RMO Practice Problem - Algebra 2

Algebra Level 5

Let ( a , b ) (a,b) be the pair of integers such that

a x 17 + b x 16 + 1 \large ax^{17} + bx^{16} +1

is divisible by x 2 x 1 x^2 - x -1 .

Then find the sum of all such pairs.

Details and Assumptions:

  • If you get ( 2 , 3 ) (2,3) and ( 3 , 4 ) (3,4) as your answer, then express your answer as sum of all those values i.e. 2 + 3 + 3 + 4 = 12 2+3+3+4=12 .

Try this set RMO Practice Problems


The answer is -610.

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

We are looking for a , b a,b such that a x 17 + b x 16 = x 16 ( a x + b ) = 1 , ax^{17} + bx^{16} = x^{16}(ax + b) = -1, where

x 2 = x + 1 x 4 = x 2 + 2 x + 1 = 3 x + 2 x 8 = 9 x 2 + 12 x + 4 = 21 x + 13 x^{2} = x + 1 \Longrightarrow x^{4} = x^{2} + 2x + 1 = 3x + 2 \Longrightarrow x^{8} = 9x^{2} + 12x + 4 = 21x + 13

x 16 = 441 x 2 + 546 x + 169 = 987 x + 610. \Longrightarrow x^{16} = 441x^{2} + 546x + 169 = 987x + 610.

Now x 2 x 1 = 0 x = 1 ± 5 2 , x^{2} - x - 1 = 0 \Longrightarrow x = \dfrac{1 \pm \sqrt{5}}{2}, so x 16 = 987 ( 1 ± 5 ) 2 + 610 = 2207 ± 987 5 2 . x^{16} = 987*\dfrac{(1 \pm \sqrt{5})}{2} + 610 = \dfrac{2207 \pm 987\sqrt{5}}{2}.

Thus a x + b = 1 x 16 a ( 1 ± 5 ) 2 + b = 2 2207 ± 987 5 ax + b = -\dfrac{1}{x^{16}} \Longrightarrow a*\dfrac{(1 \pm \sqrt{5})}{2} + b = -\dfrac{2}{2207 \pm 987\sqrt{5}}

( a + 2 b ) ± a 5 = 4 2207 ± 987 5 2207 987 5 2207 987 5 = 2207 ± 987 5 . \Longrightarrow (a + 2b) \pm a\sqrt{5} = -\dfrac{4}{2207 \pm 987\sqrt{5}} * \dfrac{2207 \mp 987\sqrt{5}}{2207 \mp 987\sqrt{5}} = -2207 \pm 987\sqrt{5}.

So for both cases we have that a = 987 a = 987 and a + 2 b = 2207 b = 1597 , a + 2b = -2207 \Longrightarrow b = -1597, and so

a + b = 987 + ( 1597 ) = 610 . a + b = 987 + (-1597) = \boxed{-610}.

what a genius . give me some of ur knowledge

Kaustubh Miglani - 5 years, 8 months ago

Log in to reply

You can shorten the solution even further by noticing: A root of x^2-x-1 must also be a root of ax^17+ bx^16 + 1

As there are two roots you get: a ϕ 17 + b ϕ 16 + 1 = 0 a* \phi^{17} + b*\phi^{16} + 1 = 0 and a ( 1 ϕ ) 17 + b ( 1 ϕ ) 16 + 1 = 0 a* (1-\phi)^{17} + b*(1-\phi)^{16} +1= 0

Subtract them from another, divide by 5 \sqrt{5} , reorder the terms and you can immediately see a and -b are Fibonacci numbers, in fact the 16th and 17th.

I really like the problem, cause there are several short and interesting solutions.

Meer Kat - 5 years, 8 months ago

An absolutely genius way to solve.. Too good !! :D

Hrishik Mukherjee - 5 years, 6 months ago

Log in to reply

Thanks! :)

Brian Charlesworth - 5 years, 6 months ago

Plus, there is no rocket science involved, 8th grade mathematics ....

