Quick! Where's the Chebyshev Table?

Geometry Level 5

cos ( 2 x ) = 2 cos 2 ( x ) 1 cos ( 3 x ) = 4 cos 3 ( x ) 3 cos ( x ) cos ( 4 x ) = 8 cos 4 ( x ) 8 cos 2 ( x ) + 1 \large{\begin{aligned} \cos(2x) &=& 2\cos^2(x) - 1 \\ \cos(3x) &=& 4\cos^3(x) - 3\cos(x) \\ \cos(4x) &=& 8\cos^4(x) - 8\cos^2(x)+1 \\ \end{aligned} }

Above shows the first 3 examples of writing cos ( n x ) \cos(nx) in terms of a polynomial of cos ( x ) \cos(x) for positive integer n n .

If we write cos ( 2015 x ) \cos(2015x) in terms of a polynomial of cos ( x ) \cos(x) , what is the coefficient of cos 3 ( x ) \cos^3(x) ?

If the value is Y Y , submit your answer as Y ÷ 2015 Y \div 2015 .


The answer is 676704.

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.

3 solutions

Isaac Buckley
Jul 9, 2015

I came up with a way to determine the coefficient of cos 3 ( x ) \cos^3(x) for cos ( ( 4 k + 3 ) x ) \cos((4k+3)x) .

We apply the standard method of using De Moivres theorem.

c o s ( ( 4 k + 3 ) x ) = [ ( c o s ( x ) + i s i n ( x ) ) 4 k + 3 ] cos((4k+3)x)=\Re[(cos(x)+isin(x))^{4k+3}]

Expanding the RHS we find the coefficient of cos 3 ( x ) \cos^3(x) can be found.

( 4 k + 3 1 ) cos ( x ) i 4 k + 2 sin 4 k + 2 + ( 4 k + 3 3 ) cos 3 ( x ) i 4 k sin 4 k ( x ) + . . . \binom{4k+3}{1}\cos(x)i^{4k+2}\sin^{4k+2}+\binom{4k+3}{3}\cos^3(x)i^{4k}\sin^{4k}(x)+...

With a little bit of simplifying this implies the coefficient of cos 3 ( x ) \cos^3(x) is

C 3 = 4 ( 4 k + 3 ) ( 2 k + 1 ) ( k + 1 ) 3 C_3=\frac{4(4k+3)(2k+1)(k+1)}{3}

Let's try for a few known values that we can test from wiki to make sure.

for k = 0 k=0 , C 3 = 4 C_3=4

for k = 1 k=1 , C 3 = 56 C_3=56

for k = 2 k=2 , C 3 = 220 C_3=220

To find it for cos ( 2015 x ) \cos(2015x) we use k = 503 k=503

Y = 2015 × 676704 \implies Y=2015\times676704

@Isaac Buckley , we really liked your comment, and have converted it into a solution. If you subscribe to this solution, you will receive notifications about future comments.

Brilliant Mathematics Staff - 5 years, 11 months ago
Kartik Sharma
Sep 19, 2015

cos ( n x ) = e i n x + e i n x 2 \displaystyle \cos(nx) = \frac{{e}^{inx} + {e}^{-inx}}{2}

We want that in terms of cos ( x ) \cos(x) .

cos ( n x ) = c o s ( x ) + i 1 cos 2 ( x ) n + c o s ( x ) i 1 cos 2 ( x ) n 2 \displaystyle \cos(nx) = \frac{{cos(x) + i\sqrt{1 - \cos^2(x)}}^{n} + {cos(x) - i\sqrt{1 - \cos^2(x)}}^{n}}{2}

Let's change some variables

T n ( z ) = z + z 2 1 n + z z 2 1 n 2 \displaystyle T_n(z) = \frac{{z + \sqrt{z^2 -1}}^{n} + {z - \sqrt{z^2 - 1}}^{n}}{2}

Now, we will use binomial theorem,

T n ( z ) = k = 0 n 2 ( n 2 k ) ( z 2 1 ) k z n 2 k \displaystyle T_n(z) = \sum_{k=0}^{\left \lfloor \frac{n}{2} \right \rfloor}{\binom{n}{2k} {(z^2 -1)}^{k} {z}^{n-2k}} [I guess that's quite simple to get. Try yourself.]

