Let $ABCD$ be a convex quadrilateral, and let $P$, $Q$, $R$, and $S$ be points on the sides $AB$, $BC$, $CD$, and $DA$, respectively. Let the line segment $PR$ and $QS$ meet at $O$. Suppose that each of the quadrilaterals $APOS$, $BQOP$, $CROQ$, and $DSOR$ has an incircle. Prove that the lines $AC$, $PQ$, and $RS$ are either concurrent or parallel to each other.
Problem
Source: 2015 IMO Shortlist G7
Tags: IMO Shortlist, geometry
08.07.2016 00:01
By ELMO 2011 Problem 1, quadrilateral $ABCD$ has incircle $\omega$. Now construct parallelograms $WPOS$, $XPOQ$. $YQOR$, $ZROS$. Now $AP-AS = WS-WP$, so we can draw a circle $\omega_A$ tangent to all four lines. Define $\omega_B$, $\omega_C$, $\omega_D$ similarly. Let $K$ be the exsimilicenter of $\omega_A$ and $\omega_C$. We claim this is the desired concurrency point. By Monge on $\omega_A$, $\omega_C$, $\omega$, we get $KAC$ collinear. By Monge on $\omega_A$, $\omega_B$, $\omega_C$ we get $KPQ$ collinear. By Monge on $\omega_A$, $\omega_D$, $\omega_C$ we get $KRS$ collinear. Thus done. Degenerate cases left as annoying exercise to reader. Also, I heard that you can use straight Cartesian coordinates on this?
08.07.2016 00:10
The first step is to show that $ABCD$ also has an incircle.
We have that $PR=PM+MU+UR=PE+RL+OS$ and $PR=PN+NT+TR=PH+RI+OQ$, so $2PR=EH+LI+OQ+OS=PR+QS$. This means that $PR=QS$, so $EH+IL=FK+GJ\implies AB+CD=AD+BC$, as desired. Now, let $\omega$ be the incircle of $ABCD$, and let $Z$ be the exsimilicenter of $\omega_1$ and $\omega_3$. By Monge on $\omega_1$, $\omega_3$, and $\omega$, $Z$ lies on $AC$. Let $\ell_1$ and $\ell_2$ be the common external tangents of $\omega_1$ and $\omega_3$. Now, by the dual of Desargues' Involution Theorem on $Z$ and $APOS$, $ZA,ZO;ZP,ZS;\ell_1,\ell_2$ are pairs of an involution. Similarly, $ZC,ZO;ZQ,ZR;\ell_1,\ell_2$ are pairs of an involution. Since these two involutions share the pairs $ZAC,ZO$ and $\ell_1,\ell_2$, they must be the same involution $\Phi$. Let $\heartsuit$ (sorry out of letters) be the intersection of $PQ$ and $RS$. By the dual of Desargues' Involution Theorem on $Z$ and $PQSR$, $ZP,ZR;ZQ,ZS;ZO,Z\heartsuit$ are pairs of an involution. But they share the pairs $ZP,ZR$ and $ZQ,ZS$ with $\Phi$, so $ZO,Z\heartsuit$ is a pair of $\Phi$. But $ZO,ZAC$ is a pair of $\Phi$, so the result follows.
08.07.2016 13:50
On an inspection,found out that my solution is almost the same solution as the first one, so not posting it again. Just wanted to remark that this problem is quite easy for a G7. But the configuration looks quite pretty and congrats to the problem creator(s) for finding such a beautiful property.
18.08.2016 00:17
Yes, it is easy after one finds the construction of $W$, $X$, $Y$, and $Z$. However, how would one be motivated to construct these points and notice that it is the concurrency point of $AC, PQ$, and $RS$? Thank you very much!
18.08.2016 15:30
Dear Mathlinkers, jus have a look at http://jl.ayme.pagesperso-orange.fr/Docs/Demir.pdf p. 26-27... Sincerely Jean-Louis
18.08.2016 17:20
The motivation comes from just length chasing all the way.Doing ELMO problems helps very much.!! This was how I came to the solution after making a very nice diagram.
07.09.2016 10:55
jayme wrote: Dear Mathlinkers, jus have a look at http://jl.ayme.pagesperso-orange.fr/Docs/Demir.pdf p. 26-27... Sincerely Jean-Louis Yeah,I have looked at .. but unfortunately , I understood nothing....Is there any copy of English version ?
30.09.2016 20:40
Hallow ^^^^^^^^^^
30.12.2016 07:45
v_Enhance wrote: Also, I heard that you can use straight Cartesian coordinates on this? Possibly. It is doable with complex numbers if you set it up correctly: Let $O$ be the origin, and let the internal and external angle bisectors of $\angle SOP$ be the real and imaginary axes. Then set the coordinates of the incenters as your variables, and let the equation $SO$ and $OP$ be $z/\overline{z} = t, z/\overline{z} = 1/t$ (so five variables in total). $DA$ is the reflection of $RP$ across the line joining the incenters of $ASOP$ and $DROS$ etc. so we can compute $A$ and the equation of $DA$. Intersect $DA$ with $SO$, which ends up being surprisingly easy to get $S$. Then compute $AS$. Similarly compute $AP, BP, BQ, CR, CQ, DS, DR$. Now verify that $\frac{AS}{AP}\cdot \frac{BP}{BQ} \cdot \frac{CR}{CQ} \cdot \frac{DS}{DR} = 1$, so we are done by two Menelaus'. It is probably doable with Cartesian with the same setup but would be messier I think.
11.07.2017 00:33
ABCDE wrote: But they share the pairs $ZP,ZR$ and $ZQ,ZS$ with $\Phi$ But why are $ZP,ZR$ and $ZQ,ZS$ (reciprocal)pairs of $\Phi$? $ZP,ZS$ and $ZQ,ZR$ indeed are but i don't see how this implies the statement. Im very sorry if this is a silly question...