Hrishik Mukherjee - 5 years, 6 months ago

hhow u know that only that one pair exists

Suneel Kumar - 5 years, 3 months ago
Chew-Seong Cheong
Sep 27, 2015

When f ( x ) = a x 17 + b x 16 + 1 f(x)=ax^{17}+bx^{16}+1 is divisible by g ( x ) = x 2 x 1 g(x)=x^2-x-1 . It means that f ( x ) f(x) has the same roots as g ( x ) g(x) . Now, let us look at x x .

\(\begin{array} {} x = x + 0 & \Rightarrow x^1 = \color{blue}{F_1}x + \color{blue}{F_0} \quad \quad \color{blue}{F_n \text{ is the } n^{th} \text{ Fibonacci number}} \\ x^2 = x + 1 & \Rightarrow x^2 = F_2x + F_1 \\ x^3 = x^2 + x = 2x+1 & \Rightarrow x^3 = F_3x+F_2 \\ x^4 = 2x^2+x = 3x + 2 & \Rightarrow x^4 = F_4x+F_3 \\ x^5 = 3x^2 + 2x = 5x + 3 & \Rightarrow x^5 = F_5x+F_4 \\ ... & \Rightarrow x^{16} = F_{16}x + F_{15} \\ ... & \Rightarrow x^{17} = F_{17}x + F_{16} \end{array} \)

a x 17 + b x 16 + 1 = 0 ( a F 17 + b F 16 ) x + a F 16 + b F 15 + 1 = 0 \begin{aligned} \Rightarrow ax^{17} + bx^{16} + 1 & = 0 \\ (aF_{17}+bF_{16})x + aF_{16} + bF_{15} + 1 & = 0 \end{aligned}

Equating the coefficients on both sides, we have:

{ a F 17 + b F 16 = 0 . . . ( 1 ) a F 16 + b F 15 = 1 . . . ( 2 ) \begin{cases} aF_{17} + bF_{16} = 0 &...(1) \\ aF_{16} + bF_{15} = - 1 & ...(2) \end{cases}

{ ( 1 ) × F 15 : a F 15 F 17 + b F 15 F 16 = 0 . . . ( 3 ) ( 2 ) × F 16 : a F 16 2 + b F 15 F 16 = F 16 . . . ( 4 ) \begin{cases} (1)\times F_{15}: & aF_{15}F_{17} + bF_{15}F_{16} = 0 &...(3) \\ (2) \times F_{16}: & aF_{16}^2 + bF_{15}F_{16} = - F_{16} &...(4) \end{cases}

{ ( 3 ) ( 4 ) : ( F 15 F 17 F 16 2 ) a = F 16 ( 1 ) a = F 16 By Cassini’s identity* ( 1 ) : F 16 F 17 + b F 16 = 0 b = F 17 \begin{cases} (3)-(4): & (\color{#3D99F6}{F_{15}F_{17}-F_{16}^2})a = F_{16} & \Rightarrow \color{#3D99F6}{(1)} a = F_{16} & \color{#3D99F6}{\text{By Cassini's identity*}} \\ (1): & F_{16}F_{17} + bF_{16} = 0 & \Rightarrow b = - F_{17} \end{cases}

a = F 16 \Rightarrow a = F_{16} and b = F 17 b = -F_{17} and a + b = F 16 F 17 = F 15 = 610 a+b = F_{16}-F_{17} = - F_{15} = \boxed{-610}

\color{#3D99F6}{* } Cassini's identity F n 1 F n + 1 F n 2 = ( 1 ) n \color{#3D99F6}{\text{ }F_{n-1}F_{n+1} - F_n^2 = (-1)^n}

This solution is not true. The equation a F 16 + b F 15 = 1 aF_{16} + bF_{15} = -1 does not have single integer solution. For instance, (a,b) = ( F 14 , F 15 ) (F_{14}, -F_{15}) holds true (and this one looks like your "Cassini's solution"). More generally, a pair with this form ( F 16 k F 15 , F 17 + k F 16 ) (F_{16} - kF_{15}, -F_{17} + kF_{16}) is a solution with k is an arbitrary integer. The last two "if/then" statements are wrong.

trongnhan khong - 5 years, 8 months ago

Log in to reply

I actually worked it out numerically first.

a x 17 + b x 16 + 1 = 0 ( a F 17 + b F 16 ) x + a F 16 + b F 15 + 1 = 0 \begin{aligned} ax^{17} + bx^{16} + 1 & = 0 \\ (aF_{17} + bF_{16})x + aF_{16} + bF_{15} + 1 & = 0 \end{aligned} .

