GeometryDifficulty 7.0National Olympiad, round 2Prove itUnited States
Let ABCD be a quadrilateral with AC=BD. Diagonals AC and BD meet at P. Let ω1 and O1 denote the circumcircle and circumcenter of triangle ABP. Let ω2 and O2 denote the circumcircle and circumcenter of triangle CDP. Segment BC meets ω1 and ω2 again at S and T (other than B and C), respectively. Let M and N be the midpoints of minor arcs SP (not including B) and TP (not including C). Prove that MN∥O1O2.
(This problem was suggested by Steve Dinh.)
Solution
Note. The result still holds without the assumption that both triangles ABP and CDP are acute. This assumption helps the contestants to focus on more specific configurations. Indeed, because triangles are acute, O1 lies inside triangle ABP and O2 lies inside triangle CDP. Points M and N lie in the region bounded by rays PB and PD. It is not difficult to see that O1MNO2 is a convex quadrilateral. (In particular, this is helpful in solution 2.) Hence we can consider the configuration (in two diagrams) shown below. For other possible configurations, our proofs can be adjusted slightly. We present two solutions. Both solutions are based on the fact that triangles ABQ and CDQ are similar isosceles triangles. Indeed, because ABPQ and DCPQ are cyclic, we have ∠QBD=∠QAC.
and ∠QDB=∠QCA. Hence triangles ACQ and BDQ are similar to each other. Because AC=BD, we conclude that triangle ACQ and BDQ are congruent to each other, implying that QA=QB and QC=QD. Because ∠BQD=∠AQC, we have ∠AQB=∠CQD and isosceles triangles ABQ and CDQ are similar to each other. In particular, ∠QBA=∠QAB=∠QCD=∠QDC.
Solution 2. For point X and line ℓ, let d(X,ℓ) denote the distance from X to ℓ. Because O1MNO2 is a convex quadrilateral by the note prior to Solution 1, it suffices to show that d(M,O1O2)=d(N,O1O2). Working on arcs along ω1, we have ∠MO1O2=∠MO1P+∠PO1O2=MP+2∠PO1Q=2SP+∠PBQ=∠PBS+∠PBQ=∠SBQ=∠CBQ. Thus, we have O1Md(M,O1O2)=sin∠MO1O2=sin∠QBC=BQd(Q,BC)ord(Q,BC)d(M,O1O2)=O1MBQ. In exactly the same way, we can show that ∠NO2O1=∠BCQ and d(Q,BC)d(N,O1O2)=O2NCQ. It suffices to show that O1MBQ=O2NCQ, which is holds because BQ and CQ are two corresponding sides of two similar (isosceles) triangles (namely, BAQ and CDQ) inscribed in circles ω1 and ω2, with radii O1M and ON, respectively.
Want a route through all this instead of an archive? The track
puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.
Source: MathNet,
licensed CC-BY-4.0.
Statement reproduced verbatim; metadata (topic, difficulty) added by this project.