28.05.2018 05:42
The trig bash is fairly straightforward. [asy][asy] unitsize(75); pair A, B, C, D, P, Q, R, S, O, Ia, Ib, Ic, Id, X = (-0.8, 1), Y = reflect((0, 0), (1, 0)) * X; real ra, rb, rc, rd; O = origin; Ia = 0.7 * (0, 1); Ib = 0.6 * (-1, 0); Ic = 0.9 * (0, -1); Id = 1.0 * (1, 0); ra = abs(foot(Ia, O, X)-Ia); rb = abs(foot(Ib, O, X)-Ib); rc = abs(foot(Ic, O, X)-Ic); rd = abs(foot(Id, O, X)-Id); P = extension(O, X, reflect(Ia, Ib) * O, reflect(Ia, Ib) * Y); Q = extension(O, Y, reflect(Ib, Ic) * O, reflect(Ib, Ic) * X); R = extension(O, X, reflect(Ic, Id) * O, reflect(Ic, Id) * Y); S = extension(O, Y, reflect(Id, Ia) * O, reflect(Id, Ia) * X); A = extension(P, reflect(P, Ia) * O, S, reflect(S, Ia) * O); B = extension(Q, reflect(Q, Ib) * O, P, reflect(P, Ib) * O); C = extension(R, reflect(R, Ic) * O, Q, reflect(Q, Ic) * O); D = extension(S, reflect(S, Id) * O, R, reflect(R, Id) * O); draw(circle(Ia, ra)^^circle(Ib, rb)^^circle(Ic, rc)^^circle(Id, rd), gray(0.6)); draw(A--B--C--D--cycle, gray(0.4)); draw(P--R^^Q--S); draw(Ia--P--Ib--Q--Ic--R--Id--S--cycle); draw(Ia--Ib--Ic--Id--cycle); draw(Ia--Ic^^Ib--Id); dot(A^^B^^C^^D^^P^^Q^^R^^S^^O^^Ia^^Ib^^Ic^^Id); label("$A$", A, dir(A - Ia)); label("$B$", B, dir(B - Ib)); label("$C$", C, dir(C - Ic)); label("$D$", D, dir(D - Id)); label("$P$", P, rotate(90, O) * dir(A - B)); label("$Q$", Q, rotate(90, O) * dir(B - C)); label("$R$", R, rotate(90, O) * dir(C - D)); label("$S$", S, rotate(90, O) * dir(D - A)); label("$O$", O, dir(290)); label("$I_A$", Ia, dir(90)); label("$I_B$", Ib, dir(180)); label("$I_C$", Ic, dir(270)); label("$I_D$", Id, dir(0)); label("$\alpha$", Ia, 4*dir(dir(Ib-Ia)+dir(O-Ia))); label( "$\beta$", Ib, 3*dir(dir(Ic-Ib)+dir(O-Ib))); label("$\gamma$", Ic, 3*dir(dir(Id-Ic)+dir(O-Ic))); label("$\delta$", Id, 5*dir(dir(Ia-Id)+dir(O-Id))); label("$\kappa$", O, 3*dir(dir(Ia-O)+dir(S-O))); label("$\overline{\alpha}$", Ib, 3*dir(dir(Ia-Ib)+dir(O-Ib))); label("$\overline{\alpha}$", P, 3*dir(dir(Ia-P)+dir(O-P))); label("$\delta$", S, 4*dir(dir(Ia-S)+dir(O-S))); label("$\kappa$", O, 3*dir(dir(Ia-O)+dir(P-O))); label("$\overline{\kappa}$", O, 3*dir(dir(Id-O)+dir(S-O))); [/asy][/asy] By two applications of Menelaus, the conclusion is equivalent to \[\frac{AP}{PB} \cdot \frac{BQ}{QC} \cdot \frac{CR}{RD} \cdot \frac{DS}{SA} = 1 \iff \frac{AS}{AP} \cdot \frac{BP}{BQ} \cdot \frac{CQ}{CR} \cdot \frac{DR}{DS} = 1.\]Let $I_A$, $I_B$, $I_C$, $I_D$ denote the incenters of $APOS$, $BQOP$, $CROQ$, $DSOR$ respectively. Introduce $\alpha = \angle OI_AI_B$, $\beta = \angle OI_BI_C$, $\gamma = \angle OI_CI_D$, $\delta = \angle OI_DI_A$, and $\kappa = \angle I_AOS$. For convenience, let $\overline{\phi}$ denote the complement of any angle $\phi$. Since $\angle I_AOI_B = \angle I_API_B = 90^{\circ}$, quadrilateral $I_AOI_BP$ is cyclic, so $\angle I_APO = \angle I_AI_BO = \overline{\alpha}$; similarly $\angle I_ASO = \delta$. It follows that \[\angle I_AAP = \angle I_AAS = 180^{\circ} - \overline{\alpha} - \delta - \kappa.\]Letting $r_A$ denote the inradius of $APOS$, observe $AP = r_A(\cot \overline{\alpha} + \cot (180^{\circ} - \overline{\alpha} - \delta - \kappa))$ and $AS = r_A(\cot \delta + \cot (180^{\circ} - \overline{\alpha} - \delta - \kappa))$, so \begin{align*} \frac{AS}{AP} & = \frac{\cot \overline{\alpha} + \cot (180^{\circ} - \overline{\alpha} - \delta - \kappa)}{\cot \delta + \cot (180^{\circ} - \overline{\alpha} - \delta - \kappa)}\\ & = \frac{\cot \overline{\alpha} - \cot (\overline{\alpha} + \delta + \kappa)}{\cot \delta - \cot (\overline{\alpha} + \delta + \kappa)}\\ & = \frac{\cos \overline{\alpha} \sin (\overline{\alpha} + \delta + \kappa) - \sin \overline{\alpha} \cos (\overline{\alpha} + \delta + \kappa)}{\sin \overline{\alpha} \sin (\overline{\alpha} + \delta + \kappa)} \div \frac{\cos \delta \sin (\overline{\alpha} + \delta + \kappa) - \sin \delta \cos (\overline{\alpha} + \delta + \kappa)}{\sin \delta \sin (\overline{\alpha} + \delta + \kappa)}\\ & = \frac{\sin(\delta + \kappa)}{\sin \overline{\alpha} \sin (\overline{\alpha} + \delta + \kappa)} \div \frac{\sin(\overline{\alpha} + \kappa)}{\sin \delta \sin (\overline{\alpha} + \delta + \kappa)}\\ & = \frac{\sin(\delta + \kappa)}{\sin(\overline{\alpha} + \kappa)} \cdot \frac{\sin \delta}{\cos \alpha}. \end{align*}Similarly, \[\frac{BP}{BQ} = \frac{\sin(\alpha + \overline{\kappa})}{\sin(\overline{\beta} + \overline{\kappa})} \cdot \frac{\sin \alpha}{\cos \beta}, \frac{CQ}{CR} = \frac{\sin(\beta + \kappa)}{\sin(\overline{\gamma} + \kappa)} \cdot \frac{\sin \beta}{\cos \gamma}, \frac{DR}{DS} = \frac{\sin(\gamma + \overline{\kappa})}{\sin(\overline{\delta} + \overline{\kappa})} \cdot \frac{\sin \gamma}{\cos \delta}.\]Since $\overline{\alpha} + \kappa$ and $\alpha + \overline{\kappa}$ are supplementary, $\sin(\overline{\alpha} + \kappa) = \sin(\alpha + \overline{\kappa})$. Recognizing similar equalities, \begin{align*} \frac{AS}{AP} \cdot \frac{BP}{BQ} \cdot \frac{CQ}{CR} \cdot \frac{DR}{DS} & = \tan \alpha \tan\beta \tan \gamma \tan \delta\\ & = \frac{OI_B}{OI_A} \cdot \frac{OI_C}{OI_B} \cdot \frac{OI_D}{OI_C} \cdot \frac{OI_A}{OI_D}\\ & = 1 \end{align*}as desired.
