Let $ABC$ be an acute triangle with incenter $I$ and circumcircle $\omega$. Suppose a circle $\omega_B$ is tangent to $BA,BC$, and internally tangent to $\omega$ at $B_1$, while a circle $\omega_C$ is tangent to $CA, CB$, and internally tangent to $\omega$ at $C_1$. If $B_2, C_2$ are the points opposite to $B,C$ on $\omega$, respectively, and $X$ denotes the intersection of $B_1C_2, B_2C_1$, prove that $XA=XI$. Proposed by Vincent Huang and Nathan Weckwerth
Problem
Source: 2017 ELMO Shortlist G4
Tags: geometry
03.07.2017 06:45
Let $\{Y,Z\}$ be the midpoints of minor arcs $AC, AB$ and $\{V,W\}$ be the midpoints of major arcs $AC, AB$. It's well-known that $BB_1\cap CC_1 \in OI$, so Pascal on $BB_1C_2CC_1B_2$ and $WC_1B_2BYZ$ implies $\{B_1C_2, C_1B_2, OI\}$ are concurrent on $YZ$, which is the perpendicular bisector of $\overline{AI}$.
03.07.2017 07:23
What are $B_2, C_2$?
03.07.2017 07:47
tastymath75025 wrote: If $B_2, C_2$ are the points opposite to $B,C$
03.07.2017 19:29
EulerMacaroni wrote: tastymath75025 wrote: If $B_2, C_2$ are the points opposite to $B,C$ You should read the problem carefully next time, the statement doesn't say $ B_2,C_2\in\odot(ABC).$
03.07.2017 19:39
DSD wrote: EulerMacaroni wrote: tastymath75025 wrote: If $B_2, C_2$ are the points opposite to $B,C$ You should read the problem carefully next time, the statement doesn't say $ B_2,C_2\in\odot(ABC).$ Ok congrats on ur reading abilities
03.07.2017 23:02
Everyone's upvoting the previous post, but no one is saying anything. Lol.
04.07.2017 00:33
Well, the thing is there is a certain "someone" who "trivialises" every problem by just writing two lines about it and wastes half of his/her posts criticising other comments. So these upvotes are just a reflection of the "repercussions" of his immature conduct.
19.06.2019 09:33
babu2001 wrote: Well, the thing is there is a certain "someone" who "trivialises" every problem by just writing two lines about it and wastes half of his/her posts criticising other comments. So these upvotes are just a reflection of the "repercussions" of his immature conduct. The first post is precisely an entire , complete , total , perfect and beautiful solution with Pascal’s theorem!
10.02.2020 10:24
Basically the same as post #02, but this is too beautiful! Denote the incircle as $\Omega$. $\textbf{Claim 01.}$ $O, I, CC_1 \cap BB_1$ are collinear. $\textit{Proof.}$ Monge Theorem on $(\omega_B, \omega, \Omega), (\omega_C, \omega, \Omega)$ respectively gives us that $BB_1 \cap CC_1$, $O$, and $I$ are collinear. $\textbf{Claim 02.}$ $X,I,O$ are collinear. $\textit{Proof.}$ Pascal Theorem on $B_1 C_2 C C_1 B_2 B$ gives us $B_1 C_2 \cap C_1 B_2 = X$, $CC_2 \cap BB_2 = O$ and $CC_1 \cap BB_1$ are collinear. Therefore, by $\textbf{Claim 01}$, we have $X,I,O$ are collinear. Now, denote $Y$ and $Z$ as the midpoints of minor arc $AB$ and $AC$ respectively. $\textbf{Claim 03.}$ $X,Y,Z$ are collinear. $\textit{Proof.}$ Notice that by Pascal on $WC_1 B_2 BYZ$ on $WC_1 \cap BY = I$, $C_1 B_2 \cap YZ$ and $B_2 B \cap WZ = O$ are collinear. This gives us $C_1 B_2, YZ$ and $OI$ are collinear. But notice that $C_1 B_2 \cap OI = X$, which gives us $X,Y,Z$ are collinear. Now, notice that $YZ$ is the perpendicular bisector of $AI$, and therefore $XA = XI$.