Equating the coefficients on both sides, we have:

{ a F 17 + b F 16 = 1597 a + 987 b = 0 . . . ( 1 ) a F 16 + b F 15 = 987 a + 610 b = 1 . . . ( 2 ) \begin{cases} aF_{17} + bF_{16} = 1597a + 987b = 0 &...(1) \\ aF_{16} + bF_{15} = 987a + 610b = - 1 & ...(2) \end{cases}

{ ( 1 ) × 610 : 974170 a + 602070 b = 0 . . . ( 3 ) ( 2 ) × 987 : 974169 a + 602070 b = 987 . . . ( 4 ) \begin{cases} (1)\times 610: & 974170a + 602070b = 0 &...(3) \\ (2) \times 987: & 974169a + 602070b = -987 &...(4) \end{cases}

{ ( 3 ) ( 4 ) : ( 1 ) a = 987 = F 16 Note that F 15 F 17 F 16 2 = 1 ( 1 ) : 1597 ( 987 ) + 987 b = 0 b = 1597 = F 17 \begin{cases} (3)-(4): & \color{#3D99F6}{(1)}a = 987 = F_{16} \quad \quad \color{#3D99F6}{\text{Note that } F_{15}F_{17}-F_{16}^2 = 1 } \\ (1): & 1597(987) + 987b = 0 \quad \Rightarrow b = - 1597 = - F_{17} \end{cases}

a + b = F 16 F 17 = F 15 = 610 a+b = F_{16} - F_{17} = - F_{15} = -610

I have done at least two other similar problems before without knowing Cassini's. I have consistently got the same results and found it beautiful.

Chew-Seong Cheong - 5 years, 8 months ago

Log in to reply

I agree with you that your idea is beautiful and comes naturally. But if you use the Cassini's as your "shortcut" like that, it may cause problems. Let's work on the system of (1) and (2). Since gcd ( F n + 1 , F n ) = 1 \gcd(F_{n+1},F_{n}) = 1 , solutions of (1) take form (a,b) = ( F 16 k , F 17 k ) (F_{16}k,-F_{17}k) . Let it take place of (a,b) in (2), by the identity we get the unique k = 1 immediately. I think that's how Cassini's used in your solution. We can't get the answer to the original problem if we just only solve (2). That's what I mean.

trongnhan khong - 5 years, 8 months ago

Log in to reply

@Trongnhan Khong Yes, you are right. I have changed the solution.

Chew-Seong Cheong - 5 years, 8 months ago
Priyanshu Mishra
Sep 27, 2015

Let p p , q q be the roots of the equation x 2 x 1 { x }^{ 2 }-x-1 . By Vieta's theorem p + q = 1 p + q = 1 and p q = 1 pq = -1 . Note that p p and q q must also be the roots of a x 17 + b x 16 + 1 = 0 \large\ a{ x }^{ 17 }+b{ x }^{ 16 }+1=0 . Thus

a p 17 + b p 16 = 1 \large\ a{ p }^{ 17 }+b{ p }^{ 16 }=-1 and a q 17 + b q 16 = 1 \large\ a{ q }^{ 17 }+b{ q }^{ 16 }=-1 .

Multiplying the first of these equations by q 16 q^{16} , the second one by p 16 p^{16} , and using the fact that p q = 1 pq = -1 , we find

a p + b = q 16 \large\ ap+b=-{ q }^{ 16 } and a q + b = p 16 \large\ aq+b=-{ p }^{ 16 } . ( 1 ) (1)

Thus

a = p 16 q 16 p q = ( p 8 + q 8 ) ( p 4 + q 4 ) ( p 2 + q 2 ) ( p + q ) \large\ a=\frac { { p }^{ 16 }-{ q }^{ 16 } }{ p-q } =({ p }^{ 8 }+{ q }^{ 8 })({ p }^{ 4 }+{ q }^{ 4 })({ p }^{ 2 }+{ q }^{ 2 })(p+q) .

