Let $\bigtriangleup ABC$ is not an isosceles triangle and is an acute triangle, $AD,BE,CF$ be the altitudes and $H$ is the orthocenter .Let $I$ is the circumcenter of $\bigtriangleup HEF$ and let $K,J$ is the midpoint of $BC,EF$ respectively.Let $HJ$ intersects $(I)$ again at $G$ and $GK$ intersects $(I)$ at $L\neq G$. a) Prove that $AL$ is perpendicular to $EF$. b) Let $AL$ intersects $EF$ at $M$, the line $IM$ intersects the circumcircle $\bigtriangleup IEF$ again at $N$, $DN$ intersects $AB,AC$ at $P$ and $Q$ respectively then prove that $PE,QF,AK$ are concurrent.
Problem
Source: VMO 2021 P3
Tags: geometry, circumcircle
25.12.2020 17:42
(a) First, note that a composition of a reflection about the $A$-angle bisector and a homothety sends $\triangle AEF$ to $\triangle ABC$. It is well-known that $I$ is the midpoint of $AH$ because $AH$ is the diameter of cyclic quadrilateral $AFHE$. Letting $A'$ be the point diametrically opposite $A$ in the circumcircle of $\triangle ABC$, we see that $H$ is sent to $A'$. Since $J$ is sent to $K$, this means that $G$ is sent to $KA'\cap (ABC)$, which is well-known to be $(ABC)\cap (AFE)$. Suppose this intersection point is $G'$. Then $GA$ and $G'A$ are reflections about the $A$-angle bisector, which means that $\widehat{G'F}=\widehat{GE}$, which means that $GG'\parallel FE$. Since $KF=KE$ ($K$ is the circumcenter of cyclic quadrilateral $BFEC$), this means that by symmetry, $HL\parallel EF$. Supposing that $L$ is sent to $L'$, then $L'A'\parallel BC$, which means that $L'$ is the reflection of the orthocenter of $\triangle ABC$. The conclusion follows. (b) The following rephrasing using different letters might be helpful: VMO 2021 P3(b), rephrased wrote: Let $\triangle ABC$ be a triangle with circumcenter $O$. Suppose $D$ is the foot of the perpendicular from $A$ to $BC$. Let $OD$ intersect $(BOC)$ at $T$ and $AO$ intersect $(BOC)$ at $X$. Denote the intersection points of $TX$ with $AB$ and $AC$ by $P$ and $Q$. Show that $BQ\cap CP$ lies on the $A$-symmedian. This could be fairly easily finished with barycentric coordinates (the only hard part is computing $T$ and $X$). Anyone has synthetic observations for this part?
25.12.2020 18:09
@above: Here is my solution for what you are left. By Ceva-Menelaus configuration, it suffices to show that $AA, BC, TX$ are concurrent. Let $\gamma$ denote the $A$-Apollonius circle, which is centered at $AA\cap BC$ and is orthogonal to $\odot(ABC)$. Let $K = AO\cap BC$. Notice that $D,K$ are inverses w.r.t. $\gamma$. Thus, by inverting this relation around $\odot(ABC)$, we find that $T,X$ are inverses w.r.t. $\gamma$. This gives the desired result.
25.12.2020 20:24
[asy][asy]/* Geogebra to Asymptote conversion, documentation at artofproblemsolving.com/Wiki, go to User:Azjps/geogebra */ import graph; size(11cm); real labelscalefactor = 0.5; /* changes label-to-point distance */ pen dps = linewidth(0.7) + fontsize(10); defaultpen(dps); /* default pen style */ pen dotstyle = black; /* point style */ real xmin = -0.1008145751460307, xmax = 23.18958555555746, ymin = -14.998195361273924, ymax = 5.7958856393983655; /* image dimensions */ draw((14.30347844591869,0.19436179096340434)--(14.518096812342169,0.4882998160206613)--(14.224158787284912,0.7029181824441386)--(14.009540420861434,0.4089801573868817)--cycle); /* draw figures */ draw(circle((13.94,-3.5), 9.085167086298226)); draw((22.190504045342827,-7.303872235464572)--(8.957992989503284,4.097359221026617)); draw((8.957992989503284,4.097359221026617)--(8.652183536235897,-6.928373716076678)); draw((5.4913341534782525,-6.84070447656981)--(12.377282661612282,1.1512734173574215)); draw((6.8822259084921065,-2.452138040581081)--(22.190504045342827,-7.303872235464572)); draw(circle((8.85891208891383,0.5250708650094217), 3.573662144551107)); draw((11.074204276861416,3.3292715791909195)--(13.84091909941054,-7.072288356017191)); draw((8.759831188324362,-3.047217491007767)--(11.074204276861416,3.3292715791909195)); draw((8.759831188324362,-3.047217491007767)--(12.174659180519138,-0.8079266398301627)); draw(circle((11.349915594162193,-3.2736087455038985), 