > 0 + 5 A > 0. (10A - 2) - (5B - 2) = 5\left( \frac{1}{ 1 + \frac{ 1}{ 1 + \frac{ 1} { \ddots } } } - \frac{1}{ 2 + \frac{ 1}{ 2 + \frac{ 1} { \ddots } } } \right) + 5 \left( 1 + \frac{1}{ 1 + \frac{ 1}{ 1 + \frac{ 1} { \ddots } } } \right) = 5\underbrace{\left( \frac1A - \frac1B\right)}_{> \; 0} + 5A > 0 .

And we're done!

Are you sure about the equations in method 2? From A = 1 + 1 A A = 1 + \frac{1}{A} and B = 2 + 1 B B = 2 + \frac{1}{B} , what we have is 10 A 5 B = 10 A 5 B 10A - 5 B = \frac{10}{A} - \frac{5}{B} . Note: I'm not sure why you multiplied by 5, or why you chose to consider 10 A 2 10A - 2 . Doesn't the integer part just cancel out?

You could just do it directly as 2 A = 2 + 2 A > 2 + 1 B = B 2A = 2 + \frac{2}{A} > 2 + \frac{1}{B} = B .

Calvin Lin Staff - 2 years, 6 months ago

Log in to reply

Are you sure about the equations in method 2?

Yup, it's a long-winded approach. But it works!

Note: I'm not sure why you multiplied by 5, or why you chose to consider 10 A 2 10A - 2 . Doesn't the integer part just cancel out?

Well, looking at your earlier comment, I realized that my working is way longer than necessary. I started out with A 2 B > 0.8218 4 5 A - 2B > 0.8218 \approx \frac45 , so I multiplied by 5 to remove all the fractions. And next thing you know, I got the expression ( 10 A 2 ) ( 5 B 2 ) (10A - 2) - (5B - 2) .

You could just do it directly as 2 A = 2 + 2 A > 2 + 1 B = B 2A=2+\frac2A>2+\frac1B=B .

Yup. I just realized that after posting my solution. Oh well.

Pi Han Goh - 2 years, 5 months ago

Log in to reply

Note: Your final equation is wrong. Specifically, ( 10 A 2 ) ( 5 B 2 ) 5 ( 1 A 1 B (10A - 2) - (5B - 2) \neq 5 ( \frac{1}{A} - \frac{1}{B} + 5A ). The LHS is equal to 10 A 5 B \frac{10}{A} - \frac{5}{B} , which follows if we make the (correct) assumption that A = 1 + 1 A A = 1 + \frac{1}{A} and B = 2 + 1 B B = 2 + \frac{1}{B} .

I have no idea how you got to the second part of the equation (esp without deriving a relationship about 1 A = \frac{1}{A} = \ddots . It also seems like you had an additional 5 in there.

Calvin Lin Staff - 2 years, 5 months ago
Affan Izzadeen
Dec 20, 2018

I just pretended that the fractions weren't infinite so a was 1+1/1 and b was 2+1/2. If you multiplied a by 2 it would be more. I'm not sure if this is how you're supposed to do it though.

Not quite, because you need to account for the rest. For example, if A A was close to 1 + 1 5 1 + \frac{1}{5} , then multiplying it by 2 would be less than 2 + 1 2 2 + \frac{1}{2} .

Calvin Lin Staff - 2 years, 5 months ago

The key for me was to learn I could scroll sideways to see the rest of the question.. I thought b= 2..

Courtenay Watson - 1 year, 2 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...