The Student Room Group

How do you prove that this is divergent?

My lecturer has challenged me to complete this question. I need to prove the series either diverges or converges. I know that the series is divergent by using divergent tests, but I need to do this by the comparison test. The problem is the -3 in the numerator and the only hint given is that I need to divide the numerator by something to make n^4 smaller.

Question.PNG

Anybody know how to do this?
(edited 4 years ago)
Original post by Brad.Allison
My lecturer has challenged me to complete this question. I need to prove the series either diverges or converges. I know that the series is divergent by using divergent tests, but I need to do this by the comparison test. The problem is the -3 in the numerator and the only hint given is that I need to divide the numerator by something to make n^4 smaller.

Anybody know how to do this?


So you just need a lower bound on (n43n6+n1)1/2\displaystyle \left( \dfrac{n^4-3}{n^6 + n - 1} \right)^{1/2} whose series diverges.

Just divide numerator and denominator by n4n^4 to obtain


(13n4n2+n3n4)1/2\displaystyle \left( \dfrac{1-3n^{-4}}{n^2 + n^{-3} - n^{-4}} \right)^{1/2}

Note that 13n4c11-3n^{-4} \geq c_1 for n2n \geq 2

and n2+n3n4<c2n2n^2 + n^{-3} - n^{-4} < c_2 n^2

where I leave it to you to determine what c1,c2c_1,c_2 are as they are constants.


Hence bound is n=2(n43n6+n1)1/2>c1c2n=21n\displaystyle \sum_{n=2}^{\infty} \left( \dfrac{n^4-3}{n^6 + n - 1} \right)^{1/2} > \sqrt{ \dfrac{c_1}{c_2} } \sum_{n=2}^{\infty} \dfrac{1}{n}

and clearly the lower bound diverges.
Original post by RDKGames
So you just need a lower bound on (n43n6+n1)1/2\displaystyle \left( \dfrac{n^4-3}{n^6 + n - 1} \right)^{1/2} whose series diverges.

Just divide numerator and denominator by n4n^4 to obtain


(13n4n2+n3n4)1/2\displaystyle \left( \dfrac{1-3n^{-4}}{n^2 + n^{-3} - n^{-4}} \right)^{1/2}

Note that 13n4c11-3n^{-4} \geq c_1 for n2n \geq 2

and n2+n3n4<c2n2n^2 + n^{-3} - n^{-4} < c_2 n^2

where I leave it to you to determine what c1,c2c_1,c_2 are as they are constants.


Hence bound is n=2(n43n6+n1)1/2>c1c2n=21n\displaystyle \sum_{n=2}^{\infty} \left( \dfrac{n^4-3}{n^6 + n - 1} \right)^{1/2} > \sqrt{ \dfrac{c_1}{c_2} } \sum_{n=2}^{\infty} \dfrac{1}{n}

and clearly the lower bound diverges.

15756496046158499407729554653160.jpg
Thankyou for your reply! Based off what you said, would something like this work?
Original post by Brad.Allison

Thankyou for your reply! Based off what you said, would something like this work?


You are working backward from what I said even though I intended for you to go *from* the LHS *to* the RHS.

So 13n41324=1316=c11-3n^{-4} \geq 1-3 \cdot 2^{-4} = \dfrac{13}{16} = c_1

and similarly proceed to show that

n2+n3n4n^2 + n^{-3} - n^{-4} has an upper bound of c2n2c_2 n^2 for some c2c_2.

P.S. And no, the bottom half of your page isn't valid so don't bother substituting in n=2 for this one.
(edited 4 years ago)
Original post by RDKGames
You are working backward from what I said even though I intended for you to go *from* the LHS *to* the RHS.

So 13n41324=1316=c11-3n^{-4} \geq 1-3 \cdot 2^{-4} = \dfrac{13}{16} = c_1

and similarly proceed to show that

n2+n3n4n^2 + n^{-3} - n^{-4} has an upper bound of c2n2c_2 n^2 for some c2c_2.

P.S. And no, the bottom half of your page isn't valid so don't bother substituting in n=2 for this one.

Oh okay, so for c2 something like this.
15756506692366835792679216622029.jpg
Original post by Brad.Allison
Oh okay, so for c2 something like this


That's not a convincing argument and not what I was suggesting. While that value of c2c_2 works, you need to refine your argument.


For all n2n \geq 2 we have that

1<n2+n3n4<n2+n3<n2+kn21 < n^2 + n^{-3} - n^{-4} < n^2 + n^{-3} < n^2 + kn^2

and really, you could just pick k=1k = 1 at this stage (because it's obvious that n3<n2n^{-3} < n^2) hence obtain c2=2 c_2 = 2.

We are effectively bounding n3<kn2n^{-3} < kn^2 so it's just a matter of having k>n5k > n^{-5} for all n2n \geq 2. So just choose k>maxn2n5=25k > \displaystyle \max_{n \geq 2} n^{-5} = 2^{-5} hence the bound n2+n3<c2n2n^2 + n^{-3} < c_2 n^2 is guaranteed if we just choose

c2=1+k>3332c_2 = 1+k > \dfrac{33}{32}.

Your answer you got through a bad argument works because 654>3332\dfrac{65}{4} > \dfrac{33}{32}.
(edited 4 years ago)
Original post by RDKGames
That's not a convincing argument and not what I was suggesting. While that value of c2c_2 works, you need to refine your argument.


For all n2n \geq 2 we have that

1<n2+n3n4<n2+n3<n2+kn21 < n^2 + n^{-3} - n^{-4} < n^2 + n^{-3} < n^2 + kn^2

and really, you could just pick k=1k = 1 at this stage (because it's obvious that n3<n2n^{-3} < n^2) hence obtain c2=2 c_2 = 2.

