Experimental... again

Algebra Level 3
  • a 3 + b 3 \sqrt[3]{a} + \sqrt[3]{b} is a root of 10 x 3 30 x 29 10x^3 - 30x - 29 .
  • a 5 + b 5 \sqrt[5]{a} + \sqrt[5]{b} is a root of 10 x 5 50 x 3 + 50 x 29 10x^5 - 50x^3 + 50x - 29 .
  • a 7 + b 7 \sqrt[7]{a} + \sqrt[7]{b} is a root of n 1 x 7 + n 2 x 5 + n 3 x 3 + n 4 x + n 5 n_1x^7 + n_2x^5 + n_3x^3 + n_4x + n_5 .

If gcd ( n 1 , n 2 , n 3 , n 4 , n 5 ) = 1 \gcd(n_1, n_2, n_3, n_4, n_5) = 1 with n 1 > 0 n_1 > 0 , what is the value of n 1 + n 2 + n 3 + n 4 + n 5 ? |n_1|+|n_2|+|n_3|+|n_4|+|n_5|?


The answer is 319.

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.

2 solutions

Mark Hennings
May 24, 2018

If ω = 1 2 + 1 2 i 3 \omega = -\tfrac12 + \tfrac12i\sqrt{3} is a primitive cube root of unity, consider the complex numbers r = u + v s = u ω + v ω 2 t = u ω 2 + v ω r \; = \; u + v \hspace{1cm} s \; = \; u\omega + v\omega^2 \hspace{1cm} t \; = \; u\omega^2 + v\omega Then we calculate r + s + t = 0 r s + r t + s t = 3 u v r s t = u 3 + v 3 r+s+t = 0 \hspace{1cm} rs + rt + st \; = \; -3uv \hspace{1cm} rst \; = \; u^3 + v^3 and hence r , s , t r,s,t will be the roots of 10 X 3 30 X 29 = 0 10X^3 - 30X - 29 = 0 provided that u v = 1 uv = 1 and u 3 + v 3 = 29 10 u^3 + v^3 = \tfrac{29}{10} , which means that u 3 , v 3 u^3,v^3 are the roots of the quadratic 0 = X 2 29 10 X + 1 = 1 10 ( 5 X 2 ) ( 2 X 5 ) 0 \; = \; X^2 - \tfrac{29}{10}X + 1 \; = \; \tfrac{1}{10}(5X - 2)(2X - 5) so that we have u 3 , v 3 = 5 2 , 2 5 u^3,v^3 = \tfrac52,\tfrac25 . Thus we deduce that a 3 + b 3 \sqrt[3]{a} + \sqrt[3]{b} is a root of 10 X 3 30 X 29 = 0 10X^3 - 30X - 29 = 0 when a = u 3 = 5 2 a =u^3= \tfrac52 , b = v 3 = 2 5 b =v^3= \tfrac25 . This is the only real root of the cubic.

If we put y = a 5 y = \sqrt[5]{a} and z = b 5 z = \sqrt[5]{b} , then (note that y z = 1 yz=1 and y 5 + z 5 = a + b = 29 10 y^5 + z^5 = a+b = \tfrac{29}{10} ) ( y + z ) 3 = ( y 3 + z 3 ) + 3 ( y + z ) ( y + z ) 5 = 29 10 + 5 ( y 3 + z 3 ) + 10 ( y + z ) \begin{aligned} (y+z)^3 & = \; (y^3 + z^3) + 3(y+z) \\ (y+z)^5 & = \; \tfrac{29}{10} + 5(y^3 + z^3) + 10(y+z) \end{aligned} so that ( y + z ) 5 = 29 10 + 5 ( ( y + z ) 3 3 ( y + z ) ) + 10 ( y + z ) (y+z)^5 \; = \; \tfrac{29}{10} + 5\big((y+z)^3 - 3(y+z)\big) + 10(y+z) and hence that 10 ( y + z ) 5 50 ( y + z ) 3 + 50 ( y + z ) 29 = 0 10(y+z)^5 - 50(y+z)^3 + 50(y+z) - 29 \; = \; 0 as stated.

