In a convex hexagon ABCDEF, AB∥DE, BC∥EF, CD∥FA, and AB+DE=BC+EF=CD+FA. Denote the midpoints of sides AB, BC, DE, EF by A1,B1,D1,E1, respectively. Prove that ∠D1OE1=21∠DEF, where O is the point of intersection of segments A1D1 and B1E1.
Solution
Define α=π−∠FAB=π−∠CDE, β=π−∠ABC=π−∠DEF, γ=π−∠BCD=π−∠EFA. Clearly α+β+γ=π, and at least two of the three are acute. Let O′ be the intersection point of B1E1 and C1F1, and O′′ the intersection point of C1F1 and A1D1 (where C1,F1 are the midpoints of CD and FA respectively). Further let α′=∠F1O′′A1=∠C1O′′D1, β′=∠A1OB1=∠D1OE1, γ′=∠B1O′C1=∠E1O′F1. Since we may translate C1F1 parallel to itself until it passes through the point O, and such a parallel translation does not change the angles of intersection between lines, we know that α′+β′+γ′=π. Moreover, by the cyclic symmetry of the problem, the statement of the problem is equivalent to β′=∠D1OE1=21∠DEF=2π−β, or to α′=2π−α, or to γ′=2π−γ. Hence, without loss of generality, we may relabel the vertices cyclically without changing the problem, so that we may assume that α and β are acute. Write m=tanα, M=tanβ; both m,M are positive.
Now we set up a coordinate system: without loss of generality let A≡(−d,−h), B≡(d,−h), D≡(d′+Δ,h), E≡(−d′+Δ,h), where d,d′,h are positive numbers, and d′≥d. After a careful algebraic computation (finding the equations of the lines BC, CD, EF, FA, computing their intersection points, and using the fact that the slopes of lines BC, CD are m,M respectively), we obtain C≡(M+mMd′+MΔ+md+2h,M+mMmd′+MmΔ−Mmd−Mh+mh), F≡(−M+mmd′−mΔ+Md+2h,M+mMmd′−MmΔ−Mmd+Mh−mh). Note that the y-coordinates of C,F must lie in the interval (−h,h) in order for the hexagon to be convex, and hence we have m2h>d′−d+Δ>−M2h,M2h>d′−d−Δ>−m2h. After some further algebraic computation, and using the above results, we obtain BC=M+mM(d′−d+Δ)+2hm2+1, EF=M+m2h−M(d′−d−Δ)m2+1, CD=M+m2h−m(d′−d+Δ)M2+1, FA=M+mm(d′−d−Δ)+2hM2+1. Hence we have d+d′=2AB+DE=2BC+EF=2CD+FA=M+m2h+MΔm2+1=M+m2h−mΔM2+1. From the last two terms of the above equality, we find that (2hΔ)2+2M−mMm+1(2hΔ)−12hΔ=0,=M−m−Mm−1+(M2+1)(m2+1)=tan2α−β, where here we assume that M=m; if M=m, then Δ=0, and we have d+d′=mhm2+1=sinβh=sinαh. Note that we discard the negative root, not only for the sake of continuity of the solution as M→m, but also because, after some computation, one can see that 2h=(d+d′)(sinα+sinβ) always holds, since 2h=BCsinβ+CDsinα=EFsinβ+FAsinα.
We can now easily derive that A1≡(0,−h),D1≡(Δ,h), B1≡(2(M+m)Md′+Md+MΔ+2md+2h,2(M+m)Mmd′+MmΔ−Mmd+2Mh), E1≡(−2(M+m)2md′−2mΔ−MΔ+Md+Md′+2h,2(M+m)Mmd′−MmΔ−Mmd+2Mh). β′ is also the angle between the vectors A1D1 and B1E1, where A1D1≡(Δ,2h), B1E1≡(−M+m(M+m)(d+d′)+2h−mΔ,−MM+mmΔ−2h). Since Δ2h=cot2α−β, the vector A1D1 makes an angle of size 2π−α+β with the horizontal axis. Furthermore, (M+m)(d+d′)+2h−mΔM(mΔ−2h)=sinα+sinβtanα+tanβ+1−tanβtan2α−βtanα(tanβtan2α−β−1)=cos2α+β(1+cosα)−sinαcos2α+β=−tan2α, that is, the vector B1E1 makes an angle of π−2α with the horizontal axis. Therefore, the angle formed between B1E1 and A1D1 is β′=π−2α−2π−α+β=2π−β. This completes the proof.
Note that when M=m, A1D1≡(0,2h), which makes an angle of 2π with the horizontal axis; and B1E1≡(−mm(d+d′)+h,h), and since −m(d+d′)+hmh=−1+cosβsinβ=−tan2β, we get that B1E1 makes an angle of π−2β with the horizontal axis. Once again we have β′=π−2β−2π−β=2π−β. This completes the proof.
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 translated into English from zh; metadata (topic, difficulty) added by this project.