Special Polynomials

What is the largest value of d d , such that for some degree d d polynomial f ( x ) f(x) with integer coefficients, f ( x ) = 1024 |f(x)|=1024 for more than d d integer values of x x ?


The answer is 7.

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.

9 solutions

Brock Thunder
Nov 6, 2013

This is not rigorous, but intended to start discussion.

Let f ( x ) = x ( x r 1 ) ( x r 2 ) . . . ( x r n ) + 1024 f(x) = x(x - r_{1})(x - r_{2})...(x - r_{n}) + 1024

Then, if all the roots are distinct, we have f ( x ) = 1024 f(x) = 1024 for n n values of x x

We want to make f ( x ) = 1024 f(x) = -1024 so let x ( x r 1 ) ( x r 2 ) . . . ( x r n ) = 2048 x(x - r_{1})(x - r_{2})...(x - r_{n}) = -2048 Note that 2048 = 2 11 2048 = 2^{11} , we want to maximize the number of terms in a distinct product of factors that will equal 2048.

Clearly this is done by choosing the factors as 1 , 1 , 2 , 2 , 4 , 4 , 32 ) 1,-1,2,-2,4,-4,32) (or any other combination of length 7). So our polynomial must have degree at most 7.

I am stuck on how to make my argument more complete and rigorous. Any hints or suggestions?

That's a great start! I'm interested in hearing what the other 100 people who solved it have to say.

Calvin Lin Staff - 7 years, 7 months ago

Log in to reply

Did the same as above and couldn't really make up anything higher than 7, so took a guess.

It is also odd that you only get d+1 roots even in lower values of d.

Mirza Baig - 7 years, 7 months ago

Even I used the same logic.

I first started to think graphically, but, to no avail. Then I tried to construct such a polynomial and ended up following the same approach as yours. Calvin Sir, do you have some other solution in your mind, or it is on the same lines as this one?

Aman Gupta - 7 years, 7 months ago

Log in to reply

The above logic isn't necessarily correct, as it assumes that we have a factorization into integers, which need not always be the case.

You need to show why d 8 d \geq 8 is not possible. This generally boils down to an argument that for integers a a and b b , we have ( a b ) f ( a ) f ( b ) (a-b) \mid f(a) - f(b) . Hence, we can't have too many pairs where f ( a ) = 1024 f(a) = 1024 and f ( b ) = 1024 f(b) = -1024 .

Calvin Lin Staff - 7 years, 6 months ago

Log in to reply

Hmm...my proof has a different approach for this.. but the more I think about this... the more uncertain I get.

Xuming Liang - 7 years, 6 months ago
Nick Haliday
May 20, 2014

It is obvious that, for any integer x x and y y , x y x - y divides f ( x ) f ( y ) f(x) - f(y) ( f ( x ) f ( y ) = i a i ( x i y i ) f(x) - f(y) = \sum_i a_i (x^i - y^i) , the a i a_i are integer, and x y x - y divides x i y i x^i - y^i ). We will use this to show that 7 is an upper bound on d d , and in the process be led toward an f f that achieves that bound.

Suppose we have some maximum degree f f , and with it two integer sequences a 0 , a 1 , , a n 1 a_0, a_1, \dots, a_{n - 1} and b 0 , b 1 , , b m 1 b_0, b_1, \dots, b_{m - 1} such that f ( a i ) = 2 10 f(a_i) = 2^{10} and f ( b i ) = 2 10 f(b_i) = -2^{10} for all i i . Each a i a_i is a unique integer, distinct from any other a j a_j or b j b_j , and vice versa. Using the property from before, we see that each difference a i b j a_i - b_j is a divisor (positive or negative) of 2 11 2^{11} .

We will now derive a contradiction in the case that both sequences have more than one element.

Take two elements from each sequence and call them a 0 , a 1 a_0, a_1 and b 0 , b 1 b_0, b_1 . Then, considering the differences between the a i a_i and b 0 b_0 , we get b 0 = a 0 + 2 x 0 b_0 = a_0 + 2^{x_0} and b 0 = a 1 + 2 x 1 b_0 = a_1 + 2^{x_1} (there are actually arbitrary signs in front of of the powers of two, but they don't break our argument, and are tedious write in TeX so I'm not going address them; we're also assuming here that x 0 > x 1 x_0 > x_1 , and that's also a minor detail that doesn't affect the argument). This gives a 1 a 0 = 2 x 1 ( 2 x 0 x 1 1 ) a_1 - a_0 = 2^{x_1} (2^{x_0 - x_1} - 1) .