4.542583543149125)); draw((8.957992989503284,4.097359221026617)--(3.017858491126391,-14.645058936866594)); draw((14.009540420861434,0.4089801573868822)--(8.85891208891383,0.5250708650094217)); draw((3.017858491126391,-14.645058936866594)--(12.377282661612282,1.1512734173574215), red); draw((6.8822259084921065,-2.452138040581081)--(13.711909250920973,0.0013490690244003325), red); draw((8.957992989503284,4.097359221026617)--(13.84091909941054,-7.072288356017191), red); draw((8.957992989503284,4.097359221026617)--(13.94,-3.5)); draw((5.4913341534782525,-6.84070447656981)--(22.190504045342827,-7.303872235464572)); draw((8.957992989503284,4.097359221026617)--(14.009540420861434,0.4089801573868822)); draw((3.017858491126391,-14.645058936866594)--(xmax, 1.3695846723447134*xmax-18.778271669618643), linetype("2 2") + blue); /* ray */ draw((6.8822259084921065,-2.452138040581081)--(16.54887845826881,3.886818611321953)); draw((8.957992989503284,4.097359221026617)--(16.54887845826881,3.886818611321953)); /* dots and labels */ dot((13.94,-3.5),linewidth(3.pt) + dotstyle); label("$O$", (14.285189483487397,-3.0570399351452817), NE * labelscalefactor); dot((8.957992989503284,4.097359221026617),linewidth(3.pt) + dotstyle); label("$A$", (8.082994531554203,4.380446957421911), NE * labelscalefactor); dot((5.4913341534782525,-6.84070447656981),linewidth(3.pt) + dotstyle); label("$B$", (4.505794953468252,-7.509237971180314), NE * labelscalefactor); dot((22.190504045342827,-7.303872235464572),linewidth(3.pt) + dotstyle); label("$C$", (22.263116600081965,-8.46144217541902), NE * labelscalefactor); dot((8.652183536235897,-6.928373716076678),linewidth(3.pt) + dotstyle); label("$D$", (8.829316745687242,-7.766590458812396), NE * labelscalefactor); dot((12.377282661612282,1.1512734173574215),linewidth(3.pt) + dotstyle); label("$E$", (12.638133562642068,1.6782458372850413), NE * labelscalefactor); dot((6.8822259084921065,-2.452138040581081),linewidth(3.pt) + dotstyle); label("$F$", (6.590350103288122,-1.8474832432744928), NE * labelscalefactor); dot((8.759831188324362,-3.047217491007767),linewidth(3.pt) + dotstyle); label("$H$", (8.031524034027786,-2.593805457407533), NE * labelscalefactor); dot((8.85891208891383,0.5250708650094217),linewidth(3.pt) + dotstyle); label("$I$", (9.112404482082534,1.0348646182048342), NE * labelscalefactor); dot((9.629754285052194,-0.6504323116118297),linewidth(3.pt) + dotstyle); label("$J$", (9.781520949925948,-1.1526315266678693), NE * labelscalefactor); dot((13.84091909941054,-7.072288356017191),linewidth(3.pt) + dotstyle); label("$K$", (14.516806722356272,-7.766590458812396), NE * labelscalefactor); dot((11.074204276861416,3.3292715791909195),linewidth(3.pt) + dotstyle); label("$G$", (11.480047368297695,3.4797132507096205), NE * labelscalefactor); dot((12.174659180519138,-0.8079266398301627),linewidth(3.pt) + dotstyle); label("$L$", (12.483722070062818,-0.946749536562203), NE * labelscalefactor); dot((11.33716828114958,0.4692131139769764),linewidth(3.pt) + dotstyle); label("$M$", (11.24843012942882,0.8289826280991681), NE * labelscalefactor); dot((14.009540420861434,0.4089801573868822),linewidth(3.pt) + dotstyle); label("$N$", (13.744749259460022,0.9576588719152095), NE * labelscalefactor); dot((3.017858491126391,-14.645058936866594),linewidth(3.pt) + dotstyle); label("$P$", (2.241093062305923,-14.200402649614468), NE * labelscalefactor); dot((13.711909250920973,0.0013490690244003325),linewidth(3.pt) + dotstyle); label("$Q$", (13.590337766880772,-0.8438085415093699), NE * labelscalefactor); dot((16.54887845826881,3.886818611321953),linewidth(3.pt) + dotstyle); label("$S$", (15.957980653095934,4.226035464842661), NE * labelscalefactor); clip((xmin,ymin)--(xmin,ymax)--(xmax,ymax)--(xmax,ymin)--cycle); /* end of picture */[/asy][/asy] Part a) We easily have the tagents at $F,E$ WRT $(I)$ intersect at thi midpoint of $BC$ so $GFLE$ is harmonic. $\implies H(LG,EF)=-1\implies H(LJ,EF)=-1$ but $J$ is the midpoint of $EF$ so $HL\parallel FE$ which claims $AL\bot EF$ since $AL\bot LH$. $\quad\blacksquare$ Part b) (not mine) By power point $M$ we have $MI.MN=ME.MF=MA.ML$ so easy to see that $\angle AND=90^o$ so we claim that $AA$ WRT $(I),ND,EF$ are concurrent at $S$ by radical axis of $(AD), \text{Euler circle}\ , (I)$ Consider $\bigtriangleup APQ$ has $\overline{FE}\cap\overline{PQ}=S$ and since $AK$ is the symmedian of $\bigtriangleup AFE$ we have $$-1=A(S,\overline{AK}\cap\overline{FE},E,F)=A(S,\overline{AK}\cap\overline{PS},Q,P)\implies AK,PE,QF\quad\text{are concurrent}.\quad\blacksquare$$
