Roots of sum of reciprocals

Algebra Level 3

1 x 1 + 1 x 2 + 1 x 3 + 1 x 4 + 1 x 5 = 1 \frac1{x-1}+\frac1{x-2}+\frac1{x-3}+\frac1{x-4}+\frac1{x-5} = 1

What is the number of real roots to the equation above in the interval [ 1 , 5 ] [1,5] ?

2 3 4 1 5

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.

8 solutions

Garrett Clarke
Jun 14, 2015

Below I will generalize the problem for the number of solutions to k = 1 n 1 x k = 1 \sum_{k=1}^{n} \frac{1}{x-k} = 1 in the interval [ 1 , n ] [1,n] and set n = 5 n=5 .

First, we begin by subtracting 1 from both sides and defining the LHS as a function.

f ( x ) = [ k = 1 n 1 x k ] 1 = 0 f(x)=[\sum_{k=1}^{n} \frac{1}{x-k}]-1=0

It's clear that there are asymptotes at each integer value in the interval [ 1 , n ] [1,n] . Since we are looking for solutions in this interval, we need to find roots to the equation only in between each integer value inside the desired interval. Finding f ( x ) f'(x) yields our desires solutions.

f ( x ) = k = 1 n 1 ( x k ) 2 f'(x)=\sum_{k=1}^{n} -\frac{1}{(x-k)^2}

It's clear that every term is negative no matter what value you set x x to be, meaning that the slope of our function is always negative. Therefore, because we have asymptotes at k k and k + 1 k+1 , our function must assume all values from ( , ) (-\infty,\infty) , and each value of x x maps to a unique value of f ( x ) f(x) in the interval [ k , k + 1 ] [k,k+1] . That means that there must be a solution to f ( x ) = 0 f(x)=0 and this solution is unique in the interval [ k , k + 1 ] [k,k+1] . f ( x ) f(x) has n 1 n-1 such intervals, meaning that f ( x ) f(x) must have n 1 n-1 roots, and these intervals are all contained within the larger interval [ 1 , n ] [1,n] , so we're done.

We have shown that the equation k = 1 n 1 x k = 1 \sum_{k=1}^{n} \frac{1}{x-k} = 1 must have n 1 n-1 solutions inside the interval [ 1 , n ] [1,n] . Finally, letting n = 5 n=5 yields our final answer: n 1 = 5 1 = 4 n-1=5-1=\boxed{4}

Moderator note:

Thanks for proving the generalization. The other way to show that there is only four roots in the interval is to prove that there is also another root in interval [ 8 , 9 ] [8,9] .

I did it using graphs. I drew the 5 asymptotes and drew a rough sketch for a function like reciprocal function and drew the line y = 1 y=1 . Counted the no. of intersections on the graph.

Gautam Jha - 5 years, 12 months ago

Log in to reply

It's always a good thought to graph the function, and your solution works perfectly in this situation. However, I want to draw your attention to the equation x 3 + 3 x 2 + 3 x + 2 = 1 x^3+3x^2+3x+2=1 . This is one example where your solution wouldn't work, because while it intersects the line y=1 just once, it actually has three roots. Gotta prove the uniqueness of each crossing. Good idea here though!

Garrett Clarke - 5 years, 12 months ago

Log in to reply

Which 3 roots does it have? I don't see it

Limanan Nursalim - 5 years, 11 months ago

Log in to reply

@Limanan Nursalim Exactly, it's something that can't be seen simply by graphing the equation.

x 3 + 3 x 2 + 3 x + 2 = 1 x^3+3x^2+3x+2=1

x 3 + 3 x 2 + 3 x + 1 = 0 x^3+3x^2+3x+1=0

( x + 1 ) ( x + 1 ) ( x + 1 ) = 0 (x+1)(x+1)(x+1)=0

The root x = 1 x=-1 has a multiplicity of 3, meaning that there are technically three roots at that single crossing of the line y = 1 y=1 .

Garrett Clarke - 5 years, 11 months ago

I made a silly mistake by ignoring discontinuity & thinking strictly decreasing.

Btw nicely explained.

Ayush Verma - 5 years, 12 months ago

Log in to reply

Thank you!