Doing the same with b 1 b_1 gives a 1 a 0 = 2 y 1 ( 2 y 0 y 1 1 ) a_1 - a_0 = 2^{y_1} (2^{y_0 - y_1} - 1) . But the uniqueness of factorizations combined with this implies that x 0 = y 0 x_0 = y_0 and y 0 = y 1 y_0 = y_1 , implying b 0 = b 1 b_0 = b_1 , contradicting our assumptions about distinctness.

So in all cases, we will have either exactly one a a such that f ( a ) = 2 10 f(a) = 2^{10} or one b b such that f ( b ) = 2 10 f(b) = -2^{10} . Without loss of generality we consider the latter case.

So we have a sequence a 0 , a 1 , , a n 1 a_0, a_1, \dots, a_{n - 1} and some b b . The maximum degree of the polynomial f ( x ) f(x) is then n n . We know that the polynomial f ( x ) 2 10 f(x) - 2^{10} , has n n roots, so, in fact, f f must have degree n n , and f ( x ) 2 10 f(x) - 2^{10} is determined up to a constant, in turn determining f ( x ) f(x) (up to that constant).

So f ( x ) 2 10 = c ( x ( b ± 2 i ) ) ( x ( b ± 2 j ) ) ( x ( b ± 2 k ) f(x) - 2^{10} = c (x - (b \pm 2^i))(x - (b \pm 2^j)) \cdots (x - (b \pm 2^k) . Plugging in b b we get that 2 11 = ± c X 2^{11} = \pm c X where X X is the product of a bunch of distinct positive and negative divisors of 2 11 2^{11} . For c c to be integer, X X needs to divide 2 11 2^{11} . So really we're looking to maximize the number of distinct divisors 2 11 2^{11} such that their product divides 2 11 2^{11} . But this is easy to do greedily! We can choose, for example, 2 3 , 2 2 , 2 1 , 2 0 , 2 0 , 2 1 , 2 2 -2^3, -2^2, -2^1, -2^0, 2^0, 2^1, 2^2 , and this will give the maximum number, namely, 7.

Close to correct solution. However, a 1 a 0 = ± 2 x 1 ( 2 x 0 x 1 ± 1 ) a_1 - a_0 = \pm 2^{x_1} (2^{x_0 - x_1} \pm 1) and a 1 a 0 = ± 2 y 1 ( 2 y 0 y 1 ± 1 ) a_1 - a_0 = \pm 2^{y_1} (2^{y_0 - y_1} \pm 1) only imply, by factorization, that x 1 = y 1 . x_1=y_1. Indeed, 2 2 1 = 2 1 + 1 , 2^2-1=2^1+1, so it may happen that x 0 x 1 = 2 , x_0-x_1=2, while y 0 y 1 = 1. y_0-y_1=1. So one has to be more careful with the signs here: they do matter, and this case is trickier than you thought.

Calvin Lin Staff - 7 years ago

|f(x)|=1024=2^10, Equation can be written as follows, f(x)=(x-a)(x-b)....-1024 or (x-a)(x-b)....+1024 were a,b,... are the 'd' integer roots of |f(x)|=1024. |f(x)|=1024=2^10=1 2 8 (-1) (-2) (-4) (-8), 1024 can only be written as multiple of up to 7 different integers. Substituting these sample values as a,b,... we get |f(x)|=1024 not only for these 7 values but also for 0 (|-1024|=1024).Thus a total of d+1 integer values this case is satisfied. Thus we can conclude that 7 is the top limit for the given case.

Sreenath Are
May 20, 2014

Consider the polynomial f(x)=(x-r 1)(x-r 2)...(x-r n)+1024. At any x=r n, the value of |f(x)| is 1024 (n solutions, degree n). For there to be more solutions, the value of f(x) must be -1024 at an integer value of x. Thus (x-r 1)(x-r 2)...(x-r n) must equal -2048, and each x-r i is an integer. The highest n for which this is possible is 7; the values of r_i are -4, -2, -1, 1, 2, 4, 32.

The form of f f considered is not the most general one: f ( x ) 1024 f(x)-1024 does not have to have all integer roots, and does not have to be monic. It is possible that other forms of f f provide bigger answers.

