Maths Olympiad Prep

Library / China National Team Selection Test /25 of 46

China National Team Selection Test

Geometry Difficulty 8.5 Shortlist Prove it China

As illustrated in Fig. 1.1, ABCDEFABCDEF is a cyclic hexagon. The extensions of ABAB and DCDC meet at GG; the extensions of AFAF and DEDE meet at HH. Let MM and NN be the circumcentres of △BCG\triangle BCG and △EFH\triangle EFH, respectively. Prove that the lines BEBE, CFCF, and MNMN are concurrent.

Figure 1
Fig. 1.1

Solutions — 4

Solution 1

Let ω\omega be the circumcircle of △BCG\triangle BCG. Let E′E' be the other intersection of the line BEBE and ω\omega, F′F' be the other intersection of the line CFCF and ω\omega, as shown in Fig. 1.2. Note that ∠BE′F′=∠BCF′=∠BCF=∠BEF\angle BE'F' = \angle BCF' = \angle BCF = \angle BEF, and hence EF∥E′F′EF \parallel E'F'; similarly, ∠CF′G=∠CBG=∠CFA\angle CF'G = \angle CBG = \angle CFA, and hence GF′∥HFGF' \parallel HF; moreover, GE′∥HEGE' \parallel HE. It follows that △E′F′G\triangle E'F'G and △EFH\triangle EFH are homothetic triangles. Let PP be the homothetic centre. Then EE′EE' and FF′FF' pass through PP; the line connecting the circumcentres of △E′F′G\triangle E'F'G and △EFH\triangle EFH, which is MNMN, passes through PP as well. Therefore, the lines BEBE, CFCF, and MNMN meet at PP. □\square

Figure 2
Fig. 1.2

Solution 2

(Edited from Wang Zihan's proof.) As in Fig. 1.3, we first apply Pascal's theorem to the cyclic hexagon ABCDEFABCDEF to deduce that GG, PP, HH are collinear. Let A′A' and D′D' be the antipodal points of AA and DD, respectively. Let BA′BA' and CD′CD' meet at G′G', D′ED'E and A′FA'F meet at H′H'. Apply Pascal's theorem to the cyclic hexagon BA′FCD′EBA'FCD'E to deduce that G′G', H′H', PP are collinear.

Since AA and A′A', DD and D′D' are antipodal points, we have A′B⊥ABA'B \perp AB, A′F⊥AFA'F \perp AF, CD′⊥CDCD' \perp CD, D′E⊥DED'E \perp DE. Notice that ∠AHH′=90∘−∠FH′H=90∘−∠FAD\angle AHH' = 90^\circ - \angle FH'H = 90^\circ - \angle FAD. Thus, the angle between AHAH and GG′GG' is given by
180∘−∠AGG′−∠BAF=90∘−∠BAF+∠BG′G=90∘−∠BAF+∠DAB=90∘−∠DAF. \begin{align*} 180^\circ - \angle AGG' - \angle BAF &= 90^\circ - \angle BAF + \angle BG'G \\ &= 90^\circ - \angle BAF + \angle DAB \\ &= 90^\circ - \angle DAF. \end{align*}
This implies HH′∥GG′HH' \parallel GG'. As MM is the midpoint of GG′GG', NN is the midpoint of HH′HH', and furthermore it is shown that GHGH, G′H′G'H' pass through PP. We infer that MNMN passes through PP, too. Therefore, BEBE, CFCF, and MNMN are concurrent. □\square

Figure 3
Fig. 1.3

Solution 3

As shown in Fig. 1.4, let PP be the intersection of BEBE and CFCF. It suffices to show that MM, PP, NN are collinear, or equivalently,
sin⁡∠MPBsin⁡∠MPC=sin⁡∠NPEsin⁡∠NPF. \frac{\sin \angle MPB}{\sin \angle MPC} = \frac{\sin \angle NPE}{\sin \angle NPF}.
Denote α=∠MPB\alpha = \angle MPB, β=∠NPE\beta = \angle NPE, γ=∠BPC=∠FPE\gamma = \angle BPC = \angle FPE. The above relation is equivalent to
sin⁡αsin⁡(γ−β)=sin⁡βsin⁡(γ−α), \sin \alpha \sin(\gamma - \beta) = \sin \beta \sin(\gamma - \alpha),
or rewritten as sin⁡γsin⁡(α−β)=0\sin \gamma \sin(\alpha - \beta) = 0, or just α=β\alpha = \beta (for the collinearity of MM, PP, NN).

