G3. Let be a convex quadrilateral and let and be points in such that and are cyclic quadrilaterals. Suppose that there exists a point on the line segment such that and . Show that the quadrilateral is cyclic.
Problem 1317
Official solution
Solution 1. Let be the point on the line such that . By hypothesis, the quadrilateral is cyclic. So if lies between and then ; the points and are on distinct sides of the line and we infer that is a cyclic quadrilateral. And if lies between and then a similar argument shows that ; but now the points and lie on the same side of , so that is a cyclic quadrilateral.
In either case we obtain the equality which implies that . So the triangles and are either homothetic or parallel-congruent. More specifically, triangle is the image of under the mapping which carries the points respectively to and is either a homothety or translation by a vector. Note that is uniquely determined by these conditions and the position of the points alone.
Let now be the point on the line such that . The same reasoning as above applies to points in place of , implying that the triangle is the image of under the same mapping . So sends the four points respectively to .
If , so that is a homothety with a centre , then the lines -i.e. the lines -are concurrent at . And since and are cyclic quadrilaterals, the equalities hold, showing that the quadrilateral is cyclic.
Finally, if , so that is a translation, then . Thus and are isosceles trapezoids. Then also is an isosceles trapezoid, hence a cyclic quadrilateral.
Solution 2. Here is another way to reach the conclusion that the lines and are either concurrent or parallel. From the cyclic quadrilateral we get
Hence . This in view of the tangent-chord theorem means that the circumcircle of triangle is tangent to the line at . Analogously, the circumcircle of triangle is tangent to at .
Suppose that the line intersects at . Since is tangent to the circle , . Also, because lie on a circle. Therefore .
It is not hard to see that this equation determines the position of the point on the line uniquely. Thus, if also cuts , say at , then the analogous equation for yields , meaning that the three lines indeed concur. In this case, as well as in the case where , the concluding argument is the same as in the first solution.
It remains to eliminate the possibility that e.g. meets at while . Indeed, would then be an isosceles trapezoid and the angle equality would force that is the midpoint of . So the length of , which is the geometric mean of the lengths of and , should also be their arithmetic mean-impossible, as . The proof is now complete.
Comment. After reaching the conclusion that the circles and are tangent to one may continue as follows. Denote the circles (PQDA), (EDA), (EBC), (QPBC) by respectively. Let be the radical axis of the pair for . As is well-known, the lines concur, possibly at infinity (let this be the meaning of the word concur in this comment). So do the lines . Note however that and both coincide with the line . Hence the pair is in both triples; thus the four lines and are concurrent.
Similarly, concur, concur, and since , the four lines and are concurrent. The lines and are present in both quadruples, therefore all the lines are concurrent. Hence the result.