1

If $\theta_1$, $\theta_2$, and $\theta_3$ are the eccentric angles of three points on the hyperbola $x^2/a^2 - y^2/b^2 = 1$ such that $\sin (\theta_1+\theta_2) + \sin (\theta_2+\theta_3) + \sin (\theta_3+\theta_1) = 0$ then prove that the normals at these points are concurrent.

P.S. My attempt was to use the same idea as for an ellipse here, but then I realised that it wouldn't do for quite obvious reasons.

Edit 1: As a clarification, do NOT use the hyperbola $xy = c^2$ to answer this question. You may use ONLY the hyperbola I mentioned above.

Edit 2: Here are my efforts so far.... Assume that the equation of normal in parametric form is

$$\frac{ax}{\sec \theta} + \frac{by}{\tan \theta} = a^2 + b^2$$

Then, taking $\tan \frac{\theta}{2} = t$ and using the familiar trigonometric identities gives a biquadratic equation in $t$ as:

$$ bkt^4 + 2(a^2 + b^2 + ah)t^3 + 2(a^2 + b^2 - ah)t - bk = 0$$

where we have substituted $(h,k)$ for $(x,y)$ as the point of intersection of the four normals at $\theta_i$ ($i= 1,2,3,4$) as the roots of the above equation.

Observe that this gives

$$\theta_1 + \theta_2 + \theta_3 + \theta_4 = 2n\pi, n \in \mathbb Z$$ but it doesn't get me near the condition at all. I cannot 'eliminate' one root to get a cubic having three roots. I need some help, particularly in eliminating one root. Any efforts towards this could be welcome.

Other attempts I tried include complex numbers and using the fact that if we have a repeated root, then the equation will have its slope zero at this particular root.

Edit 3: In reply to Blue's comment, the point of intersection comes out to be

$$P \equiv \left( \frac{(a^2 + b^2)\sec \theta_1 \sec \theta_2 \cos \left( \frac{\theta_1-\theta_2}{2} \right)} {a\cos \left( \frac{\theta_1 + \theta_2}{2} \right)}, -\frac{(a^2 + b^2)\tan \theta_1 \tan \theta_2 \tan \left( \frac{\theta_1 + \theta_2}{2} \right)}{b} \right)$$

On substituting in the third equation, we get

$$ \sin^2 \theta_1 \cos^2 \theta_2 \sin \theta_3 \cos \theta_3- \cos^2 \theta_1 \sin^2 \theta_2 \sin \theta_3 \cos \theta_3+ \\ \quad \sin^2 \theta_1 \sin \theta_2 \cos \theta_2 \sin \theta_3 \cos \theta_3+ \sin \theta_1 \cos^2 \theta_1 \sin ^2 \theta_2 \sin \theta_3- \\ \quad \sin^2 \theta_1 \sin \theta_2 \cos^2 \theta_2 \cos \theta_3- \sin \theta_1 \cos \theta_1 \sin^2 \theta_2 \cos \theta_2 \cos \theta_3 \\ = \sin^2 \theta_1 \cos \theta_1 \cos^2 \theta_2 \cos \theta_3+ \sin \theta_1 \cos^2 \theta_1 \sin \theta_2 \cos \theta_3 \cos \theta_3- \\ \quad \sin \theta_1 \cos \theta_1 \sin \theta_2 \cos^2 \theta_2 \cos \theta_3- \cos^2 \theta_1 \sin^2 \theta_2 \cos \theta_2 \cos \theta_3$$

Now I am unable to simplify. Please note that this has been homogenised by using the fact that $\sin^2 \theta + \cos^2 \theta = 1$. Yet I am unable to get to the hypothesis.