25.09.2018 02:24
rmtf1111 wrote: ABCDE wrote: But they share the pairs $ZP,ZR$ and $ZQ,ZS$ with $\Phi$ But why are $ZP,ZR$ and $ZQ,ZS$ (reciprocal)pairs of $\Phi$? $ZP,ZS$ and $ZQ,ZR$ indeed are but i don't see how this implies the statement. Im very sorry if this is a silly question... I think it's just typos in that solution. When I apply DDIT I get the following: Through $Z$ to $APOS$ one gets $ZAC \leftrightarrow ZO$, $ZS \leftrightarrow ZP$, and the two tangents. Through $Z$ to $CROQ$ one gets $ZAC \leftrightarrow ZO$, $ZQ \leftrightarrow ZR$, and the two tangents. Through $Z$ to $PQRSOK$ one gets $Z\heartsuit \leftrightarrow ZO$, $ZQ \leftrightarrow ZR$, $ZS \leftrightarrow ZP$.
18.12.2019 23:54
Here's a straightforward solution using Monge and Pappus. Let $\omega_A$, $\omega_B$, $\omega_C$, $\omega_D$ be the incircles of $APOS$, $BQOP$, $CROQ$, $DSOR$. Claim: $ABCD$ has an incircle $\omega$. Proof. Verify that $AB+CD = AD+BC$ by many applications of Pitot's theorem, as in ELMO 2011. $\blacksquare$ [asy][asy] size(350); defaultpen(fontsize(10pt)); pair A, B, C, D, P, Q, R, S, O, Ia, Ib, Ic, Id, I, E, F, G, H, X, Y, Z; pair M = (-0.8, 1), N = reflect((0, 0), (1, 0)) * M; real ra, rb, rc, rd; O = origin; Ia = 0.6 * (0, 1); Ib = 0.6 * (-1, 0); Ic = 0.9 * (0, -1); Id = 1.0 * (1, 0); ra = abs(foot(Ia, O, M)-Ia); rb = abs(foot(Ib, O, M)-Ib); rc = abs(foot(Ic, O, M)-Ic); rd = abs(foot(Id, O, M)-Id); P = extension(O, M, reflect(Ia, Ib) * O, reflect(Ia, Ib) * N); Q = extension(O, N, reflect(Ib, Ic) * O, reflect(Ib, Ic) * M); R = extension(O, M, reflect(Ic, Id) * O, reflect(Ic, Id) * N); S = extension(O, N, reflect(Id, Ia) * O, reflect(Id, Ia) * M); A = extension(P, reflect(P, Ia) * O, S, reflect(S, Ia) * O); B = extension(Q, reflect(Q, Ib) * O, P, reflect(P, Ib) * O); C = extension(R, reflect(R, Ic) * O, Q, reflect(Q, Ic) * O); D = extension(S, reflect(S, Id) * O, R, reflect(R, Id) * O); E = extension(A, B, Q, S); F = extension(B, C, P, R); G = extension(C, D, Q, S); H = extension(D, A, P, R); X = extension(E, F, G, H); Y = extension(E, H, F, G); Z = extension(F, S, P, G); I = extension(B, incenter(A, B, C), D, incenter(A, D, C)); draw(circle(Ia, ra)^^circle(Ib, rb)^^circle(Ic, rc)^^circle(Id, rd), mediumblue+linewidth(0.4)); draw(A--B--C--D--cycle, heavygreen); draw(P--R^^Q--S, heavygreen); draw(A--E--S^^Q--G--C, heavygreen); draw(A--H--P^^B--F--P, heavygreen); draw(G--X^^E--F, dashed+lightblue); draw(E--Y^^F--G, dashed+lightblue); draw(CP(I, foot(I, A, B)), lightblue); draw(C--X^^D--Y, heavycyan+dashed); draw(F--S^^G--Z, purple+dotted); dot("$A$", A, dir(120)); dot("$B$", B, dir(260)); dot("$C$", C, dir(300)); dot("$D$", D, dir(D)); dot("$O$", O, dir(270)); dot("$P$", P, dir(90)); dot("$Q$", Q, dir(180)); dot("$R$", R, dir(R)); dot("$S$", S, dir(100)); dot("$E$", E, dir(E)); dot("$F$", F, dir(F)); dot("$G$", G, dir(G)); dot("$H$", H, dir(90)); dot("$X$", X, dir(X)); dot("$Y$", Y, dir(Y)); dot(Z); [/asy][/asy] Let $E = \overline{AB}\cap \overline{QS}$, $F = \overline{BC}\cap \overline{PR}$, $G = \overline{CD}\cap\overline{QS}$, $H = \overline{DA}\cap\overline{PR}$; furthermore, let $X$ be the exsimilicenter of $\omega_A$ and $\omega_C$ and $Y$ be the exsimilicenter of $\omega_B$ and $\omega_D$. Now by Monge, we have: $\overline{XAC}$ collinear from $\omega$, $\omega_A$, and $\omega_C$; similarly, $\overline{YBD}$ collinear. $\overline{EXF}$ collinear from $\omega_A$, $\omega_B$, and $\omega_D$. Similarly, $\overline{GHX}$, $\overline{EHY}$, $\overline{FYG}$ all collinear. Let $Z = \overline{PQ}\cap \overline{RS}$; we will show that $Z$ lies on $\overline{AC}$. Pappus is actually enough to finish: From $\overline{PRF}$ and $\overline{SQG}$, we get that $\overline{FS}\cap \overline{GP}$ lies on $\overline{ZC}$. From $\overline{FHP}$ and $\overline{GES}$, we get that $\overline{FS}\cap \overline{GP}$ lies on $\overline{AX}$. Since $\overline{XAC}$ collinear, it follows that $\overline{ZAC}$ collinear, as desired.
