Polynomial Inequalities

Algebra Level 5

Suppose f ( x ) f(x) and g ( x ) g(x) are non-constant polynomials with integer coefficients, such that f f is monic and f ( g ( x ) ) = ( f ( x ) ) 2 g ( x ) . f(g(x))=(f(x))^2\cdot g(x). Suppose N N is the number of possible polynomials g , g, such that all coefficients of g g have absolute value strictly less than 1 0 100 10^{100} . Find the last three digits of N . N.

Details and assumptions

A polynomial is monic if its leading coefficient is 1. For example, the polynomial x 3 + 3 x 5 x^3 + 3x - 5 is monic but the polynomial x 4 + 2 x 3 6 -x^4 + 2x^3 - 6 is not.


The answer is 997.

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

We first let a a be the degree of f ( x ) f(x) and b b be the degree of g ( x ) g(x) . The given equation tells us that: a b = 2 a + b a b = 2 a + b b = 2 + b a b = 2 + \frac{b}{a}

Since both a a and b b are positive integers, this means that b a \frac{b}{a} is an integer, and that a a is a factor of b b . The last equation also tells us that the difference between b b and b a \frac{b}{a} (which is another factor of b b ) is 2, so b b cannot be greater than 4. (No integers above 4 are 2 more than one of their factors, as the quotient would be under 2 but above 1.)

This results in two cases: a = 2 a = 2 and b = 4 b = 4 or a = 3 a = 3 and b = 3 b = 3 . Before examining each case, we will make the general statement that f ( x ) f(x) has no constant term. This is evident if we let s s be a root of g ( x ) g(x) . (Since the polynomial equality holds on the real line, it holds in the complex plane [according to elementary complex analysis], and the fundamental theorem of algebra [also easily shown from elementary complex analysis] guarantees that g ( x ) g(x) has a root in the complex plane.) At this root, f ( s ) 2 g ( s ) = f ( s ) 2 ( 0 ) = 0 f(s)^2 g(s) = f(s)^2 (0) = 0 , so f ( g ( s ) ) = f ( 0 ) = 0 f(g(s)) = f(0) = 0 . This prevents f ( x ) f(x) from having a constant term in either case. Now we can examine each case:

  • Case I: a = 2 a = 2 and b = 4 b = 4 . In this case, f ( x ) f(x) is a monic quadratic. As mentioned above, it has no constant term, so it may be written f ( x ) = x ( x c ) f(x) = x (x - c) , for some integer c c . This gives: f ( g ( x ) ) = g ( x ) ( g ( x ) c ) ) = ( f ( x ) 2 ) g ( x ) = ( x ( x c ) ) 2 g ( x ) f(g(x)) = g(x) (g(x) - c)) = (f(x)^2) g(x) = (x (x - c))^2 g(x) The analyticity of polynomials allows us to divide through by g ( x ) g(x) (note that the equality must hold for the entire real line and therefore complex plane, so we need not worry about g ( x ) g(x) being zero), and we obtain: g ( x ) c = ( x ( x c ) ) 2 g(x) - c = (x (x - c))^2 g ( x ) = x 4 c x 3 + c 2 x 2 + c g(x) = x^4 - c x^3 + c^2 x^2 + c The equality and restrictions on f ( x ) f(x) mean that there is only one degree of freedom in finding g ( x ) g(x) . To keep the largest coefficient of g ( x ) g(x) under 1 0 100 10^{100} , we must ensure that c 2 < 1 0 100 c^2 < 10^{100} , or that c < 1 0 50 |c| < 10^{50} . This gives us 2 ( 1 0 50 1 ) + 1 = 2 1 0 50 1 2 (10^{50} - 1) + 1 = 2 \cdot 10^{50} - 1 for the number of possible g ( x ) g(x) in this case.