If we put Y = a 7 Y = \sqrt[7]{a} and Z = b 7 Z = \sqrt[7]{b} , then (this time Y Z = 1 YZ = 1 and Y 7 + Z 7 = 29 10 Y^7 + Z^7 = \tfrac{29}{10} ) ( Y + Z ) 3 = ( Y 3 + Z 3 ) + 3 ( Y + Z ) ( Y + Z ) 5 = ( Y 5 + Z 5 ) + 5 ( Y 3 + Z 3 ) + 10 ( Y + Z ) ( Y + Z ) 7 = 29 10 + 7 ( Y 5 + Z 5 ) + 21 ( Y 3 + Z 3 ) + 35 ( Y + Z ) = 29 10 + 7 [ ( Y + Z ) 5 5 ( Y 3 + Z 3 ) 10 ( Y + Z ) ] + 21 ( Y 3 + Z 3 ) + 35 ( Y + Z ) = 29 10 + 7 ( Y + Z ) 5 14 ( Y 3 + Z 3 ) 35 ( Y + Z ) = 29 10 + 7 ( Y + Z ) 5 14 [ ( Y + Z ) 3 3 ( Y + Z ) ] 35 ( Y + Z ) = 29 10 + 7 ( Y + Z ) 5 14 ( Y + Z ) 3 + 7 ( Y + Z ) \begin{aligned} (Y+Z)^3 & = \; (Y^3 + Z^3) + 3(Y+Z) \\ (Y+Z)^5 & = \; (Y^5 + Z^5) + 5(Y^3 + Z^3) +10(Y + Z) \\ (Y+Z)^7 & = \; \tfrac{29}{10} + 7(Y^5+Z^5) + 21(Y^3 + Z^3) + 35(Y+Z) \\ & = \; \tfrac{29}{10} + 7\big[(Y+Z)^5 - 5(Y^3+Z^3) - 10(Y+Z)\big] + 21(Y^3+Z^3) + 35(Y+Z) \\ & = \; \tfrac{29}{10} + 7(Y+Z)^5 - 14(Y^3+Z^3) - 35(Y+Z) \; = \; \tfrac{29}{10} + 7(Y+Z)^5 - 14\big[(Y+Z)^3 - 3(Y+Z)\big] - 35(Y+Z) \\ & = \; \tfrac{29}{10} + 7(Y+Z)^5 - 14(Y+Z)^3 + 7(Y+Z) \end{aligned} so that 10 ( Y + Z ) 7 70 ( Y + Z ) 5 + 140 ( Y + Z ) 3 70 ( Y + Z ) 29 = 0 10(Y+Z)^7 - 70(Y+Z)^5 + 140(Y+Z)^3 - 70(Y+Z) - 29 \; = \; 0 Eisenstein's Irreducibility Criterion (using p = 2 p=2 ) tells us that 29 x 7 + 70 x 5 140 x 3 + 70 x 10 29x^7 + 70x^5 - 140x^3 + 70x - 10 is irreducible over Q \mathbb{Q} , and hence 10 x 7 70 x 5 + 140 x 3 70 x 29 10x^7 - 70x^5 + 140x^3 - 70x - 29 is the minimal polynomial of Y + Z Y+Z . Thus we deduce that n 1 + n 2 + n 3 + n 4 + n 5 = 10 + 70 + 140 + 70 + 29 = 319 |n_1| + |n_2| + |n_3| + |n_4| + |n_5| \; = \; 10 + 70 + 140 + 70 + 29 \; = \; \boxed{319}


It is worth noting that, with the assumption that a , b a,b are real, they are uniquely defined (to within ordering) by the fact that a 3 + b 3 \sqrt[3]{a} + \sqrt[3]{b} is a root of the irreducible cubic 10 x 3 30 x 29 10x^3 - 30x - 29 (and, if they were not real, the rest of the question would not be uniquely defined). Thus we do not need to know about the minimal polynomial of a 5 + b 5 \sqrt[5]{a} + \sqrt[5]{b} .

@Mark Hennings , I have edited the directive of this question and have edited your solution.

Andrew Hayes Staff - 2 years, 6 months ago

Using Excel Goal Seek, I found a root of the cubic polynomial equal to 2.094015108. I used the quadratic equation to get the other two roots that were both complex numbers. Again using Excel Goal Seek, I got a root of the 5th degree polynomial equal to 2.033677641. I divided this polynomial by (x-this root) to get a quartic polynomial which I solved with a quartic polynomial root calculator online. The roots were two complex numbers, a negative number and a positive number. I then solved for a and b from a^1/3 + b^1/3 = 2.094015108 and a^1/5 + b^1/5 = 2.033677641. (Using the other positive, non-complex root of the 5th degree polynomial gave a negative solution and so could not be the root a^1/5+b^1/5.) With a and b determined, I calculated the root a^1/7+b^1/7 =r and got numbers for r^7, r^5, and r^3. I formed n1 r^7+n2 r^5+n3 r^3+n4 r+n5=0 and used Excel Solver to get the values of n1 to n5. I took the absolute value of the solutions n1 to n5 and got 319.

Phil Lawroski - 2 years, 6 months ago

After the first paragraph the steps are straightforward. But what motivanted you to do this first steps? Aren't you providing a method to solve 3rd degree equations without the x^2 term (in the form sum of 2 cubic roots)?

I confess I did the task of finding those by numerical methods.