08.09.2020 19:07
Fun problem. Posting for storage. $\textbf{Claim 01.}$ $ABCD$ has an incircle. $\textit{Proof.}$ To prove this let the incircle of $APOS$ tangent to $AD$ at $A_D$ and $AB$ at $A_B$. Define others similarly. Then, we have \begin{align*} (AB + CD) - (AD + BC) &= (AP - AS) + (BP - BQ) + (CR - CQ) + (RD - RS) \\ &= 2(OP - OS) + 2(OR - OQ) \\ &= 2(PR - QS) \end{align*}It suffices to prove $PR = QS$. However, \[ PR = PA_B + A_D D_A + RD_C = PB_A + B_C C_B + RC_D \]Therefore, we have \[ PR = \frac{1}{2} (A_B B_A + B_C C_B + C_D D_C + D_A A_D) \]Similarly, we will then get $PR = QS$. Now, let the incircle of $APOS$, $BPOQ, CQOR, RODS$ be $\omega_1, \omega_2, \omega_3, \omega_4$ respectively. Let the incircle of $ABCD$ be $\omega$. Monge Theorem on $\omega_1, \omega_3, \omega$ gives us $A, C$ and the exsimilicenter of $\omega_1$ and $\omega_3$ collinear. Let the exsimilicenter of $\omega_1$ and $\omega_3$ be $N$. Furthermore, let the two tangents of $\omega_1$ and $\omega_3$ be $\ell_a, \ell_b$. By Dual Desargues Involution Theorem on $N,APOS$, we have an involution $\varphi$ swapping $(NA, NO); (NP,NS); (\ell_a, \ell_b)$. Similarly, by DDIT on $N,CROQ$, we have an involution $\varphi'$ swapping $(NO, NC); (NQ, NR); (\ell_a, \ell_b)$. However, $NC \equiv NA$. Therefore, $\varphi = \varphi'$ as both of them swaps $(\ell_a, \ell_b)$ and $(NO, NA \equiv NC)$. Now, let $PQ \cap RS = M$. We have DDIT on $N,PQSR$ would gives us an involution $\varphi''$ which swaps $(NP,NS), (NR, NQ), (NO,NM)$. Now, notice that involution $\varphi$ swaps $(NQ,NR)$ and $(NP,NS)$. Therefore, $\varphi = \varphi''$. This means that $\varphi$ swaps $(NO,NM)$ as well. However, we know that $\varphi$ swaps $(NO, AC)$. Therefore, $N,M,A,C$ collinear, which proves that $M$ lies on $AC$.
22.12.2020 09:07
When you realize that Menelaus is sufficient to solve a G7... [asy][asy] size(8cm); real labelscalefactor = 0.5; /* changes label-to-point distance */pen dps = linewidth(0.7) + fontsize(0); defaultpen(dps); /* default pen style */ pen dotstyle = black; /* point style */ real xmin = -28.457435948614894, xmax = 24.665018487480715, ymin = -27.96024239952826, ymax = 12.618241020753967; /* image dimensions */pen qqwuqq = rgb(0,0.39215686274509803,0); pen zzttff = rgb(0.6,0.2,1); /* draw figures */draw(circle((-5.400054273068851,-7.700989482364915), 4.345816949812666), linewidth(0.8)); draw((-16.244173370203114,-12.01688134684045)--(4.726075344215098,-0.48369824585354604), linewidth(0.8) + blue); draw((-4.552623445137903,-14.206195513368455)--(-11.70228968055285,3.7580711078066984), linewidth(0.8) + qqwuqq); draw((-1.7999789027426196,-10.13524137755767)--(-9.412652225566063,-6.032102837654061), linewidth(0.8) + zzttff); draw((-8.2563457259138,-10.976309939615467)--(-6.526533602176052,-3.503708304939073), linewidth(0.8) + zzttff); draw((-16.244173370203114,-12.01688134684045)--(-6.526533602176052,-3.503708304939073), linewidth(0.8) + blue); draw((-8.59157022251616,-4.057929232327401)--(-9.412652225566063,-6.032102837654061), linewidth(0.8)); draw((-8.59157022251616,-4.057929232327401)--(-6.526533602176052,-3.503708304939073), linewidth(0.8)); draw((-9.412652225566063,-6.032102837654061)--(-10.6139120965913,-8.920359457220973), linewidth(0.8)); draw((-10.6139120965913,-8.920359457220973)--(-4.552623445137903,-14.206195513368455), linewidth(0.8)); draw((-4.552623445137903,-14.206195513368455)--(4.726075344215098,-0.48369824585354604), linewidth(0.8)); draw((-6.526533602176052,-3.503708304939073)--(4.726075344215098,-0.48369824585354604), linewidth(0.8)); draw((-16.244173370203114,-12.01688134684045)--(-1.7999789027426196,-10.13524137755767), linewidth(0.8) + blue); draw((-8.2563457259138,-10.976309939615467)--(-11.70228968055285,3.7580711078066984), linewidth(0.8) + qqwuqq); draw((-11.70228968055285,3.7580711078066984)--(-1.7999789027426196,-10.13524137755767), linewidth(0.8) + qqwuqq); /* dots and labels */dot((-6.526533602176052,-3.503708304939073),dotstyle); label("$A$", (-6.358158152299367,-2.6618310555556275), NE * labelscalefactor); dot((-9.412652225566063,-6.032102837654061),dotstyle); label("$B$", (-10.651732124154956,-5.5663075659285255), NE * labelscalefactor); dot((-8.2563457259138,-10.976309939615467),dotstyle); label("$C$", (-9.01007148785723,-12.511794873341977), NE * labelscalefactor); dot((-1.7999789027426196,-10.13524137755767),dotstyle); label("$D$", (-1.180613068591158,-10.533383337290873), NE * labelscalefactor); dot((-8.59157022251616,-4.057929232327401),linewidth(4pt) + dotstyle); label("$E$", (-8.420757413288817,-3.7141776172849386), NE * labelscalefactor); dot((-10.6139120965913,-8.920359457220973),linewidth(4pt) + dotstyle); label("$F$", (-10.441262811809093,-8.59706566370894), NE * labelscalefactor); dot((-4.552623445137903,-14.206195513368455),linewidth(4pt) + dotstyle); label("$G$", (-4.463934341186608,-15.458365246184048), NE * labelscalefactor); dot((4.726075344215098,-0.48369824585354604),linewidth(4pt) + dotstyle); label("$H$", (4.880903126969672,-0.1361993074052816), NE * labelscalefactor); dot((-11.70228968055285,3.7580711078066984),linewidth(4pt) + dotstyle); label("$I$", (-11.535703236007576,4.115280801981134), NE * labelscalefactor); dot((-16.244173370203114,-12.01688134684045),linewidth(4pt) + dotstyle); label("$J$", (-16.839529907123303,-11.459448311612666), NE * labelscalefactor); dot((-7.367051894868528,-7.134656592188692),linewidth(4pt) + dotstyle); label("$K$", (-7.200035401682816,-6.787029577534526), NE * labelscalefactor); clip((xmin,ymin)--(xmin,ymax)--(xmax,ymax)--(xmax,ymin)--cycle); [/asy][/asy] Lemma. Suppose $EFGH$ is a circumscribed quadrilateral and the touching points on side $EF,FG,GH,HE$ are $B,C,D,A$ respectively, then $$\frac{EK}{KG}=\frac{BE}{CG}$$Proof. By applying Pascal's theorem to $BBDCCA$ and $AACDDB$ we obtain that $J,F,K,H$ are collinear, by symmetry $I,E,K,G$ are collinear. Moreover, $$(I,K;E,G)\overset{B}{=}(BC,BD;BB,BG)=-1$$Therefore, by Menelaus theorem, $$\frac{EK}{KG}=\frac{EI}{IG}=\frac{EB}{BF}\frac{FC}{CG}=\frac{BE}{CG}$$$\blacksquare$ [asy][asy] size(10cm); real labelscalefactor = 0.5; /* changes label-to-point distance */pen dps = linewidth(0.7) + fontsize(0); defaultpen(dps); /* default pen style */ pen dotstyle = black; /* point style */ real xmin = -27.356605005249566, xmax = 21.147225047890352, ymin = -21.167503371439867, ymax = 20.034674845743506; /* image dimensions */ /* draw figures */draw(circle((-5.095929724659887,2.027115961420098), 2.2041302756648538), linewidth(0.8)); draw(circle((-2.924579145174239,-5.056135709280694), 4.725512814009619), linewidth(0.8)); draw(circle((2.0550590444698473,4.166553169351109), 4.797167263167613), linewidth(0.8)); draw(circle((5.938980623808618,-3.9310012117447393), 3.035012815620033), linewidth(0.8)); draw((-7.9865872209071895,-11.687338507312862)--(9.581370927343654,-5.931370365143414), linewidth(0.8)); draw((-7.9865872209071895,-11.687338507312862)--(-7.245955629908703,3.1671768111928897), linewidth(0.8)); draw((-7.245955629908703,3.1671768111928897)--(4.92312734479103,12.332046685258797), linewidth(0.8)); draw((4.92312734479103,12.332046685258797)--(9.581370927343654,-5.931370365143414), linewidth(0.8)); draw((-7.405529462161146,-0.03332390257078012)--(8.334675708951755,-1.0434948591234776), linewidth(0.8)); draw((-4.659903153931072,5.11480379994105)--(4.819056360177046,-7.4916957427587825), linewidth(0.8)); /* dots and labels */dot((-0.4534949668144329,-0.47948983622964875),dotstyle); label("$O$", (-0.27964611357737224,-0.04486770313699714), NE * labelscalefactor); dot((-5.095929724659887,2.027115961420098),dotstyle); label("$O_1$", (-4.930102937668744,2.475940668800382), NE * labelscalefactor); dot((5.938980623808618,-3.9310012117447393),dotstyle); label("$O_3$", (6.109299242884607,-3.478382554568945), NE * labelscalefactor); dot((-2.924579145174239,-5.056135709280694),dotstyle); label("$O_2$", (-4.712791871122419,-4.999560020393226), NE * labelscalefactor); dot((2.0550590444698473,4.166553169351109),dotstyle); label("$O_4$", (2.806171031380454,3.997118134624663), NE * labelscalefactor); dot((-7.245955629908703,3.1671768111928897),linewidth(4pt) + dotstyle); label("$A$", (-7.059751389822737,3.519033788222746), NE * labelscalefactor); dot((-7.9865872209071895,-11.687338507312862),linewidth(4pt) + dotstyle); label("$B$", (-7.7986090160802455,-11.34504316354594), NE * labelscalefactor); dot((9.581370927343654,-5.931370365143414),linewidth(4pt) + dotstyle); label("$C$", (9.76012516086288,-5.564568793413673), NE * labelscalefactor); dot((4.92312734479103,12.332046685258797),linewidth(4pt) + dotstyle); label("$D$", (5.109668336771508,12.689560796477695), NE * labelscalefactor); dot((-7.405529462161146,-0.03332390257078012),linewidth(4pt) + dotstyle); label("$P$", (-7.233600243059798,0.3028300033371242), NE * labelscalefactor); dot((4.819056360177046,-7.4916957427587825),linewidth(4pt) + dotstyle); label("$Q$", (4.979281696843712,-7.1292084725472185), NE * labelscalefactor); dot((8.334675708951755,-1.0434948591234776),linewidth(4pt) + dotstyle); label("$R$", (8.49972097489419,-0.6968009027759746), NE * labelscalefactor); dot((-4.659903153931072,5.11480379994105),linewidth(4pt) + dotstyle); label("$S$", (-4.495480804576093,5.474833387139679), NE * labelscalefactor); clip((xmin,ymin)--(xmin,ymax)--(xmax,ymax)--(xmax,ymin)--cycle); [/asy][/asy] We now return to the original problem. For a point $X$ and a circle $\Gamma$ let $f(X,\omega)$ be the length of tangent from $X$ to $\omega$. Let the incircle of $ASOP$ be $\omega_A$, define $\omega_B$, $\omega_C$ and $\omega_D$ similarly. By Menelaus theorem it suffices to show that $$\frac{AP}{AS}\cdot\frac{BQ}{BP}\cdot\frac{RC}{QC}\cdot\frac{SD}{DR}=1 \hspace{20pt}(1)$$Now by Ceva's theorem and the lemma, $$\frac{AP}{AS}=\frac{\sin\angle AOP}{\sin\angle SOA}\cdot\frac{\sin\angle APO}{\sin\angle ASO}=\frac{f(P,\omega_A)}{f(S,\omega_A)}\cdot\frac{OS}{OP}\cdot\frac{\sin\angle APO}{\sin\angle ASO}$$multiplying similar expressions, it suffices to show $$\prod_{cyc}\frac{f(P,\omega_A)}{f(S,\omega_A)}=1 \hspace{20pt}(2)$$Indeed, extend $AP$ and $SO$ to meet at $V$ then $\omega_A$ is the incircle of $\triangle VPO$ while $\omega_B$ is the $V-$excircle of $\triangle VOQ$, therefore, $$f(P,\omega_A)=f(O,\omega_B)=f(Q,\omega_C)$$By symmetry, $$f(R,\omega_D)=f(O,\omega_C)=f(Q,\omega_B)$$$$f(P,\omega_B)=f(O,\omega_A)=f(S,\omega_D)$$$$f(R,\omega_C)=f(O,\omega_D)=f(S,\omega_A)$$Therefore $(2)$ is true, this completes the proof.