Since

p + q = 1 p+q = 1 ,

p 2 + q 2 = ( p + q ) 2 2 p q = 1 + 2 = 3 \large\ { p }^{ 2 }+{ q }^{ 2 }={ (p+q) }^{ 2 }-2{ pq }=1+2=3

p 4 + q 4 = ( p 2 + q 2 ) 2 2 ( p q ) 2 = 9 2 = 7 \large\ { p }^{ 4 }+{ q }^{ 4 }={ ({ p }^{ 2 }+{ q }^{ 2 }) }^{ 2 }-2{ (pq) }^{ 2 }=9-2=7

p 8 + q 8 = ( p 4 + q 4 ) 2 2 ( p q ) 4 = 49 2 = 47 \large\ { p }^{ 8 }+{ q }^{ 8 }={ ({ p }^{ 4 }+{ q }^{ 4 }) }^{ 2 }-2{ (pq) }^{ 4 }=49-2=47

it follows that a = 1.3.7.47 = 987 a = 1 . 3 . 7 . 47 = 987 .

Likewise, eliminating a in ( 1 ) (1) gives

b = p 17 q 17 p q \large\ -b=\frac { { p }^{ 17 }-{ q }^{ 17 } }{ p-q }

= ( p 16 + q 16 ) ( p 14 + q 14 ) + . . . ( p 2 + q 2 ) + 1 \large\ ={ (p }^{ 16 }+{ q }^{ 16 })-({ p }^{ 14 }+{ q }^{ 14 })+\quad ...\quad -({ p }^{ 2 }+{ q }^{ 2 })+1

For n > 1 n > 1 , let k 2 n = p 2 n + q 2 n \large\ { k }_{ 2n }={ p }^{ 2n }+{ q }^{ 2n } . Then k 2 = 3 { { k }_{ 2 }=3 } and k 4 = 7 { k }_{ 4 }=7 , and

k 2 n + 4 = p 2 n + 4 q 2 n + 4 \large\ { k }_{ 2n+4 }={ p }^{ 2n+4 }{ q }^{ 2n+4 }

= ( p 2 n + 2 + q 2 n + 2 ) ( p 2 + q 2 ) p 2 q 2 ( p 2 n + q 2 n \large\ ={ (p }^{ 2n+2 }+{ q }^{ 2n+2 })({ p }^{ 2 }+{ q }^{ 2 })-{ p }^{ 2 }{ q }^{ 2 }({ p }^{ 2n }+{ q }^{ 2n }

= 3 k 2 n + 2 k 2 n \large\ =3{ k }_{ 2n+2 }-{ k }_{ 2n }

for n 3 n\ge 3 . Then k 6 = 18 { k }_{ 6 }=18 , k 8 = 47 { k }_{ 8 }=47 , k 10 = 123 { k }_{ 10 }=123 , k 12 = 322 { k }_{ 12 }=322 , k 14 = 843 { k }_{ 14 }=843 , k 16 = 2207 { k }_{ 16 }=2207 .

Hence,

b = 2207 843 + 322 123 + 47 18 + 7 3 + 1 = 1597 -b = 2207- 843+322 - 123+47- 18+7- 3+ 1 = 1597

or

( a , b ) = ( 987 , 1597 ) (a, b) = (987, -1597) .

So, a + b = 987 1597 = 610 a+b=987-1597=-610 .

Thus, answer : 610 \boxed{-610} .

汶良 林
Sep 29, 2015

why did you equated the co-efficient of x and constant term equal to ZERO?