03.03.2020 21:15
Overcomplicated it whoops. tastymath75025 wrote: Let $ABC$ be an acute triangle with incenter $I$ and circumcircle $\omega$. Suppose a circle $\omega_B$ is tangent to $BA,BC$, and internally tangent to $\omega$ at $B_1$, while a circle $\omega_C$ is tangent to $CA, CB$, and internally tangent to $\omega$ at $C_1$. If $B_2, C_2$ are the points opposite to $B,C$ on $\omega$, respectively, and $X$ denotes the intersection of $B_1C_2, B_2C_1$, prove that $XA=XI$. Proposed by Vincent Huang and Nathan Weckwerth Let $B_0, C_0$ be the midpoint of the arcs $ABC, BCA$ respectively, and let $B_0^\ast, C_0^\ast$ be the midpoints of the opposite arcs. Then, it is well known that $B_1=B_0I \cap \omega$ and $C_1=C_0I \cap \omega.$ We will show that $B_1C_2 \cap B_2C_1$ is the same as $B_0^\ast C_0^\ast \cap OI.$ This is enough since $B_0^\ast C_0^\ast$ is the perpendicular bisector of $AI.$ To do this, define $X=B_0^\ast C_0^\ast \cap B_2C_1.$ We show $X$ lies on $OI,$ which is sufficient. Apply Pascal: \begin{align*} B_0^\ast C_0^\ast CB_2C_1C_0 &\implies \{X,I,CB_2 \cap C_0B_0^\ast\} \text{ collinear}\\ CB_2BC_0B_0^\ast B_0 &\implies \{CB_2 \cap C_0B_0^\ast, O, BC_0 \cap B_0C \} \text{ collinear} \\ BC_0C_0^\ast B_0CC_2 &\implies \{BC_0 \cap B_0C, O, C_0^\ast B_0 \cap BC_2\} \text{ collinear} \\ C_0^\ast B_0B_0^\ast BC_2C &\implies \{C_0^\ast B_0 \cap BC_2,O,I \} \text{ collinear} \end{align*}Hence, $X,I,O$ are collinear. Done. $\blacksquare$ [asy][asy] /* Geogebra to Asymptote conversion, documentation at artofproblemsolving.com/Wiki go to User:Azjps/geogebra */ import graph; size(13cm); 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 = -1.273942087622393, xmax = 10.043611041162661, ymin = -4.092379271296824, ymax = 5.821334147140195; /* image dimensions */ pen xfqqff = rgb(0.4980392156862745,0,1); pen qqffff = rgb(0,1,1); pen wwzzff = rgb(0.4,0.6,1); draw((2.1344095710858766,4.40629172657896)--(1.3539239328593422,-1.5865405145415943)--(6.279422524006489,-1.6564608415452602)--cycle, linewidth(0.7) + green); /* draw figures */ draw((2.1344095710858766,4.40629172657896)--(1.3539239328593422,-1.5865405145415943), linewidth(0.7) + green); draw((1.3539239328593422,-1.5865405145415943)--(6.279422524006489,-1.6564608415452602), linewidth(0.7) + green); draw((6.279422524006489,-1.6564608415452602)--(2.1344095710858766,4.40629172657896), linewidth(0.7) + green); draw(circle((3.855801428622153,1.1348636463984123), 3.696678456531752), linewidth(0.7) + blue); draw((1.4321803332378162,3.926188134342085)--(7.504571500630715,1.7280824699370836), linewidth(0.7) + linetype("2 2") + xfqqff); draw((6.357678924384963,3.856267807338419)--(0.49514150889773734,-0.40507028560771735), linewidth(0.7) + xfqqff); draw((0.19008041152574484,1.6122744085348044)--(6.907444207716123,3.221226021392171), linewidth(0.7) + xfqqff); draw((3.189251969153812,0.002330738041550151)--(4.784446619626273,2.7127229095930656), linewidth(0.7) + linetype("2 