In △MPC\triangle MPC and △MPB\triangle MPB, apply the law of sines to obtain
BMsin⁡∠MPB=PMsin⁡∠PBM,CMsin⁡∠MPC=PMsin⁡∠PCM. \frac{BM}{\sin \angle MPB} = \frac{PM}{\sin \angle PBM}, \quad \frac{CM}{\sin \angle MPC} = \frac{PM}{\sin \angle PCM}.
As BM=MCBM = MC, it follows that
sin⁡∠MPBsin⁡∠MPC=sin⁡∠PBMsin⁡∠PCM. \frac{\sin \angle MPB}{\sin \angle MPC} = \frac{\sin \angle PBM}{\sin \angle PCM}.
Similarly, in △NPF\triangle NPF and △NPE\triangle NPE, we have
sin⁡∠NPEsin⁡∠NPF=sin⁡∠NEPsin⁡∠NFP. \frac{\sin \angle NPE}{\sin \angle NPF} = \frac{\sin \angle NEP}{\sin \angle NFP}.
Now it suffices to show
sin⁡∠PBMsin⁡∠PCM=sin⁡∠NEPsin⁡∠NFP.1◯ \frac{\sin \angle PBM}{\sin \angle PCM} = \frac{\sin \angle NEP}{\sin \angle NFP}. \qquad \textcircled{1}

The angles appearing in the numerators are supplementary, as
∠PBM+∠NEP=∠PBC+∠MBC+∠PEF+∠NFE=(180∘−∠CDE)+(90∘−∠BGC)+(180∘−∠BAF)+(90∘−∠AHD)=540∘−(∠CDE+∠BGC+∠BAF+∠AHD)=540∘−360∘=180∘, \begin{align*} \angle PBM + \angle NEP &= \angle PBC + \angle MBC + \angle PEF + \angle NFE \\ &= (180^\circ - \angle CDE) + (90^\circ - \angle BGC) \\ &\quad + (180^\circ - \angle BAF) + (90^\circ - \angle AHD) \\ &= 540^\circ - (\angle CDE + \angle BGC + \angle BAF + \angle AHD) \\ &= 540^\circ - 360^\circ = 180^\circ, \end{align*}
and thus sin⁡∠PBM=sin⁡∠NEP\sin \angle PBM = \sin \angle NEP. Analogously, sin⁡∠PCM=sin⁡∠NFP\sin \angle PCM = \sin \angle NFP. We have verified (1), and the statement. □\square