Akshansh Jaiswal - 2 years, 8 months ago
Arturo Presa
Nov 28, 2015

In what follows we are going to assume that x 2 x 1 x^2-x-1 divides a x 17 + b x 16 + 1 ax^{17}+bx^{16}+1 . Then we can see that any solution of x 2 x 1 = 0 x^2-x-1=0 must be a solution of a x 17 + b x 16 + 1 = 0 ax^{17}+bx^{16}+1=0 .

Let α \alpha represent a solution of x 2 x 1 = 0 , x^2-x-1=0, then α 2 = α + 1 , \alpha^2=\alpha+1, and therefore for any natural number n n we have that α n + 2 = α n + 1 + α n . ( 1 ) \alpha^{n+2}=\alpha^{n+1}+\alpha^n.\:\:\:\:\:\:(1)
Now we can prove using induction that for any pair of real numbers c c and d d and for any natural numbers n > 2 n>2 and k < n k<n the following is true: c α n + d α n 1 = ( F k + 1 c + F k d ) α n k + ( F k c + F k 1 d ) α n k 1 ( 2 ) c\: \alpha^n+d\:\alpha^{n-1}=(F_{k+1}\:c+F_{k}\: d) \alpha^{n-k}+(F_{k}\:c+F_{k-1}\:d)\alpha^{n-k-1}\:\:\:\:\:\:\:\:(2) where F m F_m represents the m m- term of the Fibonacci sequence defined by F 0 = 0 , F_0=0, F 1 = 1 F_1=1 and F m + 1 = F m + F m 1 . F_{m+1}=F_{m}+F_{m-1}. Let us prove first that (2) is true for k = 1 k=1 . Indeed, using (1) we get that c α n + d α n 1 c\: \alpha^n+d\:\alpha^{n-1} = c ( α n 1 + α n 2 ) + d α n 1 = ( c + d ) α n 1 + c α n 2 = =c(\alpha^{n-1}+\alpha^{n-2})+d\:\alpha^{n-1}= (c+d)\alpha^{n-1}+c\:\alpha^{n-2}= ( F 2 c + F 1 d ) α n 1 + ( F 1 c + F 0 d ) α n 2 . (F_2 \:c+F_1\: d)\alpha^{n-1}+(F_1 \:c+F_0\: d)\alpha^{n-2}. Now let us prove that if (2) is true for any number natural k < n k<n then is would be true also for k + 1 k+1 , of course, if k + 1 < n . k+1 <n. Applying ( 1 ) (1) again we get that c α n + d α n 1 = ( F k + 1 c + F k d ) α n k + ( F k c + F k 1 d ) α n k 1 = c\: \alpha^n+d\:\alpha^{n-1}=(F_{k+1}\:c+F_{k}\: d) \alpha^{n-k}+(F_{k}\:c+F_{k-1}\:d)\alpha^{n-k-1}= = ( F k + 1 c + F k d ) ( α n k 1 + α n k 2 ) + ( F k c + F k 1 d ) α n k 1 =(F_{k+1}\:c+F_{k}\: d)( \alpha^{n-k-1}+\alpha^{n-k-2})+(F_{k}\:c+F_{k-1}\:d)\alpha^{n-k-1} = ( ( F k + 1 + F k ) c + ( F k + F k 1 ) d ) ( α n k 1 ) + ( F k + 1 c + F k d ) α n k 2 =((F_{k+1}+F_k)\:c+(F_{k}+F_{k-1})\: d)( \alpha^{n-k-1})+(F_{k+1}\:c+F_{k}\:d)\alpha^{n-k-2} = ( F k + 2 c + F k + 1 d ) α n k 1 + ( F k + 1 c + F k d ) α n k 2 . =(F_{k+2}\:c+F_{k+1}\: d) \alpha^{n-k-1}+(F_{k+1}\:c+F_{k}\:d)\alpha^{n-k-2}. This completes the proof of (2)

Now using (2) with n = 17 n=17 and k = 15 k=15 we obtain that if α \alpha is a zero of x 2 x 1 = 0 x^2-x-1=0 then 0 = a α 17 + b α 16 + 1 = ( F 16 a + F 15 b ) α 2 + ( F 15 a + F 14 b ) α + 1 = 0=a\:\alpha^{17}+b\:\alpha^{16}+1=(F_{16}\:a+F_{15}\:b)\alpha^2+(F_{15} a+F_{14} b)\alpha+1= = ( 987 a + 610 b ) α 2 + ( 610 a + 377 b ) α + 1. =(987 a+610 b)\alpha^2+(610a+377b)\alpha+1.