24.02.2021 00:03
Every time I try to do geo with the intention of getting better at things which are not projective spam, I end up using projective spam. Let $\omega_A$ be the incircle of $APOS$, etc. Let $X_{AB}$ be the exsimilicenter of $\omega_A$ and $\omega_B$, etc. Let $X = X_{AC}$ for convenience. By Monge's theorem, $X = X_{AB}X_{BC} \cap X_{AD}X_{DC}$. Lemma: $X \in AC$ Proof 1 (DDIT): By DDIT on point $X_{AC}$ and quadrilateral $APOS$, there exists an involution swapping $(XA, XO), (XX_{AB}, XX_{AD})$, and the two common external tangents of $\omega_A$ and $\omega_C$. By DDIT on point $X_{AC}$ and quadrilateral $CQOR$, there exists an involution swapping $(XC, XO), (XX_{CD}, XX_{CB})$, and the two common external tangents of $\omega_A$ and $\omega_C$. But because $X = X_{AB}X_{BC} \cap X_{AD}X_{DC}$, these two involutions are the same, and because $(XA, XO)$ and $(XC, XO)$ are pairs of this involution, $XA = XC$, i.e. $X \in AC$. Proof 2 (Moving Points): Let $I_A$ be the center of $\omega_A$, etc. Fix $O, P, Q, A, B, C$ and vary $I_D$ projectively along the internal bisector of $\angle POQ$. Then $X_{AD} = I_AI_D \cap OP$ moves projectively, so line $X_{AD}S$, as the second tangent from $X_{AD}$ to $\omega_A$, has degree 2. Thus as $PB$ is also tangent to $\omega_A$, $A = BP \cap X_{AD}S $ has degree 1 (i.e. moves projectively.) Similarly $C$ has degree 1, so to show that $X, A, C$ are collinear, we only need to check the claim for $(0+1+1)+1 = 3$ choices of $I_D$. But for $I_D$ equal to the reflection of $I_B$ over $O$, it is obvious by symmetry, while for $I_D$ such that $\omega_A, \omega_C, \omega_D$ share a common tangent, it is again obvious. For $I_D = O$, we have $A = OS \cap BP = X_{AB}$, and similarly $C = X_{BC}$. Then the claim follows by Monge. Now, we use Pappus. By Pappus on $PQX_{BC}SRX_{CD}$, $PQ \cap RS, QX_{BC} \cap RX_{CD} = C, X_{BC}S \cap X_{CD}P$ are collinear, so it suffices to show $X_{BC}S \cap X_{CD}P \in AC$. By Pappus on $PX_{AB}X_{BC}SX_{AD}X_{CD}$, $PX_{AB} \cap SX_{AD} = A, X_{AB}X_{BC} \cap X_{AD}X_{CD} = X, X_{BC}S\cap X_{CD}P$ are collinear But we know $X \in AC$, so we are done. $\blacksquare$
12.09.2021 22:29
A projective type solution: As seen in many of the above solutions, we have that quad $ABCD$ has an incircle. Let $W = \overline{AB}\cap \overline{QS}$, $X = \overline{BC}\cap \overline{PR}$, $Y = \overline{CD}\cap\overline{QS}$, $Z = \overline{DA}\cap\overline{PR}$. Then by Monge's theorem, we obtain that lines $AC,WX,YZ$ concur (at the exsimilicenter of incircles of quad $APOS,CROQ$), in particular it suffices to prove the following lemma: Lemma: Let $A,B,C,D$ be any four points and $\ell_1,\ell_2$ be any two lines. Let $\ell_1$ and $\ell_2$ intersect lines $AB,BC,CD,DA$ at points $P,X,R,Z$ and $W,Q,Y,S$, respectively. Then $$\overline{AC},\overline{PQ},\overline{RS} \text{ concur} ~~ \iff ~~ \overline{AC},\overline{WX},\overline{YZ} \text{ concur}$$ proof: Because of symmetry, WLOG assume $\overline{AC},\overline{WX},\overline{YZ} \text{ concur}$. Consider the following argument: $\overline{AC},\overline{WX},\overline{YZ} \text{ concur}$. Then $\triangle AWY, \triangle CXZ$ are perspective. Points $B = \overline{AW} \cap \overline{CX} ~,~ \ell_1 \cap \ell_2 = \overline{WY} \cap \overline{XZ} ~,~ \overline{AY} \cap \overline{CZ}~$ lie on some line $\ell$. Pappus on $\{A,S,Z\},\{C,R,Y\}$ gives points $\overline{AR} \cap \overline{CS} ~,~ \ell_1 \cap \ell_2 = \overline{SY} \cap \overline{RZ} ~,~ \overline{AY} \cap \overline{CZ}$ are collinear. All these three points must lie on $\ell$ as the last two points lie on $\ell$. So points $B = \overline{AP} \cap \overline{CQ} ~,~ \ell_1 \cap \ell_2 ~,~ \overline{AR} \cap \overline{CS}~$ are collinear (as they all lie on $\ell$). Then $\triangle APR, \triangle CQS$ are perspective. Consequently, $ \overline{AC},\overline{PQ},\overline{RS} \text{ concur}$. This completes the proof of the lemma. $\blacksquare$
26.09.2021 21:49
Let \(I_A\), \(I_B\), \(I_C\), \(I_D\) be the incenters of \(APOS\), \(BQOP\), \(CROQ\), \(DSOR\). We begin with this estimate: Claim: We have \[\frac{SA}{AP}\cdot\frac{SO}{OP}=\left(\frac{SI_A}{PI_A}\right)^2\]and cyclic variations. Proof. This is immediate upon noting by law of sines that \[ \frac{AS}{\sin\frac{A+S}2}=\frac{SI_A}{\sin\frac A2},\quad \frac{AR}{\sin\frac{A+R}2}=\frac{RI_A}{\sin\frac A2},\quad \frac{OS}{\sin\frac{A+R}2}=\frac{SI_A}{\sin\frac O2},\quad \frac{RO}{\sin\frac{A+S}2}=\frac{RI_A}{\sin\frac O2}. \]\(\blacksquare\) Now we may observe that \begin{align*} \frac{AP}{PB}\cdot\frac{BQ}{QC}\cdot\frac{CR}{RD}\cdot\frac{DS}{SA} &=\prod_\mathrm{cyc}\left(\frac{SA}{AP}\cdot\frac{SO}{OP}\right) =\prod_\mathrm{cyc}\left(\frac{SI_A}{PI_A}\right)^2 =\prod_\mathrm{cyc}\left(\frac{PI_B}{PI_A}\right)^2\\ &=\prod_\mathrm{cyc}\left(\frac{\sin\angle POI_B}{\sin\angle POI_A}\right)^2 =\prod_\mathrm{cyc}\left(\frac{\sin\angle SOI_A}{\sin\angle POI_A}\right)^2 =1, \end{align*}so by Menelaus on \(\triangle ABC\) and \(\triangle ADC\) we have that \(\overline{PQ}\) and \(\overline{RS}\) intersect \(\overline{AC}\) at the same point.
