2
$\begingroup$

Problem:

Given the following parabola, construct a tangent to point $P$. Justify the construction.

enter image description here

Solution:

Draw the line $PA$ which is perpendicular to the axis and intersects that axis at $B$. Mark off the distance $OB$ and use it to find point $C$. The line $PC$ is tangent to the parabola.

enter image description here

Basically my only justification that CP is tangent is that it looks really tangent. I have tried justifying my claim but can't seem to get anywhere. I have noticed that a circle can be constructed through points $PAC$. Let point $D$ be where $\circ PAC$ intersects the axis of the parabola. Then $\triangle CPD$ is an inscribed right triangle. This fact seems like it could be useful, but I'm not sure.

  • 1
    I can think of a couple of ways to prove this geometrically, but it depends on what properties you’re allowed to use. For instance, it’s a special case of the [two-points-two-tangents property](https://en.wikipedia.org/wiki/Parabola#2-points-2-tangents-property_of_a_parabola), which is itself a corollary of [Pascal’s theorem](https://en.wikipedia.org/wiki/Pascal%27s_theorem). It looks like one might be able to prove it using the [reflective property](https://en.wikipedia.org/wiki/Parabola#Proof_of_the_reflective_property), too.2017-02-26
  • 1
    Neat construction! Note, though, that it fails at the vertex.2017-02-26
  • 0
    Interesting point, @amd. Say that $P$ is on the vertex. Then we can simply construct the perpendicular to the axis of the parabola, and we have arrived at the tangent. So we arrive at the tangent even faster :)2017-02-26
  • 0
    AP is the polar of C with respect to the parabola.2017-02-27

4 Answers 4

3

Justification: Consider a general parabola $y=ax^2$. Don't need to consider the $bx+x$ term as these terms only describe horizontal and vertical translations but they don't change the parabola's shape. Consider point of tangency $P$ with coordinates $(t,at^2)$ and point $B$ follows $(0,at^2)$. From here your point $C$ is then $(0,-at^2)$. Now the slope of tangent $PC$ can be found with the basic slope formula "rise over run" using the points $P$ and $C$, which then becomes $m=\frac{at^2--at^2}{t-0}$ which simplifies to $2at$. This is the same value as when you take the derivative of $y=ax^2$ and plug in $x=t$. PS, this is a really neat construction, I am thinking of putting this question to my students...

  • 0
    This is a great justification using analytic geometry. Do you know how we might justify it using purely geometrical means?2017-02-26
  • 1
    Yes, I know for sure that this can also be proven using properties of the parabola, meaning using focus and directrix, and no calculus. But....I have forgotten at the moment. Maybe somebody else on the forum can pitch in...:)2017-02-26
3

Your supposition is correct.For parabola parametrized by $ (x,y)= (2 a t, a t^2)$ the slope at any point is $dy/dx=t$. It can be seen that tangent bisects the $x$ abscissa at tangent to vertex .. and the parabola vertex itself bisects (ordinate-central cutting point) line $ BOC.$

3

Let point $C$ be such that $O$ is the midpoint of $BC$ on the parabola's axis.

I propose to take advantage of the interpretation of the parabola as a quadratic Bézier curve (https://www.futurelearn.com/courses/maths-linear-quadratic-relations/0/steps/12129). A basic property of the quadratic Bézier curve defined by $P,C,A$ ($P$ as origin, $A$ as endpoint, and $C$ as control point or "handle") is that the summit $O$ of the arc of parabola it defines is the midpoint of the midpoints of line segments $[CP]$ and $[CA]$.

The essential property of Bézier curve defined by $P,C,A$ in our case is that $CP$ is tangent to the curve in $P$.

The parametric equations of the curve can be given in the following form:

$$\binom{x}{y}=(1-t)^2\binom{x_P}{y_P}+2t(1-t)\binom{x_C}{y_C}+t^2\binom{x_A}{y_A}$$

for $0 \leq t \leq 1$. If $t=0$, we are in $P$. If $t=1$ we are in $A$.

It is important to check that the parabola passes through the midpoint $O$ of $BC$.

This will happen when $t=1/2$. In this case, the current point of the parabola is

$$\binom{x}{y}=\dfrac12\binom{x_C}{y_C}+(\dfrac14\binom{x_P}{y_P}+\dfrac14\binom{x_A}{y_A})=\dfrac12\binom{x_C}{y_C}+\dfrac12\binom{x_B}{y_B}$$

(using rules of barycentrical computations), i.e., the midpoint of line segment $CB$, that is to say point $O$, as desired.

  • 0
    I haven't heard of Bezier curves before this, but they look really interesting. Thank you.2017-02-26
1

EternusVia. Here is another, but NON calculus approach: Consider point of tangency $P(t,at^2)$. Tangent PC passing through P is then of the form $y=mx+at^2-mt $ (verify!).Now intersecting the tangent with the parabola gives us the equation $ax^2=mx+at^2-mt,$ which simplifies to $ax^2-mx-at^2+mt=0.$ This equation is to have only one solution (at $P$) and so the Discriminant $D=b^2-ac,$ of this equation must be zero. This gives $m^2-4a(-at^2+mt)=0$. Simplify gives $m^2-4amt+4a^2t^2=0$. Surprise...surprise, this equation is factorable as a perfect square: $(m-2at)^2=0$ from which $m=2at$ , etc...

  • 0
    I think that the Bezier curve interpretation I propose is equivalent to your presentation.2017-02-26
  • 0
    That's a very cool link. I need to study that when I get a break...2017-02-26