Garrett Clarke - 5 years, 12 months ago
Lu Chee Ket
Oct 15, 2015

The equation becomes:

x^5 - 20 x^4 + 145 x^3 - 480 x^2 + 724 x - 394 = 0

x1 = 8.390541233048952

x2 = 4.540394425688128

x3 = 3.434736666495783

x4 = 2.356631854844214

x5 = 1.277695819922924

Substituted back to L.H.S. of original equation and found all give the R.H.S. of 1, since 8.390541233048952 is out of interval between 1 to 5, there remained with only 4 real roots as introduced above.

Do not be miss leaded by no radical solution and stop solving quintic. A progress and product should be there for things we can solve. Explain by Tao only for those we cannot see or express.

Lu Chee Ket - 5 years, 8 months ago
Anirban Karan
Jun 18, 2015

Let f ( x ) = ( n = 1 5 1 x n ) 1 f(x)=(\displaystyle\sum_{n=1}^{5}\frac{1}{x-n})-1 . Now take a small number ϵ > 0 \epsilon>0 . It is very easy to show that lim ϵ 0 f ( 1 + ϵ ) + \displaystyle\lim_{\epsilon \to 0} f(1+\epsilon) \to +\infty and lim ϵ 0 f ( 2 ϵ ) \displaystyle\lim_{\epsilon \to 0} f(2-\epsilon) \to -\infty . So there exists at least one real root of f(x) in [ 1 , 2 ] [1,2] . Now f ( x ) = n = 1 5 ( 1 x n ) 2 f'(x)=-\displaystyle\sum_{n=1}^{5}(\frac{1}{x-n})^2 and it is not zero for any real finite value of x as it is sum of inverse of squares. So there is no maximum or minimum of f(x). This means there is exactly one root of f(x) in [ 1 , 2 ] [1,2] . By same logic one can show there exists exactly one root in each of the intervals [ 2 , 3 ] , [ 3 , 4 ] , [ 4 , 5 ] [2,3],[3,4],[4,5] . So there are 4 roots of f(x) in [ 1 , 5 ] [1,5] .

  1. Write as xf = 1 (excuse for writing functions where they belong :) ), then xf - 1 = 0.
  2. Differentiate and we have f' < 0, so f always has a negative slope.
  3. Next evaluate xf-1 at zero - we get 0f -1 = -1 - 1/2 - 1/3 - 1/4 -1/5 being lazy I let DrRacket do the summation: > (+ -1 -1/2 -1/3 -1/4 -1/5 -1) -197/60
  4. So xf-1 starts life with x=0 at a bit less than -3, with poles at 1, 2, 3, 4 and 5.
  5. Each time xf-1 dives, it has to come up for air (otherwise how would it be able to dive again??) and we see that xf-1 = -infinity at 1-, 2-, 3- 4- and 5-, and +infinity at1+, 2+, 3+ and 4+ and is continuous monotonic strictly decreasing over the ranges 1+ to 2-, 2+ to 3-, 3+ to 4-, and 4+ to 5-, so there are 4 roots.
Daniel Cao Labora
Jun 28, 2015

Multiply by (x-1)(x-2)(x-3)(x-4)(x-5) both sides of the equation. The roots are still the same. Now you have a polynomial of degree 5 so the first expression has 5 roots. In (-inf,1) the function always decreases and in (5,inf) always increases. With that info it is a trivial check gives that there are no roots on (-inf,1) and there is only one root in (5,inf). It is also a trivial check by Bolzano that there is one root in each of these intervals (1,2), (2,3), (3,4), (4,5). There are no more roots by the Fundamental Theorem of Algebra so 4 is the answer

Jack Lam
Jun 27, 2015

Before we begin, first we will prove a useful and relevant result.