26.12.2020 16:35
a. It's well known that KF and KE are both tangent to $(I)$. Hence, $GL$ is symmedian of $\triangle FGE$. Therefore, \[\angle FAL=\angle FGL=\angle MGE=\angle HGE=\angle HAE=90^{\circ}-\angle AHE=90^{\circ}-\angle AFE\]which implies $AL\perp EF$. b. Define $X=PQ\cap EF$, $Y=PE\cap QF$, $Z=AY\cap EF$, $T=AH\cap FE$. Define also $X'$ as the intersection of $EF$ and line parallel to $BC$ through $A$. Notice that \[(X,M;F,E)\stackrel{N}{=}(D,I;F,E)\stackrel{I}{=}(T,P_{\infty};F,E)=A(T,P_{\infty};F,E)\]Moreover, from reflection across the internal bisector of $\angle BAC$, we have \[A(T,P_{\infty};F,E)=A(M,X';E,F)=(M,X';E,F)=(X',M;F,E)\]$(X,M;F,E)=(X',M;F,E)$, so $X=X'$. Hence, $AX\parallel BC$. Now, notice that $-1=(X,Z;P,Q)\stackrel{A}{=}(P_{\infty},AZ\cap BC;B,C)$. Therefore, we must have $K=AZ\cap BC$ which finish the problem.
02.02.2021 06:24
ACGNmath wrote: (a) First, note that a composition of a reflection about the $A$-angle bisector and a homothety sends $\triangle AEF$ to $\triangle ABC$. It is well-known that $I$ is the midpoint of $AH$ because $AH$ is the diameter of cyclic quadrilateral $AFHE$. Letting $A'$ be the point diametrically opposite $A$ in the circumcircle of $\triangle ABC$, we see that $H$ is sent to $A'$. Since $J$ is sent to $K$, this means that $G$ is sent to $KA'\cap (ABC)$, which is well-known to be $(ABC)\cap (AFE)$. Suppose this intersection point is $G'$. Then $GA$ and $G'A$ are reflections about the $A$-angle bisector, which means that $\widehat{G'F}=\widehat{GE}$, which means that $GG'\parallel FE$. Since $KF=KE$ ($K$ is the circumcenter of cyclic quadrilateral $BFEC$), this means that by symmetry, $HL\parallel EF$. Supposing that $L$ is sent to $L'$, then $L'A'\parallel BC$, which means that $L'$ is the reflection of the orthocenter of $\triangle ABC$. The conclusion follows. (b) The following rephrasing using different letters might be helpful: VMO 2021 P3(b), rephrased wrote: Let $\triangle ABC$ be a triangle with circumcenter $O$. Suppose $D$ is the foot of the perpendicular from $A$ to $BC$. Let $OD$ intersect $(BOC)$ at $T$ and $AO$ intersect $(BOC)$ at $X$. Denote the intersection points of $TX$ with $AB$ and $AC$ by $P$ and $Q$. Show that $BQ\cap CP$ lies on the $A$-symmedian. This could be fairly easily finished with barycentric coordinates (the only hard part is computing $T$ and $X$). Anyone has synthetic observations for this part? Why can b) be rephrased this way??
06.03.2021 09:10
Part (a) : $BCEF$ is cyclic with center $K \implies KJ \perp EF$ Also, $IJ \perp EF \implies I,J,K$ are collinear. $JI.JK=JF.JE=JG.JH \implies IHGK$ cyclic. $\angle HIK = \angle HGK = \angle HGL = \angle HAL \implies IK \parallel AL$ So, $AL \perp EF$ Part (b) same as DNCT1.
19.07.2024 18:31
a) It's easy to see that $GK$ is $G$ - symmedian of $\triangle GEF,$ so $GH, GL$ are isogonal conjugate in $\angle{EGF}$. Then $\angle{LAE} = \angle{LGE} = \angle{HGF} = \angle{HAF}$ or $AL \perp EF$ b) Let $T$ be intersection of tangent at $A$ of $(AEF)$ with $EF,$ $DT$ intersects $(DEF)$ again at $N',$ $AH$ intersects $EF$ at $U$. We have $\dfrac{N'E}{N'F} = \dfrac{TE}{TF} \cdot \dfrac{DF}{DE} = \dfrac{AE^2}{AF^2} \cdot \dfrac{UF}{UE} = \dfrac{AE^2}{AF^2} \cdot \dfrac{UF}{UE} = \dfrac{ME}{MF}$. So $N'M$ is bisector of $\angle{EN'F}$ or $N', M, I$ are collinear. Hence $N' \equiv N$. From this, if we let $V \equiv PE \cap QF$ then $A(PQ, VT) = - 1$. But $A(PQ, KT) = A(FE, KT) = - 1$ so $A, V, K$ are collinear or $AK, PE, QF$ concur at $V$