2") + qqffff); draw((6.357678924384963,3.856267807338419)--(1.3539239328593422,-1.5865405145415943), linewidth(0.7) + red); draw((1.4321803332378162,3.926188134342085)--(6.279422524006489,-1.6564608415452602), linewidth(0.7) + red); draw((7.504571500630715,1.7280824699370836)--(0.8041586495281816,-0.9514987285953467), linewidth(0.7) + red); draw((0.49514150889773734,-0.40507028560771735)--(7.52152244571856,0.6574528842620202), linewidth(0.7) + red); draw((6.907444207716123,3.221226021392171)--(0.8041586495281816,-0.9514987285953467), linewidth(0.5) + wwzzff); draw((0.19008041152574484,1.6122744085348044)--(7.52152244571856,0.6574528842620202), linewidth(0.5) + wwzzff); draw((0.19008041152574484,1.6122744085348044)--(6.279422524006489,-1.6564608415452602), linewidth(0.5) + wwzzff); draw((1.3539239328593422,-1.5865405145415943)--(6.907444207716123,3.221226021392171), linewidth(0.5) + wwzzff); /* dots and labels */ dot((2.1344095710858766,4.40629172657896),dotstyle); label("$A$", (1.9968597117451548,4.678000774794904), NE * labelscalefactor); dot((1.3539239328593422,-1.5865405145415943),dotstyle); label("$B$", (1.0561423800686478,-1.8201851009397263), NE * labelscalefactor); dot((6.279422524006489,-1.6564608415452602),dotstyle); label("$C$", (6.526775478125873,-1.9504382699410885), NE * labelscalefactor); dot((3.189251969153812,0.002330738041550151),linewidth(4pt) + dotstyle); label("$I$", (3.1980833814243868,-0.4452905392586798), NE * labelscalefactor); dot((3.855801428622153,1.1348636463984123),linewidth(4pt) + dotstyle); label("$O$", (3.7190960574298364,1.3927264010938771), NE * labelscalefactor); dot((6.907444207716123,3.221226021392171),linewidth(4pt) + dotstyle); label("$B_0^\ast$", (6.932007559463446,3.18732561844598), NE * labelscalefactor); dot((0.19008041152574484,1.6122744085348044),linewidth(4pt) + dotstyle); label("$C_0^\ast$", (-0.4634779249472484,1.5374521444287241), NE * labelscalefactor); dot((0.8041586495281816,-0.9514987285953467),linewidth(4pt) + dotstyle); label("$B_0$", (0.5351297040631976,-1.3715352966017005), NE * labelscalefactor); dot((7.52152244571856,0.6574528842620202),linewidth(4pt) + dotstyle); label("$C_0$", (7.626691127470712,0.4664816437508563), NE * labelscalefactor); dot((7.504571500630715,1.7280824699370836),linewidth(4pt) + dotstyle); label("$B_1$", (7.699053999138136,1.6098150160961477), NE * labelscalefactor); dot((0.49514150889773734,-0.40507028560771735),linewidth(4pt) + dotstyle); label("$C_1$", (0.11542504839214061,-0.6334340055939809), NE * labelscalefactor); dot((1.4321803332378162,3.926188134342085),linewidth(4pt) + dotstyle); label("$C_2$", (1.0561423800686478,4.099097801455516), NE * labelscalefactor); dot((6.357678924384963,3.856267807338419),linewidth(4pt) + dotstyle); label("$B_2$", (6.410994883457995,3.9688446324541538), NE * labelscalefactor); dot((4.784446619626273,2.7127229095930656),linewidth(4pt) + dotstyle); label("$X$", (4.732176260773767,2.94129185477674), NE * labelscalefactor); clip((xmin,ymin)--(xmin,ymax)--(xmax,ymax)--(xmax,ymin)--cycle); /* end of picture */ [/asy][/asy]