Given P ( x ) = i = 1 n ( x x i ) P(x) = \prod_{i=1}^n (x - x_i)

Prove that P ( x ) P ( x ) = i = 1 n 1 x x i \frac{P'(x)}{P(x)} = \sum_{i=1}^n \frac{1}{x - x_i}

l n ( P ( x ) ) = i = 1 n l n ( x x i ) ln(P(x)) = \sum_{i=1}^n {ln(x - x_i)}

Differentiate both sides wrt. x

P ( x ) P ( x ) = i = 1 n 1 x x i \frac{P'(x)}{P(x)} = \sum_{i=1}^n \frac{1}{x - x_i}

Moving back to the original problem:

Let P ( x ) = ( x 1 ) ( x 2 ) ( x 3 ) ( x 4 ) ( x 5 ) P(x) = (x-1)(x-2)(x-3)(x-4)(x-5)

\therefore It follows that P ( x ) P ( x ) = 1 \frac{P'(x)}{P(x)} = 1 is out equivalent problem

P ( x ) = P ( x ) P'(x) = P(x)

\therefore

...It follows, that I have no idea what I'm doing...

Only that I took something completely unrelated and crammed it into the problem.

Anyone have any ideas?

Moderator note:

Here's one way to complete your solution according to the approach that you have chosen:

  1. Apply the intermediate value theorem to P ( x ) P ( x ) P(x) - P'(x) , at the values where P ( x ) = 0 P(x) = 0 .
  2. Show that P ( x ) P'(x) is alternating, and hence there are at least n 1 n - 1 solutions in the range [ 1 , n ] [1, n ] .
  3. Show that there is another solution in the range [ n , 2 n ] [ n, 2n] .
  4. Hence, we have found n n roots of the n n degree polynomial P ( x ) P ( x ) P(x) - P'(x) , thus that is all of them.

It is very clear that x=1, 2, 3, 4, 5 are asymptotes of the function f(x) f ( x ) = k = 1 5 1 x k 1. I n v e s t i g a t i n g n e a r t h e a s y m p t o t e s , f ( n + . 01 ) > 0 , f ( n + . 99 ) < 0 , f o r n = 1 , 2 , 3 , 4 , In each of the FOUR interval between 1, 2, 3, 4 .. f(x) changes signs. So there are FOUR real roots for x [1,5] \displaystyle f(x)= \sum_{k=1}^5\dfrac 1 {x-k} -1.\\Investigating~near ~ the~asymptotes,\\f(n+.01)>0, f(n+.99)<0, for~~ n=1, ~2,~3,~4,\\\text {In each of the FOUR interval between 1, 2, 3, 4 .. f(x) changes signs.} \\\text{So there are FOUR real roots for x [1,5]} Surprisingly, the notes I had posted has vanished!! \color{#D61F06}{\text {Surprisingly, the notes I had posted has vanished!! }}
In the notes I had given exact value of the fifth root between 8 and 9. I had also reasoned whey the fifth root is not between 5 and 6 as is the case of other roots. When the function is graphed it shows only one intersection with x-axis between asymptotes.
It may be noted that a reciprocal graph of the type y = 1 x always have ONLY ONE intersection between asymptotes. y= \dfrac 1 x \text{always have ONLY ONE intersection between asymptotes. }

Moderator note:

Like the other solution, you have only shown that there is 5 change in signs. How do you know that the change in sign in the your interval doesn't change twice?

Mohsin Shiraz
Jun 11, 2015

The given function can be modified to : (x - 2) (x - 3) (x - 4)(x - 5) + (x - 1) (x - 3)(x - 4)(x - 5) + (x - 1) (x - 2) (x - 4)(x - 5) +(x - 1) (x - 2) (x - 3)(x - 5) + (x - 1) (x - 2)(x - 3) (x - 4) - (x - 1) (x - 2) (x - 3)(x - 4)(x - 5) = f(x)

When x = 1, f(x) is +ve
x = 2, f(x) is -ve
x = 3, f(x) is +ve
x = 4, f(x) is -ve
x = 5, f(x) is +ve

Since there are 5 sign changes, 4 roots are there.

Moderator note:

Almost complete. You have only shown that there is 5 change in sign which means that there is at least 4 real roots. What is needed to prove that there can't be more than 4 roots? Hint: Can there be other roots that fall outside of the given range?

Bonus question : Can you generalize this? That is, prove that there is exactly n 1 n-1 roots for the equation m = 1 n 1 x m = 1 \displaystyle \sum_{m=1}^n \frac1{x-m} = 1 in the interval [ 1 , n ] [1,n] .

exactly sir......i thinked it same way............ :)

Madhukar Thalore - 6 years ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...