Calvin Lin Staff - 7 years ago
Tristan Pollner
May 20, 2014

If we write that f ( x ) 1024 = g ( x ) f(x)-1024=g(x) , where g ( x ) g(x) has degree d d , then it is possible (for a certain polynomial g ( x ) g(x) ) that there are d d solutions to the equation we must satisfy.

However, this is not good enough. We need at least one extra solution, for f ( x ) = 1024 f(x)=-1024 .

Then, we can say there must be at least one solution to g ( x ) = 2048 g(x) = -2048 . Therefore, because g ( x ) g(x) must have integer roots, we are looking for the maximum possible number of integers that multiply to 2048 -2048 , which will be our answer.

It turns out this maximum can be achieved by ( 8 ) ( 4 ) ( 1 ) ( 1 ) ( 2 ) ( 4 ) ( 8 ) (-8)(-4)(-1)(1)(2)(4)(8) , so our answer is d = 7 d=7 .

The statement " g ( x ) g(x) must have integer roots" is not justified: it is possible that some roots of g g are not integers.

The product does not have to equal to 2048 , -2048, it can divide it, because one may have g ( x ) = C ( x a 1 ) . . . ( x a d ) ) g(x)=C(x-a_1)...(x-a_d)) with some integer constant C . C.

These mistakes aside, the solution is very clearly written.

Calvin Lin Staff - 7 years ago
黎 李
May 20, 2014

No -8

No solution provided, just a hint on how the answer was obtained.

Calvin Lin Staff - 7 years ago
Anderson Silva
May 20, 2014

Without loss of generality, let f ( x ) = C ( x a 1 ) ( x a 2 ) ( x a d ) 1024 f(x)=C(x-a_{1})(x-a_{2})⋅⋅⋅(x-a_{d})-1024 . f ( x ) = 1024 i f x = a 1 , a 2 , , a d |f(x)|=1024 if x=a_{1},a_{2}, \ldots, a_{d} . Another value of x x , say x 0 x_{0} , gives f ( x ) |f(x)| =1024 if f ( x 0 ) = 2048 f(x_{0})=2048 . This happens when f ( x 0 = C ( ± 1 ) ( ± 2 ) ( ± 2 2 ) ( 2 3 ) f(x_{0}=C(\pm 1)(\pm 2)(\pm 2^2)(2^3) , where C = 4 C=4 . Adding another term of (-8) gives C = 1 2 C=\frac{1}{2} . Hence, the max value of d is 7 7 .

1) "Without loss of generality" is incorrect: it is not clear why all roots of f ( x + 1024 ) f(x+1024) are integers.

2) What are a 1 , . . . , a d ? a_1,...,a_d?

3) Why is f ( x 0 ) = C ( ± 1 ) ( ± 2 ) ( ± 2 2 ) ( 2 3 ) f(x_{0})=C(\pm 1)(\pm 2)(\pm 2^2)(2^3) ?

Calvin Lin Staff - 7 years ago
Randy Couture
May 20, 2014

Let f(x)=g(x)-1024 (+1024 is another approach to the solution). Let a 1 , a 2 , , a d a_{1}, a_2, \ldots, a_d be the roots of g such that g ( x ) = C ( x a 1 ) ( x a 2 ) ( x a d ) g(x)=C(x-a_{1})(x-a_{2}) \cdot \cdot \cdot (x-a_{d}) . g vanishes if x is any root of f. If g ( x 0 ) = 2048 g(x_{0})=2048 for some x 0 x_{0} , then f ( x 0 ) = 1024 f(x_{0})=1024 . This x 0 x_{0} is the (d+1)th value of x such that f ( x ) = 1024 |f(x)|=1024 . Since the polynomial has integer coefficients, Since 2048 = 2 11 2048=2^{11} , each term of g ( x 0 g(x_{0} has to be a power of 2. The highest value of d occurs when g ( x 0 ) = C ( 1 ) ( 1 ) ( 2 ) ( 2 ) ( 4 ) ( 4 ) ( 8 [ o r 8 ] ) = 2048 g(x_{0})=C(1)(-1)(2)(-2)(4)(-4)(8 [or -8])=2048 where C=4 [or -4].) Had we let g ( x 0 ) = C ( 1 ) ( 1 ) ( 2 ) ( 2 ) ( 4 ) ( 4 ) ( 8 ) ( 8 ) = 2048 g(x_{0})=C(1)(-1)(2)(-2)(4)(-4)(8)(-8)=2048 , C becomes non-integral. Hence, the answer is 7 7 .