10.01.2022 07:41
Let the incircles of $APOS,BQOP,CROQ,$ and $DSOR$ be $\omega_1,\omega_2,\omega_3,$ and $\omega_4,$ respectively. Let $\omega_1$ and $\omega_2$ touch $\overline{AB}$ at $A_1$ and $A_2.$ Let $\omega_1$ touch $\overline{PR}$ and $\overline{QS}$ at $P_1$ and $Q_1.$ Similarly define $B_1,B_2,$ etc and $P_2,Q_2,$ etc. WLOG let $A_1$ be closer to $P$ than $A_2$ and let $A_3$ be closer to $Q$ than $A_4.$ Notice $$2PP_1+P_1P_2=A_1A_2=Q_1Q_2=OP_2+OP_1=2OP_2+P_1P_2$$so $PP_1=OP_2.$ Hence, $$PR-QS=(PA_1+A_1A_3+RA_3)-(QQ_3+Q_3Q_1+SQ_1)=(PA_1-QQ_3)+(RA_3-SQ_1)=0.$$By Pitot, $DR=OR+SD-OS$ etc, so $$AB+CD=2(OR+OP-OS-OQ)+SD+CQ+BQ+AS=BC+AD+2(PR-QS)=BC+AD.$$Therefore, $ABCD$ has an incircle, $\omega.$ Let $W=\overline{AB}\cap\overline{QS},$ $X=\overline{BC}\cap\overline{PR},$ $Y=\overline{CD}\cap\overline{QS},$ $Z=\overline{DA}\cap\overline{PR},$ and let $T$ be the exsimilicenter of $\omega_1$ and $\omega_3.$ By Monge $\omega_1,\omega_3,$ and $\omega,$ we see $T,A,$ and $C$ are collinear. By Monge on $\omega_1,\omega_2,$ and $\omega_3,$ we see $T$ lies on $\overline{WX}$ and similarly $\overline{YZ}.$ Let $U=\overline{PQ}\cap\overline{RS}$ and $V=\overline{SX}\cap\overline{PY}.$ By Pappus on $\{X,P,Z\}$ and $\{W,Y,S\},$ we see $V,A,$ and $T$ are collinear. By Pappus on $\{P,R,X\}$ and $\{Y,S,Q\},$ we see $V,U$ and $C$ are collinear. Hence, we ascertain that $U,A,$ and $C$ are collinear. $\square$
10.01.2022 18:59
We easily see, that $ABCD$ has incircle, so exsimilicenter $K$ of incircles $APOS,CROQ$ lies on $AC$ by Monge. Let $QS$ meet $AB,CD$ at $W,X$ and $PR$ meet $AD,BC$ at $Y,Z$ respectively. By Monge $K=XZ\cap WY.$ Pappus on $(SWX),(PYZ)$ gives $SZ\cap PX\in AC,$ analogously $QZ\cap RX\in AC.$ Pappus on $(SQX),(PRZ)$ gives $PQ\cap RS\in \overline{(SZ\cap PX)(QZ\cap RX)}= AC,$ so done.
25.02.2022 16:30
Firstly, we prove that $ABCD$ has an incircle $\omega$. Let $X= AB \cap SQ, Y= BC \cap PR, Z= CD \cap SQ, W= AD \cap SR$. By Pitot Theorem and some length bash, we have that $ABCD$ has an incircle if and only if $PR=QS$. Let $P_A,P_B$ be the touching point of $P$ WRT $\omega_A, \omega_B$, respectively, where $\omega_A$ is the incircle of $APOS$ and $\omega_B, \omega_C, \omega_D$ are defined similarly. Observe that $OP_A=OS_A, OR_C=OQ_C$, so $PR=QS \iff PP_A+RR_C=SS_A+QQ_C$. However, observe that $\omega_B, \omega_D$ are the incircle and $Y-$excircle of $YOQ$, respectively, so $QQ_C=OQ_B$. Similarly, $SS_A=OS_D, RR_C= OR_D, PP_A= OP_B$, but $OP_B=OQ_B, OS_D=OR_D$, so we have the desired equality. Therefore, $ABCD$ has an incircle $\omega$. Let $U$ be the exsimilicenter of $\omega_A, \omega_C$. By Monge's Theorem on $\omega, \omega_A, \omega_C$, we know that $U$ lies on $AC$. Furthermore, by Monge's Theorem on $\omega_A, \omega_B, \omega_C$ and $\omega_C, \omega_D, \omega_A$ we know that $U= XY \cap ZW$. Now, by Pappus Theorem on $\{WYR\}, \{XZQ\}$, we have that $WZ \cap XY=U, WQ \cap XR, YQ \cap ZR=C$ are collinear. Moreover, since $U \in AC$, $WQ \cap XR \in AC$. Again, by Pappus Theorem on $\{WPR\}, \{XSQ\}$, we have that $WS \cap XP= A, WQ \cap XR, PQ \cap SR$ are collinear. But since $WQ \cap XR$ lies on $AC$, we have that $PQ \cap SR$ lies on $AC$. Therefore, $AC, PQ, SR$ are concurrent. $\blacksquare$
06.06.2023 06:05
Surprisingly, combining spartacle's start with the standard DDIT finish to this problem gives a solution that uses only DDIT and obvious applications of Monge's theorem, which is written up below. Let $\omega_A$, $\omega_B$, $\omega_C$, $\omega_D$ be the incircles of $\square ASOP$, $\square BPOQ$, $\square CQOR$, and $\square DROS$, respectively. Let $T_{AB} = AB\cap QS$, $T_{CD}=CD\cap QS$, $T_{DA}=AD\cap PR$, and $T_{BC}=BC\cap PR$, so that $T_{AB}$ is the exsimilicenter of the incircles of $\omega_A$ and $\omega_B$, and similarly for three other points. Thus, Monge's theorem implies that $X = T_{AB}T_{DA}\cap T_{BC}T_{CD}$ is the exsimilicenter of $\omega_A$ and $\omega_C$. Let $XU$ and $XV$ be the common external tangents of $\omega_A$ and $\omega_C$. Then, we apply DDIT at $X$ twice: On $APOS$, we get an involution $\Phi_1$ swapping $\{XA, XO\}$, $\{XP, XS\}$, $\{XT_{AB}, XT_{DA}\}$, and $\{XU, XV\}$. On $CQOR$, we get an involution $\Phi_2$ swapping $\{XC, XO\}$, $\{XQ, XR\}$, $\{XT_{BC}, XT_{CD}\}$, and $\{XU, XV\}$. From the last two pairs, we get that $\Phi_1=\Phi_2$. Thus, the first pair gives $X\in AC$. Finally, if $Z=PR\cap QS$, then by applying DDIT on $PQSR$, we get an involution $\Phi_3$ swapping $\{XP, XS\}$, $\{XQ, XR\}$, and $\{XO, XZ\}$. From the first two pairs, we get $\Phi_1=\Phi_2=\Phi_3$, so the last pair gives $Z\in AC$, as desired.