Therefore the polynomial a x 17 + b x 16 + 1 ax^{17}+bx^{16}+1 is divisible by x 2 x 1 x^2-x-1 if and only if ( 987 a + 610 b ) x 2 + ( 610 a + 377 b ) x + 1 (987 a+610 b)x^2+(610a+377b)x+1 is divisible by x 2 x 1. x^2-x-1. Then 987 a + 610 b = 1 987 a+610 b=-1 and 610 a + 377 b = 1. 610a+377b=-1. The only solution of this system of equations is ( a , b ) = ( 987 , 1597 ) . (a, b)=(987, -1597). So the answer to the question is a + b = 987 1597 = 610. a+b=987-1597=-610.

For a x 17 + b x 16 + 1 ax^{17}+bx^{16}+1 to be divisible by x 2 x 1 x^2-x-1 , a λ 17 + b λ 16 + 1 = 0 a \lambda^{17}+b \lambda^{16} + 1 = 0 should be true for λ \lambda being the roots of x 2 x 1 = 0 x^2-x-1=0

We can easily get λ = 1 ± 1 + 4 2 = 1 2 ± 5 2 \lambda = \frac{1 \pm \sqrt{1+4}}{2} = \frac{1}{2} \pm \frac{\sqrt{5}}{2} .

Substituting for λ \lambda we have two linear equations in a , b a,b as

( 1 2 + 5 2 ) 17 a + ( 1 2 + 5 2 ) 16 b = 1 \left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{17}a+\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{16}b=-1

( 1 2 5 2 ) 17 a + ( 1 2 5 2 ) 16 b = 1 \left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{17}a+\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{16}b=-1

Using Cramer's rule,

a = ( 1 2 5 2 ) 16 + ( 1 2 + 5 2 ) 16 ( 1 2 + 5 2 ) 17 ( 1 2 5 2 ) 16 ( 1 2 5 2 ) 17 ( 1 2 + 5 2 ) 16 = 987 a = \frac{-\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{16}+\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{16}}{\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{17}\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{16}-\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{17}\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{16}}=987

b = ( 1 2 5 2 ) 17 ( 1 2 + 5 2 ) 17 ( 1 2 + 5 2 ) 17 ( 1 2 5 2 ) 16 ( 1 2 5 2 ) 17 ( 1 2 + 5 2 ) 16 = 1597 b = \frac{\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{17}-\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{17}}{\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{17}\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{16}-\left(\frac{1}{2}-\frac{\sqrt{5}}{2}\right)^{17}\left(\frac{1}{2}+\frac{\sqrt{5}}{2}\right)^{16}}=-1597

Thus, giving a + b = 610 a+b = \boxed{-610}

I have done exactly the way you did. There is only one pair namely (987, -1597).

Lu Chee Ket - 5 years, 8 months ago
Sriram Vudayagiri
Sep 27, 2015

This is not a solution.

Use long division and you will find that remainder in each step follows a pattern. Find the pattern and find remainder finally. If the final remainder is mx+n, then m=n=0. Using this, try to get the values of a and b.

Elaborating:- m = F ( 17 ) × a + F ( 16 ) × b m \ = \ F(17) \times \ a \ + \ F(16) \times \ b

n = F ( 16 ) × a + F ( 15 ) × b + 1 n \ = \ F(16) \times \ a \ + \ F(15) \times \ b \ + \ 1

Solve m = n = 0 m \ = \ n \ = \ 0 as stated to get a a and b b

F ( n ) F(n) is the nth Fibonacci term.

Kunal Verma - 5 years, 8 months ago

sir can you please show that how a pattern is followed

Deepansh Jindal - 5 years, 1 month ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...