Quadrilateral in a Cubic Polynomial

Geometry Level 5

A monic polynomial of degree 3 has three unique integer roots, P,Q and R between 0 and 36 inclusive. A quadrilateral BFEG has been constructed as follows: F and G are the two extrema and I is the inflection point. The blue line is normal to the curve at I. This normal intersects the polynomial in two additional places, B and E.

If the minimum area of the quadrilateral can be expressed as a c b \frac{a}{c}\sqrt{b} , where a and c are coprime and b is square free, then provide the sum a + b + c a + b + c .


The answer is 25.

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.

4 solutions

Mark Hennings
Jun 1, 2020

Since a cubic has order 2 2 rotational symmetry about its point of inflection, the quadrilateral is a parallelogram, and hence has area B F × B G \left| \overrightarrow{BF} \times \overrightarrow{BG} \right| Suppose that the polynomial has distinct integer roots P < Q < R P < Q < R , and hence is ( X P ) ( X Q ) ( X R ) (X-P)(X-Q)(X-R) . It is easy to evaluate the coordinates of F F and G G , and to solve the equation of the cubic simultaneously with that of the normal to the cubic at I I (where x = 1 3 ( P + Q + R ) x = \tfrac13(P+Q+R) ) to find the coordinates of B B - both of these calculations simply require us to solve quadratics. After much algebra and simplification, we obtain the surprising result that the area of the quadrilateral is A ( X ) = 2 ( 27 + 2 X 2 ) 9 + X 2 27 3 X where X = P 2 + Q 2 + R 2 P Q P R Q R A(X) \; =\; \frac{2(27 + 2X^2)\sqrt{9 + X^2}}{27\sqrt{3}X} \hspace{1cm}\mbox{where} \hspace{1cm} X = P^2 + Q^2 + R^2 - PQ - PR - QR We note that 2 X = ( Q P ) 2 + ( R Q ) 2 + ( R P ) 2 2X = (Q-P)^2 + (R-Q)^2 + (R-P)^2 ; since Q P , R Q 1 Q-P,R-Q \ge 1 and R P 2 R-P \ge 2 we see that X 3 X \ge 3 for all allowed values of P , Q , R P,Q,R . Moreover d A d X = 2 ( 4 X 4 + 18 X 2 243 ) 27 3 X 2 9 + X 2 \frac{dA}{dX} \; = \; \frac{2(4X^4 + 18X^2 - 243)}{27\sqrt{3}X^2\sqrt{9+X^2}} and hence d A d X > 0 \frac{dA}{dX} > 0 for all X 3 X \ge 3 . Thus the smallest possible value of A A occurs when X = 3 X=3 , which happens when P , Q , R P,Q,R are consecutive integers. The minimum value of A A is thus 10 3 2 3 = 10 9 6 \tfrac{10}{3}\sqrt{\tfrac23} = \tfrac{10}{9}\sqrt{6} , making the answer 10 + 9 + 6 = 25 10 + 9 + 6 = \boxed{25} .

NIce solution! A slip of the pen: x = 1 3 ( P + Q + R ) x = \frac{1}{3}\left( {P + Q + R} \right) .

Thanos Petropoulos - 11 months, 3 weeks ago

Log in to reply

Thank you...

Mark Hennings - 11 months, 3 weeks ago
Piotr Idzik
May 27, 2020

Here is my computer assisted solution heavily relying on sympy module for python.

I consider a polynomial f L , R ( x ) = ( L + x ) x ( x R ) , f_{L, R}(x) = \left(L + x\right) x \left(x - R\right), where L L and R R are some natural numbers (i.e. I shift the polynomial from the problem in such a way, that 0 0 is its middle root).

After few calculations (cf. code below) it turns out, that the area of the quadrilateral of our interest expressed as a function of L L and R R has the form: A ( L , R ) = 2 3 L 4 + 2 L 3 R + 3 L 2 R 2 + 2 L R 3 + R 4 + 9 ( 2 L 4 + 4 L 3 R + 6 L 2 R 2 + 4 L R 3 + 2 R 4 + 27 ) 81 ( L 2 + L R + R 2 ) . A(L, R) = \frac{2 \sqrt{3} \sqrt{L^{4} + 2 L^{3} R + 3 L^{2} R^{2} + 2 L R^{3} + R^{4} + 9} \left(2 L^{4} + 4 L^{3} R + 6 L^{2} R^{2} + 4 L R^{3} + 2 R^{4} + 27\right)}{81 \left(L^{2} + L R + R^{2}\right)}. It attains its minimum for L = R = 1 L = R = 1 . The minimal value is 10 6 9 \frac{10 \sqrt{6}}{9} , which gives the answer 25.