  • Case II: a = 3 a = 3 and b = 3 b = 3 . As mentioned above, f ( x ) f(x) cannot have a constant term, so it must be of the form: f ( x ) = x ( x c 1 ) ( x c 2 ) f(x) = x (x - c_1) (x - c_2) . Note that c 1 c 2 c_1 c_2 and c 1 + c 2 c_1 + c_2 must be integers, but c 1 c_1 and c 2 c_2 alone need not be (for the moment). Again, we can plug this in: f ( g ( x ) ) = g ( x ) ( g ( x ) c 1 ) ( g ( x ) c 2 ) f(g(x)) = g(x) (g(x) - c_1) (g(x) - c_2) = ( f ( x ) 2 ) g ( x ) = ( x ( x c 1 ) ( x c 2 ) ) 2 g ( x ) = (f(x)^2) g(x) = ( x(x - c_1)(x - c_2) )^2 g(x) Dividing, as we could previously: ( g ( x ) c 1 ) ( g ( x ) c 2 ) = ( x ( x c 1 ) ( x c 2 ) ) 2 (g(x) - c_1)(g(x) - c_2) = ( x(x - c_1)(x - c_2) )^2 At this juncture, one may use the quadratic formula to find g ( x ) g(x) , but one may circumvent that by noting that both sides are polynomials in x x and the right side is a perfect square, so the left side must be one as well. Either way, one arrives at the crucial breakthrough, which is that g ( x ) g(x) is only a valid polynomial when c 1 = c 2 = c c_1 = c_2 = c : ( g ( x ) c ) 2 = ( x ( x c ) 2 ) 2 (g(x) - c)^2 = ( x (x - c)^2) ^2 Equal squares do not mean equal values; they mean equal or opposite values: g ( x ) c = ± ( x ( x c ) 2 ) g(x) - c = \pm ( x (x - c)^2 ) g ( x ) = c ± ( x 3 2 c x 2 + c 2 x ) g(x) = c \pm ( x^3 - 2 c x^2 + c^2 x ) Again, we find that we only need to set c 2 < 1 0 100 c^2 < 10^{100} to ensure that the magnitudes of g ( x ) g(x) 's coefficients lie strictly under 1 0 100 10^{100} . Again, we have c < 1 0 50 |c| < 10^{50} . In this case, however, all values of c c yield 2 different solutions for g ( x ) g(x) : one with the plus sign, and one with the minus sign. (Even c = 0 c = 0 results in different solutions.) Thus, instead of 2 1 0 50 1 2 \cdot 10^{50} - 1 solutions, we have twice that, or 2 ( 2 1 0 50 1 ) = 4 1 0 50 2 2 (2 \cdot 10^{50} - 1) = 4 \cdot 10^{50} - 2 solutions.

In total, therefore, we have 6 1 0 50 3 6 \cdot 10^{50} - 3 solutions. This number is the difference of 6 1 0 50 6 \cdot 10^{50} , or 6 followed by lots of zeros, and 3. Thus, the difference ends in lots of nines followed by a 7. The last three digits of the number of solutions is 997 \boxed{997} , the correct answer.

Nice!

Zi Song Yeoh - 7 years, 7 months ago

Log in to reply

Great job, Alexander!

Alexander Borisov - 7 years, 7 months ago

Log in to reply

Am I the only one to be amused by another case of dopplegangers?

I guess we should make it easier to differentiate people.

Calvin Lin Staff - 7 years, 7 months ago

Log in to reply

@Calvin Lin You are not the only one :)

Alexander Borisov - 7 years, 7 months ago
Mark Hennings
Oct 28, 2013

If m m is the degree of f f , and n n the degree of g g , then m n = 2 m + n mn = 2m+n , so that ( m 1 ) ( n 2 ) = 2 (m-1)(n-2) = 2 . Since m , n 1 m,n \ge 1 we deduce that either m = 2 , n = 4 m=2,n=4 , or else m = n = 3 m=n=3 .

  1. If m = 2 , n = 4 m=2,n=4 , then f ( X ) = X 2 + a X + b f(X) = X^2 + aX + b , where g ( X ) 2 + a g ( X ) + b = f ( X ) 2 g ( X ) g(X)^2 + ag(X) + b \; = \; f(X)^2g(X) so g ( X ) g(X) divides b b , and hence b = 0 b=0 . Thus we deduce that g ( X ) = f ( X ) 2 a = X 4 + 2 a X 3 + a 2 X 2 a g(X) \; = \; f(X)^2 - a \; = \; X^4 + 2aX^3 + a^2X^2 - a and we want 2 a , a 2 , a |2a|,|a^2|,|-a| all to be less than 1 0 100 10^{100} . Thus a < 1 0 50 |a| < 10^{50} . There are 2 × 1 0 50 1 2\times10^{50}-1 polynomials of this type.

