An Inversion problem from PiMuEpsilon

Often times in math, a given problem can be solved in multiple ways using algebra, geometry, calculus etc. and we try to find the shortest proof among them. But say if we are interested in Geometry, we would still look for a geometrical proof even if it is more cumbersome. This post is a perfect example of such an instance.

Let's say we a given a semicircle $C_1$ with diameter $AB$ and a circle $C_2$ with center $P$ internally tangent to $C_1$. Let the two (half) chords of $C_1$ tangent to $C_2$ and perpendicular to diameter $AB$ meet $AB$ at $R$ and $S$ respectively. Let the line perpendicular to $AB$ and passing through $P$ meet $AB$ at $Q$. We asked to prove that $PQ$ is the geometric mean of $AR$ and $SB$.

An algebraic proof is easy to find but we attempt to find a geometric one. To do that, we begin by constructing a few circles.

We first construct a circle tangent at and passing through $B$ and also is orthogonal to the blue circle, which can be constructed by drawing a tangent from $B$ to the blue circle and using the distance from $B$ and the tangent point as the radius. This is shown in orange in the next figure.

We construct another circle that is internally tangent to the green circle at $B$ and externally tangent to the blue circle. We also reflect this circle w.r.t line $PQ$. Both these are shown in violet in the following figure.

A little thought shows that the violet circle and the green circle must be inversions of each other under the orange circle.

We finally contruct another circle using the following three points (i) intersection of line $PQ$ and the orange circle and (ii) the points of intersection of the green circle and the mirrored violet circle. This is shown in red in the following figure.

We will now show that the mirrored violet circle and the green circle are inversion of each other under the red circle.

To this effect, we construct an inversion circle with center at the (lower) point of intersection of line $PQ$ and the violet circle, and make the inversion circle orthogonal to the violet circles. The inversion circle is shown in black in the following figure.

The violet circles and line $PQ$ are all invariant under this inversion.

Both the orange and the red circle pass through the center of inversion. Therefore they become straight lines. These lines can be drawn using the points of intersection of the respective circles with the inversion circle. Note that the point of intersection of the lines must lie on line $PQ$.

Line $PQ$ and the orange circle are both orthogonal to the blue circle. Therefore, their inversions will also be so. This shows the point of intersection of the orange line and line $PQ$ is the center of the blue circle's inversion. Using the fact the blue circle is tangent to both the (invariant) violet circles, we can also construct the 'inverted' blue circle.

Because the red line also passes through the point of intersection of the orange line and line $PQ$ (because the red circle passes through the intersection of the orange circle and line $PQ$), we see that the red line is orthogonal to the 'inverted' blue circle. That means, the red circle and the blue circle are actually orthogonal to each other.

As we know that the green circle and the violet circle are inversion of each other under the orange circle, we get to see that the inversion of the green circle must be the reflection of the (invariant) violet circle w.r.t the orange line.

Also, the red circle passes through the point of intersection of the green circle and the mirrored violet circle which must also be the case after inversion. Therefore we see that the 'inverted' green circle and the mirrored violet circle are reflections of each other w.r.t red line.

That clearly means that the green circle and the mirrored violet circle are actually inversions of each other under the red circle.

Because of the said inversion, we see that $FR^2=RA \cdot RA'$

But $FR=PQ$. Also, $RA'=SB$ by symmetry. Therefore, $PQ^2=AR \cdot SB$ as claimed.

Until then
Yours Aye
Me