Ng Chung Tak
  • 18,990
  • If $\theta$ is the eccentric angle of a point $P$ on the hyperbola, then $P$ has coordinates $(a\sec\theta,b\tan\theta)$. Can you continue from there? – Blue Mar 11 '20 at 06:47
  • Hey Blue! I used this idea and parameterized as you said but it didn't work with what you did with the ellipse.... If it isn't too much, could you please tell me how we may generate these three co normal points? – Puneet Singh Mar 11 '20 at 07:39
  • Don't bother with what I did with the ellipse; that problem started with different assumptions (namely, the points belonged to some circle). Here, just start with the parameterization by eccentric angles, get the equations for the normals, and determine the condition under which the normals have a common point. – Blue Mar 11 '20 at 09:13
  • Yeah I tried but just wasn't happening. I'm getting a biquadratic equation, I can't 'eliminate' a root because I'm getting four roots but require three only... – Puneet Singh Mar 11 '20 at 12:34
  • Include your work in the question, and someone may be able to give specific advice. This will also help people avoid wasting time duplicating your effort. – Blue Mar 11 '20 at 12:44
  • Done! with some help... – Puneet Singh Mar 11 '20 at 13:35
  • Another way to proceed is "simply" to take the normal equations for $\theta_1$ and $\theta_2$, find their point of intersection, substitute that point into the normal equation for $\theta_3$, and then simplify, simplify, simplify. It's a bit of an algebraic and trigonometric slog, but it works. – Blue Mar 11 '20 at 13:55
  • Hey Blue, please review what I just edited as I cannot add it here. Thanks very much! – Puneet Singh Mar 11 '20 at 15:28
  • Check the equation if it is right, coz as I remember condition for Conormal points should be $\theta_1+\theta_2 +\theta_3 +\theta_4$ should be odd multiple of $\pi$ – Mathsmerizing Mar 11 '20 at 16:20

1 Answers1

2

Start with $$\frac{ax}{\text{sec}\ \theta_1} + \frac{by}{\text{tan}\ \theta_1} = a^2 + b^2$$

$$\frac{ax}{\text{sec}\ \theta_2} + \frac{by}{\text{tan}\ \theta_2} = a^2 + b^2$$

$$\frac{ax}{\text{sec}\ \theta_3} + \frac{by}{\text{tan}\ \theta_3} = a^2 + b^2$$

When are these three straight lines concurrent?

Equate the following determinant to $0$ and simplify!

\begin{vmatrix} \frac{a}{\sec \theta_1} & \frac{b}{\tan \theta_1} & a^2 + b^2 \\ \frac{a}{\sec \theta_2} & \frac{b}{\tan \theta_2} & a^2 + b^2 \\ \frac{a}{\sec \theta_3} & \frac{b}{\tan \theta_3} & a^2 + b^2 \notag \end{vmatrix}

Note:

In other words, you need to equate the following to $0$ and simplify:

\begin{vmatrix} \frac{1}{\sec \theta_1} & \frac{1}{\tan \theta_1} & 1 \\ \frac{1}{\sec \theta_2} & \frac{1}{\tan \theta_2} & 1 \\ \frac{1}{\sec \theta_3} & \frac{1}{\tan \theta_3} & 1 \notag \end{vmatrix}

A few more steps:

$1.$ Multiply $R_1$ by $\sin \theta_1$, $R_2$ by $\sin \theta_2$ and $R_3$ by $\sin \theta_3$

$2.$ Try to show that the above simplifies to $-2 \sin \frac{\theta_1 - \theta_2}{2} \sin \frac{\theta_2 - \theta_3}{2} \sin \frac{\theta_3 - \theta_1}{2}[\sin (\theta_1 + \theta_2) + \sin (\theta_2 + \theta_3) + \sin (\theta_3 + \theta_1)] = 0$

$3.$ Argue that $\sin (\theta_1 + \theta_2) + \sin (\theta_2 + \theta_3) + \sin (\theta_3 + \theta_1) = 0$

In fact, you need to prove the converse of this result. That is also true and you can write the proof "bottom-up"

PTDS
  • 3,464
  • 1
    Hey PTDS! It doesn't give me the condition yet. The most simplification I am getting is $$\text{sin}2\theta_1\ \text{sin}(\theta_2 + \theta_3) + \text{sin}2\theta_2\ \text{sin}(\theta_3 + \theta_1) + \text{sin}2\theta_3\ \text{sin}(\theta_1 - \theta_2) = 0$$ – Puneet Singh Mar 11 '20 at 23:53
  • I have added a few more steps. Let me know if you could follow. – PTDS Mar 12 '20 at 00:34
  • Thanks for your help! I was unable to simplify earlier but this really helped. – Puneet Singh Mar 12 '20 at 00:53