  2. If m = n = 3 m=n=3 then f ( X ) = x 3 + a X 2 + b X + c f(X) = x^3 + aX^2 + bX + c where g ( X ) 3 + a g ( X ) 2 + b g ( X ) + c = f ( X ) 2 g ( X ) g(X)^3 + ag(X)^2 + bg(X) + c \; = \; f(X)^2g(X) so g ( X ) g(X) divides c c , and hence c = 0 c=0 , and so g ( X ) 2 + a g ( X ) + b = f ( X ) 2 g(X)^2 + ag(X) + b \; = \; f(X)^2 Comparing coefficients of X 6 X^6 , we see that either g ( X ) g(X) is monic or g ( X ) -g(X) is monic. Moreover a g ( X ) + b = ( f ( X ) g ( X ) ) ( f ( X ) + g ( X ) ) ag(X) + b \; = \; (f(X) - g(X))(f(X) + g(X)) If g ( X ) g(X) is monic then f ( X ) + g ( X ) f(X)+g(X) has degree 3 3 and leading coefficient 2 2 , and a g ( X ) + b ag(X) + b is at most cubic. Thus we deduce that f ( X ) g ( X ) f(X) - g(X) is constant, and hence g ( X ) = f ( X ) + d g(X) = f(X) + d for some constant d d . Comparing powers of g ( X ) g(X) in the above identity, we deduce that a = 2 d , b = d 2 a=-2d, b=d^2 , so that f ( X ) = X 3 2 d X 2 + d 2 X f(X) = X^3 - 2dX^2 + d^2X , and hence g ( X ) = X 3 2 d X 2 + d 2 X d g(X) = X^3 - 2dX^2 + d^2X - d . Since we want 2 d , d 2 , d |-2d|,|d^2|,|-d| all to be less than 1 0 100 10^{100} , we require d < 1 0 50 |d| < 10^{50} . Thus there are 2 × 1 0 50 1 2\times10^{50}-1 polynomials g ( X ) g(X) of this type. Similarly, if we assume that g ( X ) -g(X) is monic, then we deduce that f ( X ) = X 3 + 2 d X 2 + d 2 X f(X) = X^3 + 2dX^2 + d^2X and g ( X ) = X 3 2 d X 2 d 2 X d g(X) = -X^3 - 2dX^2 - d^2X - d for some d < 1 0 50 |d| < 10^{50} , and so we obtain another 2 × 1 0 50 1 2\times10^{50}-1 polynomials.

Thus there are N = 6 × 1 0 50 3 N = 6\times10^{50}-3 polynomials g ( X ) g(X) of the desired type, and the last three digits of N N are 997 997 .

Nicely done once again!

Alexander Borisov - 7 years, 7 months ago

I mostly did it the same as you, but here's one difference that might interest you: in part (2), when I got to g ( X ) 2 + a g ( X ) + b = f ( X ) 2 g(X)^2+ag(X)+b = f(X)^2 I instead wrote it as:

( g ( X ) h ) 2 + k = f ( X ) 2 (g(X)-h)^2+k =f(X)^2

which means that

k = ( f ( X ) + g ( X ) h ) ( f ( X ) g ( X ) + h ) k= (f(X) +g(X)-h) (f(X) -g(X)+h)

But ( f ( X ) + g ( X ) h ) (f(X) +g(X)-h) and ( f ( X ) g ( X ) + h ) (f(X) -g(X)+h) cannot both be constant, so one of them is zero (and so is k k ). So

g ( X ) = h ± f ( X ) = h ± X ( X h ) 2 g(X) =h\pm f(X) =h\pm X(X-h)^2

and h h has to be an integer.

And so on ...

Peter Byers - 7 years, 7 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...