17.03.2020 21:05
Attachments:

01.08.2020 12:49
We define some points first . Let $E$ and $F$ denote the midpoint of arcs $\widehat{ABC}$ and $\widehat {ACB}$ respectively. Let $E',F'$ denote their respective antipodes on $\odot(ABC)$ . Set $D \stackrel{\text{def}}{:=} \overline{BB_1}\cap \overline{CC_1}.$ Note that $ \overline{E'F'}$ is the perpendicular bisector of $ \overline{AI}$ . So it suffices to prove that $X \in \overline{E'F'}$ . Set $:=$$$X_1 \stackrel{\text{def}}{:=} \overline{B_1C_2}\cap \overline{E'F'} \quad X_2 \stackrel{\text{def}}{:=} \overline{B_2C_1}\cap \overline{E'F'}.$$ Claim $:=$ $X \in \overline{OI}$ Proof $:=$ We will use the well known fact that $D \in \overline{OI}$ . Note that by Pascal on $B_1C_2CC_1B_2B$, we have $:=$ $$ \underbrace {\overline{B_1C_2}\cap \overline{B_2C_1}}_{X} -\underbrace {\overline{C_2C}\cap \overline{B_2B}}_{O} -\underbrace {\overline{CC_1}\cap \overline{BB_1}}_{D} \quad \text{collinear} \quad \implies X\in \overline {OD} \equiv \overline {OI}$$ Claim $:=$ $X_1,X_2 \in \overline{OI}$ Proof $:=$ We will use the well known fact that $EB_1 \cap FC_1 \equiv I$ By Pascal on $E'F'CC_2B_1E$ , we have $:=$ $$ \underbrace {\overline{B_1C_2}\cap \overline{E'F'}}_{X_1} -\underbrace {\overline{F'C}\cap \overline{B_1E}}_{I} -\underbrace {\overline{CC_2}\cap \overline{EE'}}_{O} \quad \text{collinear} \quad \implies X_1 \in \overline {OI}$$ Similarly , $X_2 \in \overline {OI}$ . Hence $X \equiv X_1 \equiv X_2$ and we conclude $\blacksquare$
Attachments:

12.04.2021 18:36
Let $M$ and $D$ be the midpoints of arc $ABC$ and arc $AC$, and define $N$ and $E$ similarly for $C$. It is known that $M,I,B_1$ are collinear and so are $N,I,C_1$. Since $DE$ is the perpendicular bisector of $AI$, it suffices to show that $X$ lies on $DE$. Let $O$ be the circumcenter of $\triangle ABC$. Then by Pascal's theorem on hexagon $CC_2B_1MDE$, it follows that $CC_2\cap MD=O$, $C_2B_1\cap DE$, and $B_1M\cap EC=I$ are collinear so $OI$, $DE$ and $C_2B_1$ are concurrent. Similarly, $OI,DE$ and $C_1B_2$ are also concurrent and we're done.
04.06.2021 10:39
What do opposite here means ?? anyone ??
26.08.2021 15:35
@above, it means the antipode. I will use Wizard 32's diagram. From,Pascal on $C_{2}B_{1}C_{0}^{*}B_{2}C_{1}B_{0}^{*}$ $CC_{0}^{*}C_{0}BB_{0}^{*}B_{0}$ and $B_{1}BC_{0}C_{1}CB_{0}$, We get, $X--O--I--BC_{0} \cap CB_{0}--BB_{1} \cap CC_{1}$ Now since we know $C_{0}^{*}B_{0}^{*}$ is the perpendicular bisector of $AI$,it remains to show that $X$ lies on $C_{0}^{*}B_{0}^{*}$ Let $X'=C_{2}B_{1} \cap C_{0}^{*}B_{0}^{*}$. Now, combining the above result with Pascal on $CC_{0}^{*}B_{0}^{*}CC_{2}B_{1}$ and $BB_{0}^{*}CC_{1}C_{0}B_{1}$ gives $X'=X$ and hence we are done. So much Pascal....