T n ( z ) = k = 0 n 2 ( n 2 k ) ( 1 z 2 ) k z n \displaystyle T_n(z) = \sum_{k=0}^{\left \lfloor \frac{n}{2} \right \rfloor}{\binom{n}{2k} {(1-{z}^{-2})}^{k} {z}^{n}}

T n ( z ) = k = 0 n 2 ( n 2 k ) m = 0 k ( k m ) ( 1 ) m z 2 m z n \displaystyle T_n(z) = \sum_{k=0}^{\left \lfloor \frac{n}{2} \right \rfloor}{\binom{n}{2k} \sum_{m=0}^{k}{\binom{k}{m} (-1)^m {z}^{-2m}}{z}^{n}}

Now, changing order of summation,

T n ( z ) = k = 0 n 2 n 2 ( n k 1 ) ! ( n 2 k ) ! k ! ( 1 ) k ( 2 z ) n 2 k \displaystyle T_n(z) = \sum_{k=0}^{\left \lfloor \frac{n}{2} \right \rfloor}{\frac{n}{2} \frac{(n-k-1)!}{(n-2k)! k!} {(-1)}^{k} {(2z)}^{n-2k}}

Now, the problem can easily be done since we know the series and hence, the coefficients.

I will give a much more detailed proof of "changing order of summation" part tomorrow. It's late now.

Hobart Pao
Aug 24, 2015

I solved the problem by observation.

  • Let's call each chebyshev polynomial cos n x \cos nx P n P_{n} My observations are the following:

  • Only odd numbers of n n yield chebyshev polynomials with x 3 x^{3}

  • The abs. value of coefficients of x 3 x^{3} for n=3, 5, 7, 9 are: 4, 20, 56, and 120. These can be written as 4 x 1, 4 x 5, 4 x 14, and 4 x 30. Notice: 1, 5, 14, 30 are in the form x = 1 n 2 0.5 x 2 \displaystyle \sum_{x=1}^{\frac{n}{2}-0.5} x^{2} . So, for P 2015 P_{2015} , our abs. value of the coefficient will be in the form 4 x = 1 1007 x 2 4 \displaystyle \sum_{x=1}^{1007} x^{2} . Use the formula 1 6 z ( z + 1 ) ( 2 z + 1 ) \dfrac{1}{6} z(z+1)(2z+1) to compute the power sum with exponent of 2 to get 1 6 ( 1007 ) ( 1008 ) ( 2015 ) \dfrac{1}{6} (1007)(1008)(2015) and multiply that by 4 to obey the rule.

  • P 3 , 7 , 11... P_{3, 7, 11...} have a positive value of the coefficient for x 3 x^{3} (this is an arithmetic progression in the form T y = n = 3 + 4 ( y 1 ) T_{y} = n= 3+4(y-1) and P 5 , 9 , 13... P_{5, 9, 13...} have negative value of coefficient for x 3 x^{3} (this is an arithmetic progression in the form T y = n = 5 + 4 ( y 1 ) T_{y}= n = 5+4(y-1) . So, 2015 = 3 + 4 ( y 1 ) 2015 = 3+4(y-1) gives a whole number result, no decimals, so our coefficient is positive.

The answer calls for dividing the coefficient of x 3 x^{3} in P 2015 P_{2015} by 2015 2015 so we have ( 4 ) ( 1007 ) ( 1008 ) ( 2015 ) ( 6 ) ( 2015 ) \dfrac{(4)(1007)(1008)(2015)}{(6)(2015)} which is equal to ( 4 ) ( 1007 ) ( 1008 ) ( 6 ) \dfrac{(4)(1007)(1008)}{(6)} , and the answer is 676704 \boxed{676704} .

Let me know if I screwed up any variable names because I initially solved the problem without the formulas, I only made the formulas fancy for posting the solution.

Hobart Pao - 5 years, 9 months ago

Log in to reply

You need to prove that all the sequences you've found follows a certain progression. It may appear to be true for small values of n n , but that doesn't guarantee that it works for all n n . So your solution does not work.

Pi Han Goh - 5 years, 9 months ago

Log in to reply

I'll prove it with induction when I get a chance?

Hobart Pao - 5 years, 9 months ago

Log in to reply

@Hobart Pao Sure. By all means! =D

Pi Han Goh - 5 years, 8 months ago

Why not use this expression ?

It is also quite easy to prove it. It is the most straight forward thinking to use euler and binomial, I guess.

Kartik Sharma - 5 years, 8 months ago

Log in to reply

POST POST POST!

Pi Han Goh - 5 years, 8 months ago

Log in to reply

I will post tomorrow.

Kartik Sharma - 5 years, 8 months ago

Log in to reply

@Kartik Sharma I CAN'T WAIT! I CAN'T WAIT!

Pi Han Goh - 5 years, 8 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...