And the code:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
"""
solution of:
    https://brilliant.org/problems/quadrilateral-in-a-cubic-polynomial/
"""
import itertools
import sympy
def get_area_expr():
    """
    returns an expression for the area of the quadrilateral described in:
        https://brilliant.org/problems/quadrilateral-in-a-cubic-polynomial/
    """
    x_sym = sympy.symbols('x', real=True)
    l_sym = sympy.symbols('L', positive=True)
    r_sym = sympy.symbols('R', positive=True)
    poly_fun = (x_sym+l_sym)*x_sym*(x_sym-r_sym)
    def get_point(in_x):
        """
        returns the coordinates of a point on the graph of the function poly_fun for given in_x
        """
        return sympy.Point2D(in_x, sympy.simplify(poly_fun.subs(x_sym, in_x)))

    #we determine point_f and point_g
    poly_der = sympy.simplify(sympy.diff(poly_fun, x_sym))
    tmp_sol_data = list(sympy.solveset(poly_der, x_sym))
    assert len(tmp_sol_data) == 2
    assert tmp_sol_data[0] < tmp_sol_data[1]
    point_f = get_point(tmp_sol_data[0])
    point_g = get_point(tmp_sol_data[1])

    #we determine point_i
    poly_dder = sympy.simplify(sympy.diff(poly_der, x_sym))
    tmp_sol_list = list(sympy.solveset(poly_dder, x_sym))
    point_i = get_point(tmp_sol_list[0])

    #we determine the slope of the straigth line passing through point_i
    slope_i = poly_der.subs(x_sym, point_i[0])
    line_i = sympy.simplify(-1/slope_i*(x_sym-point_i[0])+point_i[1])

    #we determine point_b and point_e
    tmp_sol_data = sympy.solveset(poly_fun-line_i, x_sym)
    assert point_i[0] in tmp_sol_data
    assert len(tmp_sol_data) == 3
    tmp_sol_data = list(tmp_sol_data-sympy.FiniteSet(point_i[0]))
    tmp_sol_data = [sympy.simplify(_) for _ in tmp_sol_data]
    assert tmp_sol_data[0] < tmp_sol_data[1]
    point_b = get_point(tmp_sol_data[0])
    point_e = get_point(tmp_sol_data[1])

    poly_data = sympy.geometry.polygon.Polygon(point_b, point_g, point_e, point_f)
    return sympy.simplify(poly_data.area), l_sym, r_sym

AREA_EXPR, L_SYM, R_SYM = get_area_expr()
MIN_VAL = sympy.oo

for (l, r) in itertools.product(range(1, 37), repeat=2):
    cur_val = AREA_EXPR.subs([(L_SYM, l), (R_SYM, r)])
    if cur_val < MIN_VAL:
        MIN_VAL = cur_val
        MIN_ARG = (l, r)

print(f"Minimal area: {MIN_VAL} for l={MIN_ARG[0]}, r={MIN_ARG[1]}")

Nice use of sympy.

Fletcher Mattox - 1 year ago

As Mark Hennings has already pointed out in his solution, the cubic has order 2 rotational symmetry about its point of inflection I Ι , hence, the quadrilateral is a parallelogram. Working in 2D, its area A A can be given by A = det ( F E , F B ) A = \left| {\det \left( {\overrightarrow {FE} ,\overrightarrow {FB} } \right)} \right| .

We notice that a translation of the cubic by vector I O \overrightarrow {IO} , ( O O is the origin) does not affect the compound shape of the cubic and the parallelogram, but it simplifies quite a bit the algebraic manipulations needed.

Due to the symmetry, the zeros of the translated cubic g g are 0 0 , ± c \pm c for some positive real number c c .
Hence, g ( x ) = x ( x 2 c 2 ) = x 3 c 2 x g\left( x \right) = x\left( {{x^2} - {c^2}} \right) = {x^3} - {c^2}x .
Consequently, g ( x ) = 3 x 2 c 2 g ( 0 ) = c 2 slope of B 1 E 1 = 1 c 2 equation of B 1 G 1 : y = 1 c 2 x g'\left( x \right) = 3{x^2} - {c^2} \Rightarrow g'\left( 0 \right) = - {c^2} \Rightarrow {\text{slope of }}{{\text{B}}_1}{{\text{E}}_1}{\text{ = }}\frac{1}{{{c^2}}} \Rightarrow {\text{ equation of }}{{\text{B}}_1}{{\text{G}}_1}{\text{: }}y = \frac{1}{{{c^2}}}x .
To find points E 1 {E_1} and B 1 {B_1} , we solve simultaneously { y = 1 c 2 x y = x 3 c 2 x \left\{ {\begin{matrix} {y = \frac{1}{{{c^2}}}x} \\ {y = {x^3} - {c^2}x} \end{matrix}} \right.
and we easily get E 1 ( c 4 + 1 c , c 4 + 1 c 3 ) {E_1}\left( {\frac{{\sqrt {{c^4} + 1} }}{c},\;\frac{{\sqrt {{c^4} + 1} }}{{{c^3}}}} \right) , B 1 ( c 4 + 1 c , c 4 + 1 c 3 ) {B_1}\left( { - \frac{{\sqrt {{c^4} + 1} }}{c},\; - \frac{{\sqrt {{c^4} + 1} }}{{{c^3}}}} \right) .