Pau Cantos - 2 years, 6 months ago

Log in to reply

If you consider the cubic equation X 3 3 a X 2 + b X c = 0 X^3 - 3aX^2 + bX - c = 0 as an equation in X a X-a , it becomes ( X a ) 3 + d ( X a ) + e = 0 (X-a)^3 + d(X-a) + e = 0 . In other words, it is fine to consider only cubic equation with no X 2 X^2 term, since every cubic equation can be converted into that form by a simple translation.

My method is perfectly general, and is Cardano's solution for the cubic, viewed through the lens of Galois Theory - this last gives the reason for the r , s , ω r,s,\omega formalism when describing the roots of the cubic.

Mark Hennings - 2 years, 6 months ago

Log in to reply

It's good to know this, but I also wonder some more things:

First, I don't understand what you say in the last coment, I mean, there's only one root r of the cubic polinomial, but r=(a)^1/3+(b)^1/3 does no define a and b. You need the whole system right?

On the other hand, don't you think there must be some other easyer solution without having to find the roots at any time? I mean, the solution was more or less predictible, there is some patern underlinding the succession of coefficients in the succession of polinomials. As a bonus, could we find a closed form for those? Another interesting thing that I found on this polinomials is that they all share -9 and -49 as relative maximum-minimum values respectively. Can we use this anywhere? Can we prove that this property will continue.

Pau Cantos - 2 years, 6 months ago

Log in to reply

@Pau Cantos If ω \omega is a primitive cube root of unity, then consider a = r + s b = r ω + s ω 2 c = r ω 2 + s ω a \; = \; r + s \hspace{1cm} b \; = \; r\omega + s\omega^2 \hspace{1cm} c \; = \; r\omega^2 + s\omega Then a + b + c = 0 a b + a c + b c = 3 r s a b c = r 3 + s 3 a+b+c \; = \; 0 \hspace{1cm} ab + ac + bc \; = \; -3rs \hspace{1cm} abc \; = \; r^3 + s^3 (check these calculations). Thus a , b , c a,b,c are the roots of the cubic equation X 3 + β X γ = 0 X^3 + \beta X - \gamma = 0 provided that r s = 1 3 β r 3 + s 3 = γ rs \; = \; -\tfrac13\beta \hspace{1cm} r^3 + s^3 \; = \; \gamma and hence provided that r 3 , s 3 r^3,s^3 are the roots of the quadratic equation X 2 γ X 1 27 β 3 = 0 X^2 - \gamma X - \tfrac{1}{27}\beta^3 \; = \; 0 Find a root of this quadratic equation, and take its cube root. Call that number r r . Then define s = β 3 r s = -\tfrac{\beta}{3r} , and we have found r , s r,s such that a , b , c a,b,c become the roots of the cubic X 3 + β X γ = 0 X^3 + \beta X - \gamma = 0 .

You have two choices for which solution of the quadratic to take, and then three choices for how to take its cube root. Thus there are six choices for r r , and a matching choice for s s in each case. If we look carefully, each of these six choices for r r still gives the same numbers a , b , c a,b,c as the roots of the cubic, but each time the three roots come in a different order.

Mark Hennings - 2 years, 6 months ago

Log in to reply

@Mark Hennings Sorry, I thing I didn't explained myself clear enought. I understand all this work. But what I mean is that then, aren't you assuming that having r=u^(1/3)+v^(1/3)=a^(1/3)+b^(1/3) means a=u and b=v or the other way arround? But this is not true there is one degree of freedom left, thus infinitely many solutions (only using the first information). I mean, a=0.5055, b=2.18385 (aprox.) also fills the first requeriment.

Pau Cantos - 2 years, 6 months ago

Log in to reply

@Pau Cantos I showed that r = a 1 3 + b 1 3 r = a^{\frac13} + b^{\frac13} is a root of the cubic precisely when a , b = 5 2 , 2 5 a,b = \tfrac52,\tfrac25 ;. This means that the only uncertainty about r r (apart from choosing which is which between a a and b b ) is choosing which cube root to take. As I commented in my solution, if you choose not to take the real cube root for either a a or b b , then the rest of the question is not true (the polynomial satisfied by a 1 5 + b 1 5 a^{\frac15} + b^{\frac15} is not the one specied, to be precise. This means that there is no slack in the problem.

Mark Hennings - 2 years, 6 months ago
Vinod Kumar
Nov 26, 2018

Root is {5/2}^(1/7) + {2/5}^(1/7) and the polynomial is (Guessed and verified with WolframAlpha)

10x^7 - 70x^5 + 140x^3 + 70x - 29,

Answer=(10+70+140+70+29)=319

I really like and respect your urge to develop your critical thinking even at this age. 🙂

Praveen Kumar - 2 years, 6 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...