1
$\begingroup$

How do we prove that there is always a unique parabola (with equation $y=ax^2+bx+c$) that passes through 3 distinct points $P_1 (p_1,q_1), P_2 (p_2,q_2), P_3(p_3,q_3)$ ?

If I choose to use matrices and row reduction, how do I deal with the fact that the coordinates might be zero?

$\endgroup$
  • $\begingroup$ First prove that there is a parabola that passes through $(p_1,q_1)$, $(p_2,0)$, and $(p_3,0)$. (Once you have that, you also know there's one through $(p_1,0)$, $(p_2,q_2)$, and $(p_3,0)$ by relabeling, and similarly for the last coordinate; then you can just add the equations together.) $\endgroup$ – Akiva Weinberger Aug 31 '15 at 1:57
  • $\begingroup$ The statement is only true if you assume your three points are not colinear. Otherwise, consider the points $(0,0),(-1,0),(1,0)$. Clearly there is no such parabola as that would be a quadratic equation with 3 roots. (Unless you consider a straight line to be a degenerate parabola, which generally it isn't :)) $\endgroup$ – Alan Aug 31 '15 at 1:58
  • 1
    $\begingroup$ @Alan Technically, that's $y=0x^2+0x+0$, though I see your point. $\endgroup$ – Akiva Weinberger Aug 31 '15 at 1:59
  • $\begingroup$ OP, you should prove that there is no (proper) parabola through $(-1,0)$, $(0,0)$, and $(1,0)$, by the way. $\endgroup$ – Akiva Weinberger Aug 31 '15 at 2:00
  • $\begingroup$ Not is unique frodriguezdiaz.blogspot.mx/2013/11/… $\endgroup$ – jimbo Aug 31 '15 at 2:13
1
$\begingroup$

Proving that there is a parabola should be no hassle. To show uniqueness, suppose $y = a_1x^2 + b_1x + c_1$ and $y = a_2x^2 + b_2x + c_2$ fit the points. Then the parabola $y = (a_2 - a_1)x^2 + (b_2 - b_1)x + (c_2 - c_1)$ goes through the points $(p_1,0),(p_2,0),(p_3,0)$. This implies the quadratic has three roots, which is only possible if $a_1 = a_2$, $b_1 = b_2$, and $c_1 = c_2$.

$\endgroup$
2
$\begingroup$

You have a set of points $(p_i,q_i)$ for $i=1,2,3$ with $p_i\neq p_j$ for $i\neq j$.

You can prove existence of a solution by deriving the determinant of a Vandermonde matrix of size 3: $\prod_{1\leq i<j\leq 3} (p_i-p_j)$ and seeing that it is not singular from your assumptions above.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.