For the point F 1 F_1 ,
g ( x ) = 0 3 x 2 c 2 = 0 x < 0 x = c 3 g ( c 3 ) = ( c 3 ) 3 c 2 ( c 3 ) = 2 c 3 3 3 } F 1 ( c 3 , 2 c 3 3 3 ) \left. {\begin{matrix} {g'\left( x \right) = 0 \Leftrightarrow 3{x^2} - {c^2} = 0\mathop \Leftrightarrow \limits^{x < 0} x = - \frac{c}{{\sqrt 3 }}} \\ {g\left( { - \frac{c}{{\sqrt 3 }}} \right) = {{\left( { - \frac{c}{{\sqrt 3 }}} \right)}^3} - {c^2}\left( { - \frac{c}{{\sqrt 3 }}} \right) = \frac{{2{c^3}}}{{3\sqrt 3 }}} \end{matrix}} \right\} \Rightarrow {F_1}\left( { - \frac{c}{{\sqrt 3 }},\;\frac{{2{c^3}}}{{3\sqrt 3 }}} \right)

Thus, F 1 E 1 = ( c 4 + 1 c + c 3 , c 4 + 1 c 3 2 c 3 3 3 ) \overrightarrow {{F_1}{E_1}} = \left( {\frac{{\sqrt {{c^4} + 1} }}{c} + \frac{c}{{\sqrt 3 }},\frac{{\sqrt {{c^4} + 1} }}{{{c^3}}} - \frac{{2{c^3}}}{{3\sqrt 3 }}} \right) and F 1 B 1 = ( c 4 + 1 c + c 3 , c 4 + 1 c 3 2 c 3 3 3 ) \overrightarrow {{F_1}{B_1}} = \left( { - \frac{{\sqrt {{c^4} + 1} }}{c} + \frac{c}{{\sqrt 3 }}, - \frac{{\sqrt {{c^4} + 1} }}{{{c^3}}} - \frac{{2{c^3}}}{{3\sqrt 3 }}} \right) .

Finally, A = det ( F E , F B ) = det ( F 1 E 1 , F 1 B 1 ) = 2 3 c 4 + 1 ( 2 c 2 3 + 1 c 2 ) A = \left| {\det \left( {\overrightarrow {FE} ,\overrightarrow {FB} } \right)} \right| = \left| {\det \left( {\overrightarrow {{F_1}{E_1}} ,\overrightarrow {{F_1}{B_1}} } \right)} \right| = \frac{2}{\sqrt 3}\sqrt {{c^4} + 1} \left( {\frac{{2{c^2}}}{3} + \frac{1}{{{c^2}}}} \right) .

Now, due to the concavity of the cubic, the distance between P P and R R cannot exceed the length of segment M N MN , which is parallel to the x-axis, through the point of inflection I I .
Hence, 2 R P M N = L K = 2 c 2 \leqslant R - P \leqslant \overline {MN} = \overline {LK} = 2c . Thus, c 1 c \geqslant 1 .

If we label c 2 = x {c^2} = x , we have a function for the area of the parallelogram: A ( x ) = 2 3 x 2 + 1 ( 2 x 3 + 1 x ) , x 1 A\left( x \right) = \frac{2}{{\sqrt 3 }}\sqrt {{x^2} + 1} \left( {\frac{{2x}}{3} + \frac{1}{x}} \right), x \geqslant 1 . For x 1 x \geqslant 1 , A ( x ) = 2 ( 4 x 4 + 2 x 2 3 ) 3 3 x 2 x 2 + 1 > 0 A'\left( x \right) = \frac{{2\left( {4{x^4} + 2{x^2} - 3} \right)}}{{3\sqrt 3 {x^2}\sqrt {{x^2} + 1} }} > 0 . Thus, A ( x ) A\left( x \right) is increasing, so A min = A ( 1 ) {A_{\min }} = A\left( 1 \right) , which occurs when c = 1 2 = 1 c = \sqrt {{1^2}} = 1 .