The solution is incomplete. It is not obvious that all roots of g ( x ) g(x) are integers. It is also not clear why there cannot be more than one integer value x 0 x_0 such that g ( x 0 ) = 2048. g(x_0)=2048.

Calvin Lin Staff - 7 years ago
Xuming Liang
Nov 19, 2013

I will try my best to achieve all the level 5 expectations for this solution, although I have a feeling that my solution will be very similar to but more complicated than the one already submitted. Also I am new to English writing so sorry if some sentences are hard to understand.

First of all, f ( x ) \mid f(x)\mid =1024 is equivalent to f ( x ) 1024 = 0 f(x)-1024=0 and f ( x ) = 1024 f(x)=-1024 .

To approach this problem, I tried constructing a polynomial that satisfies the problem. To do this, let's assume that r i r_i ,where i = 1 , 2 , . . . , d i=1,2,...,d are distinct(repeated integers are counted once), are the integer solutions to the first equation. This polynomial can be written as: f ( x ) = a ( x r 1 ) ( x r 2 ) . . . ( x r d ) + 1024 f(x)=a(x-r_1)(x-r_2)...(x-r_d)+1024 for some integer a a (note that f ( x ) f(x) has integer coefficients). Now consider the second equation with this polynomial:

f ( x ) = a ( x r 1 ) ( x r 2 ) . . . ( x r d ) + 1024 = 1024 f(x)=a(x-r_1)(x-r_2)...(x-r_d)+1024=-1024

a ( x r 1 ) ( x r 2 ) . . . ( x r d ) = 1024 = 2 11 a(x-r_1)(x-r_2)...(x-r_d)=-1024=-2^{11} ($)

Since there are already d solution for the first equation, so to fulfill the problem's condition that we want more than d solutions, we hope for at least one integer solution for ($).

Since the LHS of ($) are all integers, therefore they must all be factors of and multiply to get 2 11 -2^{11} . Since we want to maximize d, we have to find the largest number d such that 2 11 -2^{11} can be expressed as the product of d distinct integers(in doing so we'll set a = 1 a=1 ). To answer this sub-question, one can easily come up with and check these 7 integers : S = 1 , 1 , 2 , 2 , 2 2 , 2 2 , 2 5 S=1,-1,2,-2,2^2,-2^2,2^5 . This means that one of the x r i x-r_i corresponds with one of the factors. In other words, we have 7 equations: x r i = p i x-r_i=p_i where p i p_i is an element of S S and i = 1 , 2 , . . . , 7 i=1,2,...,7 .

Since there are 7 equations with 8 variables, there's at most one solution for x if one r i r_i is numerically defined. Therefore, we can find r 1 , r 2 , . . . , r 7 r_1,r_2,...,r_7 to make the second equation to have one integer solution(i.e. they are a permutation of S S and x = 0 x=0 ). Hence this suggests that the answer should be 7 7 .

At this point, I've summarized how I arrived at the solution. To ensure that 7 is indeed the largest integer, let's assume that d 8 d\ge 8 . Since it's shown above that we can only have most 7 distinct factors on the LHS of ($), there are d 7 1 d-7\ge 1 factors that will reoccur. Thus if we list out the 7 distinct equations again, by the same argument shown above, we will get at most 1 solution for x. This makes the total number of distinct solutions at most 7 + 1 = 8 7+1=8 , which doesn't fulfill the requirements for d 8 d\ge 8 .

As predicted...this is indeed similar to the above solution.. Needs check for rigorousness!

Xuming Liang - 7 years, 6 months ago

An addition/typo to the proof: it should be 7 LINEAR equations.

Xuming Liang - 7 years, 6 months ago

Now that I thought about it....the whole thing seems to be a mistake! What I said wasn't enough for the number of solutions to x.

Xuming Liang - 7 years, 6 months ago

I'm not so sure what your last paragraph mean. There could be d d solutions to f ( x ) = 1024 f(x) = 1024 and d d solutions to f ( x ) = 1024 f(x) = - 1024 . I do not see any argument for why given d 8 d\geq 8 , there cannot be more than d d integer solutions in total.

Note that you're also making an assumption that the polynomial must be completely factored into integers, which need not be the case. We could have quadratic terms like x 2 + 2 x + 3 x^2 + 2x + 3 which doesn't allow us to factorize it into ( x r 1 ) ( x r 2 ) (x-r_1)(x-r_2) , where r 1 , 2 r_1, _2 are integers.

Calvin Lin Staff - 7 years, 6 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...