Figure 4
Fig. 1.4

Solution 4

Assume that the circumcircle of △ABC\triangle ABC is the unit circle in the complex plane. We use lowercase letters to represent the complex numbers corresponding to the points in the plane (for instance, aa is the complex number for the point AA, and so on). Let BE∩CF=XBE \cap CF = X. We have
x∈CF⇔x−cx−f=(x−cx−f)‾⇔xxˉ−fˉx−cxˉ+cfˉ=xxˉ−cxˉ−fxˉ+cfˉ⇔(c−f)xˉ−(cˉ−fˉ)x=cfˉ−cfˉ⇔xˉ+cfˉxˉ=cˉ+fˉ. \begin{align*} x \in CF &\Leftrightarrow \frac{x-c}{x-f} = \overline{\left(\frac{x-c}{x-f}\right)} \\ &\Leftrightarrow x\bar{x} - \bar{f}x - c\bar{x} + c\bar{f} = x\bar{x} - c\bar{x} - f\bar{x} + c\bar{f} \\ &\Leftrightarrow (c-f)\bar{x} - (\bar{c} - \bar{f})x = c\bar{f} - c\bar{f} \\ &\Leftrightarrow \bar{x} + c\bar{f}\bar{x} = \bar{c} + \bar{f}. \end{align*}
In the same manner, xˉ+bˉeˉxˉ=bˉ+eˉ\bar{x} + \bar{b}\bar{e}\bar{x} = \bar{b} + \bar{e}. Thus,
x=bˉ+eˉ−cˉ−fˉbˉeˉ−cˉfˉ.1◯ x = \frac{\bar{b} + \bar{e} - \bar{c} - \bar{f}}{\bar{b}\bar{e} - \bar{c}\bar{f}}. \qquad \textcircled{1}
On the other hand, △BMC\triangle BMC is an isosceles triangle, ∠BMC\angle BMC is twice the angle between the lines CDCD and ABAB, and this implies that
c−mb−m=(d−ca−b)2/∣d−ca−b∣2=d−ca−b/dˉ−cˉaˉ−bˉ=cdab. \frac{c-m}{b-m} = \left(\frac{d-c}{a-b}\right)^2 / \left|\frac{d-c}{a-b}\right|^2 = \frac{d-c}{a-b} / \frac{\bar{d}-\bar{c}}{\bar{a}-\bar{b}} = \frac{cd}{ab}.
Consequently,
m=bc(a−d)ad−bc=dˉ−aˉcˉdˉ−aˉbˉ.2◯ m = \frac{bc(a-d)}{ad-bc} = \frac{\bar{d}-\bar{a}}{\bar{c}\bar{d}-\bar{a}\bar{b}}. \qquad \textcircled{2}
In the same way, we can derive
n=(a−d)efaf−be=dˉ−aˉdˉeˉ−aˉfˉ.3◯ n = \frac{(a-d)ef}{af-be} = \frac{\bar{d}-\bar{a}}{\bar{d}\bar{e}-\bar{a}\bar{f}}. \qquad \textcircled{3}
To prove that BEBE, CFCF, and MNMN are concurrent, it is equivalent to prove that MM, XX, NN are collinear, namely
x−nx−m=(x−nx−m)∈R.(4) \frac{x-n}{x-m} = \left( \frac{x-n}{x-m} \right) \in \mathbb{R}. \qquad (4)
From (1)–(3), it follows that
x−n=1(bˉeˉ−cˉfˉ)(dˉeˉ−aˉfˉ)[(bˉ+eˉ−cˉ−fˉ)(dˉeˉ−aˉfˉ)−(dˉ−aˉ)(bˉeˉ−cˉfˉ)]=1(bˉeˉ−cˉfˉ)(dˉeˉ−cˉfˉ)[bˉdˉeˉ+dˉeˉ2−dˉeˉfˉ−cˉdˉeˉ−aˉbˉfˉ−aˉeˉfˉ+aˉfˉ2+aˉcˉfˉ−bˉdˉeˉ+aˉbˉeˉ+cˉdˉfˉ−aˉcˉfˉ]=1(bˉeˉ−cˉfˉ)(dˉeˉ−aˉfˉ)(eˉ−fˉ)(aˉbˉ−aˉfˉ−cˉdˉ+dˉeˉ), \begin{align*} x - n &= \frac{1}{(\bar{b}\bar{e} - \bar{c}\bar{f})(\bar{d}\bar{e} - \bar{a}\bar{f})} [ (\bar{b} + \bar{e} - \bar{c} - \bar{f})(\bar{d}\bar{e} - \bar{a}\bar{f}) - (\bar{d} - \bar{a})(\bar{b}\bar{e} - \bar{c}\bar{f}) ] \\ &= \frac{1}{(\bar{b}\bar{e} - \bar{c}\bar{f})(\bar{d}\bar{e} - \bar{c}\bar{f})} [ \bar{b}\bar{d}\bar{e} + \bar{d}\bar{e}^2 - \bar{d}\bar{e}\bar{f} - \bar{c}\bar{d}\bar{e} - \bar{a}\bar{b}\bar{f} - \bar{a}\bar{e}\bar{f} + \bar{a}\bar{f}^2 \\ &\quad + \bar{a}\bar{c}\bar{f} - \bar{b}\bar{d}\bar{e} + \bar{a}\bar{b}\bar{e} + \bar{c}\bar{d}\bar{f} - \bar{a}\bar{c}\bar{f} ] \\ &= \frac{1}{(\bar{b}\bar{e} - \bar{c}\bar{f})(\bar{d}\bar{e} - \bar{a}\bar{f})} (\bar{e} - \bar{f})(\bar{a}\bar{b} - \bar{a}\bar{f} - \bar{c}\bar{d} + \bar{d}\bar{e}), \end{align*}
while
m−n=dˉ−aˉ(cˉdˉ−aˉbˉ)(dˉeˉ−aˉfˉ)(aˉbˉ−aˉfˉ−cˉdˉ+dˉeˉ). m - n = \frac{\bar{d} - \bar{a}}{(\bar{c}\bar{d} - \bar{a}\bar{b})(\bar{d}\bar{e} - \bar{a}\bar{f})}(\bar{a}\bar{b} - \bar{a}\bar{f} - \bar{c}\bar{d} + \bar{d}\bar{e}).
Hence,
x−nx−m=(eˉ−fˉ)(cˉdˉ−aˉbˉ)(dˉ−aˉ)(bˉeˉ−cˉfˉ). \frac{x - n}{x - m} = \frac{(\bar{e} - \bar{f})(\bar{c}\bar{d} - \bar{a}\bar{b})}{(\bar{d} - \bar{a})(\bar{b}\bar{e} - \bar{c}\bar{f})}.
In the last fraction, the unit complex numbers a,b,c,d,ea, b, c, d, e, and ff all appear exactly once in the numerator and in the denominator; moreover, in each parenthesis, the coefficients of the homogeneous terms sum to zero (one minus one). So, the fraction is real, which justifies (4), and the argument as well.

Want a route through all this instead of an archive? The track puts 2,604 problems in a working order, from Junior Challenge level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement reproduced verbatim; metadata (topic, difficulty) added by this project.