Since 2 R P 2 c = 2 2 \leqslant R - P \leqslant 2c = 2 , it turns out that the zeros P P , Q Q , R R of the initial cubic are three consecutive integers, i.e. ( P , Q , R ) { ( 0 , 1 , 2 ) , ( 1 , 2 , 3 ) , , ( 34 , 35 , 36 ) } \left( {P,Q,R} \right) \in \left\{ {\left( {0,1,2} \right),{\text{ }}\left( {1,2,3} \right),{\text{ }} \ldots ,{\text{ }}\left( {34,35,36} \right)} \right\} No matter which triad is the actual zeros of the cubic, the minimum area of the parallelogram is A ( 1 ) = 10 6 9 A\left( 1 \right) = \frac{{10\sqrt 6 }}{9} .

For the answer, a = 10 a=10 , b = 6 b=6 , c = 9 c=9 , hence, a + b + c = 25 a+b+c=\boxed{25} .

the area asked will be minimum when the three zeros of the polynomial are the closest. So, let us choose the polynomial y = x ( x 1 ) ( x 2 ) = x 3 3 x 2 + 2 x y=x(x-1)(x-2)=x^3-3x^2+2x .

So the coordinates of I I are ( 1 , 0 ) (1,0) obtained from the condition d 2 y d x 2 = 0 \dfrac{d^2y}{dx^2}=0 and the equation of the curve.

Coordinates of F F and G G are ( 3 3 3 , 2 3 9 ) (\dfrac{3-\sqrt 3}{3},\dfrac{2\sqrt 3}{9})

and ( 3 + 3 3 , 2 3 9 ) (\dfrac{3+\sqrt 3}{3},\dfrac{-2\sqrt 3}{9}) respectively, obtained from the condition d y d x = 0 \dfrac{dy}{dx}=0 and the equation of the curve.

Slope of B E \overline {BE} is 1 1 , and it's equation is y = x 1 y=x-1 . Solving this and the equation of the curve we get the coordinates of B B and E E as

( 1 2 , 2 ) (1-\sqrt 2,-\sqrt 2 ) and ( 1 + 2 , 2 ) (1+\sqrt 2, \sqrt 2) .

Hence the required area is 10 6 9 \dfrac{10\sqrt 6}{9} , a = 10 , b = 6 , c = 9 a=10,b=6,c=9 and a + b + c = 10 + 6 + 9 = 25 a+b+c=10+6+9=\boxed {25} .

I have the same solution. But "the area asked will be minimum when the three zeros of the polynomial are the closest" it is open question. I try оther a,b,m for polynom mx(x-a)(x-b) and see that minimum area for m=1, a=1, b=2.

Yuriy Kazakov - 1 year ago

Log in to reply

So the proposition is justified. Isn't it? The area is minimum when the three zeros are closest!

Log in to reply

Yes minimum when the three zeros are closest - but I have no rigorous justification for it.

Yuriy Kazakov - 1 year ago

Log in to reply

@Yuriy Kazakov See my proof for why the area is minimized when the roots are consecutive.

Mark Hennings - 1 year ago

Log in to reply

@Mark Hennings Great, thanks!

Yuriy Kazakov - 1 year ago

What about a coefficient in front of your polynomial? When considering y = a x ( x 1 ) ( x 2 ) y = ax(x - 1)(x - 2) , there are smaller values for the area than A = 10 6 9 2.722 A = \frac{10\sqrt{6}}{9} \approx 2.722 for different values of a a . The smallest I found was when a = 3 + 57 2 a = \frac{\sqrt{3 + \sqrt{57}}}{2} , which leads to an area of A = 1 18 858 + 114 57 2.303 A = \frac{1}{18}\sqrt{858 + 114 \sqrt{57}} \approx 2.303 . I think the question is flawed, and should include that the polynomial must be a monic polynomial (where the leading coefficient is 1 1 ).

David Vreken - 1 year ago

Log in to reply

Quite right. I've edited it. Thank you!

Fletcher Mattox - 1 year ago

Log in to reply

Great, thanks!

David Vreken - 1 year ago

Fine rezult.

Yuriy Kazakov - 1 year ago

I've just come and so sorry for not answering you. Yes. That is the truth. Because of excitement, I missed the constant factor :).

Is this college mathematics?

Joshua Olayanju - 1 year ago

Log in to reply

Yes, I would say so. This problem would be difficult to do without using derivatives, which is taught in Calculus. Calculus is a college class, although some advanced students do take Calculus in their last years of high school.

David Vreken - 1 year ago

Ok , that is why i couldn't solve this

Joshua Olayanju - 1 year ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...