Processing math: 100%

Saturday, May 18, 2024

A nice Geometric Optimisation Problem - Part I

Consider two circles of different radii intersecting at B and D. If we now draw a line KL through B as shown in the following figure, the question of determining the position at which the sum of BK and BL maximised is one of the famous non trivial problems in Geometric Optimisation.

We can solve this by noting that \triangle KDL remains self similar. Because both \angle BKD and \angle BLD are angles subtended by the same chord BD in the respective circles, the angles in the triangle remain constant irrespective of the position of the line KL

 Therefore KL will be maximised when DK (or DL) is as large as possible which happens when they are the respective diameters of the given circles. We could also see that, at this position KL and BD will be perpendicular to each other.

A natural follow up question would be find the position of KL such that the product of BK and BL is maximised. Find a geometric construction of the position along with a geometric proof for the same proved more challenging that I had anticipated. That proof will be the subject of this post.

Before I start, I also want to point out that the proof here may seem unnecessarily complicated. But if anyone can share the shorter and elegant geometric proof for this problem, I would be really grateful.

To begin with, if we construct the circle passing through K, D and L, and extend the segment BD so that it intersects the circumcircle at N, then by the power-of-the-point theorem, we know that

BK \cdot BL = BD \cdot BN

Therefore, maximising the product of BK and BL is the same as maximising BD \cdot BN. But BD is a constant independent of the position of the segment KL. Hence, the problem we are after reduces to that of maximising the segment BN or DN.

To maximise DN, we turn our attention to the locus of the centre of the circumcircle, say J (Take a wild guess as to what the locus might be before continuing).

Let A and C be the centres of the respective circles and let the line AC meet the circles at Q and R as shown below.

Then, for any general position of KL, \angle CAD=2\angle CQD=\angle BQD=\angle BKD

Also, because LD is a chord on both the circumcircle and (one of) the given circle, it's perpendicular bisector pass through both C and J. Therefore, \angle BKD=\angle LKD=\angle DJC.

We've seen that the segment CD subtends the same angle at both A and J in any general position. Therefore, J lies on the circle containing the points A, D and C i.e. the circumcircle of \triangle ADC.

It should be easy to see that DN (which is a chord on the circumcircle of \triangle KDL) is maximised when the projection of DJ along the line BD is at its highest. This happens when J is at the 'top' of the circle i.e. at the intersection of the \triangle ACD's circumcircle and the perpendicular bisector of AC.

Now that we know the position of J, we can draw a circle with J as centre and JD as radius. Then, the intersection of this circle with the given circles gives us points K and L such that the product BK \cdot BL is maximised.

Even though we have now shown the geometric construction of KL to maximize the product, it's not as clean as the statement 'KL must be perpendicular to BD' we had in the maximising-the-sum case.

Do we even have such a relation in the product case? We'll see about that in the next post.

Hope you enjoyed this. See ya later.

Until then
Yours Aye
Me

No comments:

Post a Comment