28.12.2022 23:03
Here's another way of interpreting the Pascals, lol. Let $B_3=B_2I\cap\omega,C_3=C_2I\cap\omega$ and $M,M'$ be the midpoints of small and large arc $AC$ respectively and $N,N'$ be the same for arc $AB$. We first prove the following claim: Claim: $B_3N',BC$ and $C_3M'$ are concurrent at $T$. Proof: We use complex numbers with $\omega$ as the unit circle. With $A=a^2,B=b^2,C=c^2$, then $I=-ab-bc-ca,N'=ab$. One can see that $b_3=\frac{a^2bc+abc^2+a^2c^2-ab^2c}{ab+b^2+bc-ac}$ from $\angle BB_3I=90^\circ$ and then check that $B_3N'\cap BC$ is symmetric in $b$ and $c$ (it's $\frac{a(b^2+c^2)(ab+bc+ca)}{a^2(b+c)+2abc+bc(b+c)}$). Let $\varphi$ be the composition of involutions on $\omega$ with centers $I,T$ and again $I$. They're collinear so by Ping Pong lemma, $\varphi$ is also an involution on $\omega$. Also, note $\varphi(M)=N,\varphi(B_1)=C_2$ and $\varphi(C_1)=B_2$. Thus, $MN,B_1C_2$ and $B_2C_1$ all go through the center of $\varphi$ (which is $X$) which is equivalent to $XA=XI$. Remark: Note we also proved that $T,I$ and $X$ are collinear (by using Ping Pong lemma).
23.08.2023 00:44
Let $S_B$ and $S_C$ denote the midpoints of arcs $AC$ and $BC$ of $\omega$ not containing the third point, and $M_B, M_C$ be their antipodes. Also, let $O$ be the center of $(ABC)$. It's well known that $C_1, I$, and $M_C$ are collinear, as are $B_1, I$, and $M_C$. Additionally, by fact $5$, we find that $S_BS_C$ is the perpendicular bisector of $\overline{AI}$, so it suffices to show that $X$ lies on this line. Now, let $X_1 = \overline{S_BS_C} \cap \overline{B_1C_2}$ and $X_2 = \overline{S_BS_C} \cap \overline{B_2C_1}$. By Pascal's on $S_CS_BM_BB_1C_2C$ and $S_BS_CM_CC_1B_2B$, we find that $X_1 = X_2 = \overline{OI} \cap \overline{B_1C_2} \cap \overline{B_2C_1} = X$, as desired.
04.09.2023 06:34
oops Let $M_A,M_B,N_B,M_C,N_C$ be the midpoints of arcs $\widehat{BC},\widehat{AC},\widehat{ABC},\widehat{AB},\widehat{ACB}$ respectively, let $O$ be the circumcenter, let $T = OI\cap (BIC)$, and let $S = B_2I\cap (ABC)$. By Pascal's on $N_BB_1C_2N_CC_1B_2$, $X$ lies on $OI$ (since $C_2N_C\cap B_2N_B$ is the Bevan point). By Pascal's on $N_CSB_2BCM_C$, $P = N_CS\cap BC$ lies on $OI$. Then by PoP at $P$, $SITN_C$ is cyclic, so the negative inversion at $I$ fixing $(ABC)$ swaps $T$ and $X$. Since $\triangle M_AIT$ is isosceles, it follows that $\triangle XAI$ is isosceles, as desired.