We are effectively bounding n3<kn2n^{-3} < kn^2 so it's just a matter of having k>n5k > n^{-5} for all n2n \geq 2. So just choose k>maxn2n5=25k > \displaystyle \max_{n \geq 2} n^{-5} = 2^{-5} hence the bound n2+n3<c2n2n^2 + n^{-3} < c_2 n^2 is guaranteed if we just choose

c2=1+k>3332c_2 = 1+k > \dfrac{33}{32}.

Your answer you got through a bad argument works because 654>3332\dfrac{65}{4} > \dfrac{33}{32}.

Hi, sorry I'm struggling to understand this as it's not usually how we complete the questions. If you have time, could you try and explain this is as much detail as possible?
Original post by Brad.Allison
Hi, sorry I'm struggling to understand this as it's not usually how we complete the questions. If you have time, could you try and explain this is as much detail as possible?


In order to find the lower bound for (13n4n2+n3n4)1/2\left( \dfrac{1-3n^{-4}}{n^2 + n^{-3} - n^{-4}} \right)^{1/2} you need to do two things;

(a) find a lower bound for 13n41-3n^{-4}

(b) find an upper bound for n2+n3n4n^2 + n^{-3} - n^{-4}.


For (a), it is obvious that 13n41-3n^{-4} is an increasing function for n2n \geq 2 therefore we are justified to sub in n=2n=2 to obtain the lower bound. Hence 13n413161-3n^{-4} \geq \dfrac{13}{16}.

For (b), it is less so obvious as to how this function behaves. We cannot immediately say that this is a decreasing function, so we cannot substitute anything in. But we can notice by inspection that n4n^{-4} is being taken away, therefore n2+n3n^2 + n^{-3} is a bigger number.
Hence n2+n3n4<n2+n3n^2 + n^{-3} - n^{-4} < n^2 + n^{-3}.
Now also notice that for n2n \geq 2 we have that n3n^{-3} is a very small number. We are adding it onto n2n^2. But we add on n2n^2 instead, then this is obviously going to give as a bigger result because n2n^2 is much bigger than n4n^{-4}. Therefore;
n2+n3n4<n2+n3<n2+n2=2n2n^2 + n^{-3} - n^{-4} < n^2 + n^{-3} < n^2 + n^2 = 2n^2.

Hence the upper bound here is 2n22n^2. And so the overall bound becomes; (13n4n2+n3n4)1/2>(1332n2)1/2=13321n\left( \dfrac{1-3n^{-4}}{n^2 + n^{-3} - n^{-4}} \right)^{1/2} > \left( \dfrac{13}{32n^2} \right)^{1/2} = \sqrt{\dfrac{13}{32}} \cdot \dfrac{1}{n}

and its series is a divergent one, hence by comparison test n2(n43n6+n1)1/2\displaystyle \sum_{n \geq 2} \left( \dfrac{n^4-3}{n^6 + n - 1} \right)^{1/2} diverges.




I merely introduced kk to show you that we can bound n3n^{-3} by *not just* n2n^2 but by any kn2kn^2 as long as k>132k > \dfrac{1}{32}, and since your method was incorrect, it still obtained a valid coefficient because this inequality was satisfied by it.
(edited 4 years ago)
Original post by RDKGames
In order to find the lower bound for (13n4n2+n3n4)1/2\left( \dfrac{1-3n^{-4}}{n^2 + n^{-3} - n^{-4}} \right)^{1/2} you need to do two things;

(a) find a lower bound for 13n41-3n^{-4}

(b) find an upper bound for n2+n3n4n^2 + n^{-3} - n^{-4}.


For (a), it is obvious that 13n41-3n^{-4} is an increasing function for n2n \geq 2 therefore we are justified to sub in n=2n=2 to obtain the lower bound. Hence 13n413161-3n^{-4} \geq \dfrac{13}{16}.

For (b), it is less so obvious as to how this function behaves. We cannot immediately say that this is a decreasing function, so we cannot substitute anything in. But we can notice by inspection that n4n^{-4} is being taken away, therefore n2+n3n^2 + n^{-3} is a bigger number.
Hence n2+n3n4<n2+n3n^2 + n^{-3} - n^{-4} < n^2 + n^{-3}.
Now also notice that for n2n \geq 2 we have that n3n^{-3} is a very small number. We are adding it onto n2n^2. But we add on n2n^2 instead, then this is obviously going to give as a bigger result because n2n^2 is much bigger than n4n^{-4}. Therefore;
n2+n3n4<n2+n3<n2+n2=2n2n^2 + n^{-3} - n^{-4} < n^2 + n^{-3} < n^2 + n^2 = 2n^2.

Hence the upper bound here is 2n22n^2. And so the overall bound becomes; (13n4n2+n3n4)1/2>(1332n2)1/2=13321n\left( \dfrac{1-3n^{-4}}{n^2 + n^{-3} - n^{-4}} \right)^{1/2} > \left( \dfrac{13}{32n^2} \right)^{1/2} = \sqrt{\dfrac{13}{32}} \cdot \dfrac{1}{n}

and its series is a divergent one, hence by comparison test n2(n43n6+n1)1/2\displaystyle \sum_{n \geq 2} \left( \dfrac{n^4-3}{n^6 + n - 1} \right)^{1/2} diverges.




I merely introduced kk to show you that we can bound n3n^{-3} by *not just* n2n^2 but by any kn2kn^2 as long as k>132k > \dfrac{1}{32}, and since your method was incorrect, it still obtained a valid coefficient because this inequality was satisfied by it.

Thank you for taking the time to explain this. That made perfect sense!

Quick Reply

Latest