
10-19-2009, 10:24 AM
|
|
|
Heaven is full of goodness and icosahedrons
|
|
|
Re: IA on Quartic Functions
Well I can certainly prove your conjecture, I didn't begin with a general quatric. Make of that what you will...
Any function of a particular "W" shape for which a conjecture might hold can be translated freely without loss of character. It may be translated such that it has inflection points at x=a and x=-a for some a.
Then the second derivative of the quatric is given:
f''(x) = b(x+a)(x-a)
f''(x) = b(x^2-a^2)
Anti-differentiate
f'(x) = b(1/3*x^3 - a^2*x + c)
Anti-differentiate
f(x) = b(1/12*x^4 - a^2/2*x^2 + cx + d)
Solve at the inflection points:
f(a) = b(1/12*(a)^4 - a^2/2*(a)^2 + c(a) + d)
f(-a) = b(1/12*(-a)^4 - a^2/2*(-a)^2 + c(-a) + d)
Solve for the slope (m) of the line connecting the inflection points:
f(a) - f(-a) = 2abc
a - (-a) = 2a
m = bc
Solve for slope/intercept form:
y = mx + i
f(a) = bc(a) + i
b(1/12*(a)^4 - a^2/2*(a)^2 + c(a) + d) = bca + i
i = b/12*(a)^4 - ba^4/2 + bd
i = (-5b/12)(a^4) + bd
y = bcx + (-5b/12)(a^4) + bd
Solve for intersections of line with f(x)
b/12*x^4 - ba^2/2*x^2 + bcx + bd = bcx + (-5b/12)(a^4) + bd
x^4 - 6(a^2)*x^2 = -5(a^4)
x^4 - 6(a^2)*x^2 + 5(a^4) = 0
let w = x^2
w^2 - 6(a^2)w + 5(a^4) = 0
w = 6(a^2)/2 +/- sqrt([-6(a^2)]^2 - 4(5(a^4))]/2
w = 3(a^2) +/- sqrt(36a^4 - 20a^4) / 2
w = 3(a^2) +/- sqrt(16a^4)/2
w = 3(a^2) +/- 2(a^2)
w1 = 5(a^2), w2 = (a^2)
x1 = +sqrt(w1) = sqrt(5)a = Projection of S onto x-axis
x2 = -sqrt(w1) = -sqrt(5)a = Projection of P onto x-axis
x3 = +sqrt(w2) = a = Projection of R onto x-axis
x4 = -sqrt(w2) = -a = Projection of Q onto x-axis
(Note that our original inflection points have fallen out: beauty).
Then there is some constant t such that the distance between points is given:
|PQ| = t|x2x4|
|QR| = t|x4x3|
|RS| = t|x3x1|
|PQ| = ta(sqrt(5) - 1)
|QR| = ta(2)
|RS| = ta(sqrt(5) - 1)
Gives ratios of phi. QED
This proof extends to any Quatric with two distinct real inflection points. We cannot draw a line to intersect a quatric four times if that quatirc has fewer than two distinct real inflection points.
Edit: Made my linear algebra a little more clear (hopefully)...
|