13.02.2024 03:27
Define the midpoints of arcs $AC$, $ABC$, $AB$, $ACB$ be $M$, $N$, $P$, and $Q$. Also let $O$ be the circumcenter, and $IO$ be line $\ell$. Then $BB_1 \cap CC_1$ is known to be the exsimilicenter of the incircle and circumcircle, so it lies on $\ell$. Pascal's on $B_1BB_2C_1CC_2$ tells us $X \in \ell$. Pascal's on $BB_2C_1QPM$ tells us $X \in MP$, the perpendicular bisector of $AI$. $\blacksquare$
13.02.2024 06:07
Let $M_B$ be the midpoint of arc $AC$ and $M_C$ be the midpoint of arc $AB$. Let $N_C$ and $N_B$ be the antipodes of $M_C$ and $M_B$. It is well known that $N_C$, $I$, and $C_1$ are collinear and similarly $N_B$, $I$, and $B_1$ are collinear. It is sufficient to show that $X$ lies on the perpendicular bisector of $AI$ that is $M_BM_C$(This is because $M_BA=M_BI$ and $M_CA=M_CI$). It is equivalent to show that $M_BM_C\cap B_1C_2$ and $M_BM_C\cap C_1B_2$ both lie on $IO$, where $O$ is the circumcenter of $ABC$. But this is just two applications of pascal; pascal on $C_2B_1N_BM_BM_CC$ and on $B_2C_1N_CM_CM_BB$.
Attachments:

03.03.2024 03:43
Let $Y = \overline{BB_1} \cap \overline{CC_1}$. It is well-known (say, see EGMO 7.42) that $Y$ lies on $\overline{OI}$. Also, let $M_B$ and $M_C$ be the minor arc midpoints. Now, we apply Pascal twice: Pascal on $B_1C_2CC_1B_2B$ implies $X$, $O$ (the circumcenter), $Y$ collinear, hence $X$ lies on $\overline{OI}$, and Pascal on $B_2C_1QM_CM_BB$ implies that $I$, $O$, and $\overline{M_BM_C} \cap \overline{B_2C_1}$ are collinear. The second condition implies that $X$ lies on $\overline{M_BM_C}$, i.e. $XA=XI$.
05.03.2024 09:06
Let $D$ be the second intersection of $AI$ with $\omega$. Define $E,F$ similarly for $B,C$, respectively. Let $M_A,M_B,M_C$ be the mid-points of arcs $BAC,CBA,ACB$, respectively. It is well-known that $M_BB_1$ and $M_CC_1$ intersect at $I$. Let $I'=\overline{B_2M_B}\cap\overline{C_2M_C}$. Since $I=\overline{BE}\cap\overline{CF}$, reflecting about $O$, $B$ goes to $B_2$, $E$ goes to $M_B$, $C$ goes to $C_2$ and $F$ goes to $M_C$. Hence, $I$ goes to $I'$. So, $I,O,I'$ are collinear. Claim: $X\in IO$. Proof. Applying Pascal's Theorem on $(C_2B_1M_BB_2C_1M_C)$, we get that $C_2B_1\cap B_2C_1=X$, $B_1M_B\cap C_1M_C=I$, $M_BB_2\cap M_CC_2=I'$ are collinear. Hence, $X,I,I'$ are collinear. Since $I,O,I'$ are collinear as well, the claim follows. Claim: $X\in EF$. Proof. Applying Pascal's Theorem on $(B_1C_2CFEM_B)$, we get that $B_1C_2\cap FE$, $C_2C\cap EM_B=O$, $CF\cap M_BB_1=I$ are collinear. Hence, $B_1C_2\cap FE\in IO$. However, $X=B_1C_2\cap IO$, hence $X\in EF$, as claimed. Note that $E$ is the center of circle $(AIC)$ and $F$ is the center of $(AIB)$. Hence, $EF$ is the perpendicular bisector of $\overline{AI}$. Since $X\in EF$, we have that $XA=XI$, as desired. $\blacksquare$
Attachments:

01.10.2024 17:32
solved with Kanad, AN1729, Om245