14.12.2023 12:25
Just spam pitot and monge
Claim 1: $ABCD$ has an incircle
Now we spam monge's to every triple of circles Let $\omega_1,\omega_2,\omega_3,\omega_4,\omega$ denote the incircles of $APOS,BPOQ,CQOR,DSOR$ and $ABCD$ respectively Let $G = SQ \cap CD$, $E = SQ \cap AB$, $F = PR \cap BC$, $H= PR \cap AD$, $X$ and $Y$ be the exsimilicenters of $\omega_1,\omega_3$ and $\omega_2,\omega_4$ respectively. By Monge's we get the following collinearities: $X-H-G$,$E-H-Y$,$F-Y-G$,$Y-B-D$,$X-A-C$,$F-X-E$ Now we apply Pappus's theorem to $FPR$ and $GSQ$ to get $FS \cap PG,PQ \cap RS,C$ collinear $HPF$ and $ESG$ to get $FS \cap PG, X, A$ collinear. Hence we're done.
13.02.2024 10:39
Absolute menace of a problem. Denote the incircles by $\omega_i$ in order $(APOS)$, $(BQOP)$, $(CROQ)$, and $(DSOR)$. [asy][asy] /* Geogebra to Asymptote conversion, documentation at artofproblemsolving.com/Wiki go to User:Azjps/geogebra */ import graph; size(15cm); 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 = -242.01839310183192, xmax = 300.6595372247246, ymin = -219.00412571765153, ymax = 217.92382753010781; /* image dimensions */ pen qqwuqq = rgb(0,0.39215686274509803,0); pen qqwwzz = rgb(0,0.4,0.6); pen qqqqcc = rgb(0,0,0.8); pen yqqqyq = rgb(0.5019607843137255,0,0.5019607843137255); /* draw figures */ draw(circle((-19.510087421793767,15.483380442716857), 12.420427186115836), linewidth(0.7) + qqwuqq); draw(circle((23.757733137651748,-22.790543966197177), 17.97873865459749), linewidth(0.7) + qqwuqq); draw(circle((-18.900985864109355,-19.45082012833311), 21.906757789886868), linewidth(0.7) + qqwuqq); draw(circle((15.735355764827185,19.704793976591137), 22.54588940582299), linewidth(0.7) + qqwuqq); draw((-29.98348329057844,24.484239246786483)--(-47.42866961804265,-41.75333159450325), linewidth(0.7) + blue); draw((-47.42866961804265,-41.75333159450325)--(43.22917793503198,-40.50229801939051), linewidth(0.7) + blue); draw((43.22917793503198,-40.50229801939051)--(35.66773566157664,52.89871075414429), linewidth(0.7) + blue); draw((35.66773566157664,52.89871075414429)--(-29.98348329057844,24.484239246786483), linewidth(0.7) + blue); draw((-34.97379312581012,5.536548174507169)--(40.55195611766193,-7.432519872606985), linewidth(0.7) + qqwwzz); draw((10.449861635083561,-40.95463641443643)--(-11.638625889740442,32.42406787818682), linewidth(0.7) + qqwwzz); draw((10.449861635083561,-40.95463641443643)--(55.33808465897373,-190.07483487934255), linewidth(0.7) + qqwwzz); draw((55.33808465897373,-190.07483487934255)--(43.22917793503198,-40.50229801939051), linewidth(0.7) + qqqqcc); draw((40.55195611766193,-7.432519872606985)--(219.00880851615148,-38.07662582869403), linewidth(0.7) + qqwwzz); draw((219.00880851615148,-38.07662582869403)--(43.22917793503198,-40.50229801939051), linewidth(0.7) + qqqqcc); draw((-34.97379312581012,5.536548174507169)--(-62.744011092931494,10.305171460027486), linewidth(0.7) + qqwwzz); draw((-62.744011092931494,10.305171460027486)--(-29.98348329057844,24.484239246786483), linewidth(0.7) + qqqqcc); draw((-29.98348329057844,24.484239246786483)--(-20.30758245566413,61.222635442471756), linewidth(0.7) + qqqqcc); draw((-20.30758245566413,61.222635442471756)--(-11.638625889740442,32.42406787818682), linewidth(0.7) + qqwwzz); draw(circle((2.065995514721633,-3.059151367356906), 38.007559614983926), linewidth(0.7) + qqwuqq); draw((-62.744011092931494,10.305171460027486)--(168.20097673778162,-151.43235197670788), linewidth(0.7) + linetype("2 2") + yqqqyq); draw((168.20097673778162,-151.43235197670788)--(-20.30758245566413,61.222635442471756), linewidth(0.7) + linetype("2 2") + yqqqyq); draw((-53.26139988871275,45.1466652419549)--(55.33808465897373,-190.07483487934255), linewidth(0.7) + linetype("2 2") + yqqqyq); draw((-53.26139988871275,45.1466652419549)--(219.00880851615148,-38.07662582869403), linewidth(0.7) + linetype("2 2") + yqqqyq); draw((-116.194992146551,101.009082617485)--(168.20097673778162,-151.43235197670788), linewidth(0.7) + yqqqyq); draw((-116.194992146551,101.009082617485)--(-207.04159900674418,181.64822758767713), linewidth(0.7) + yqqqyq); draw((-207.04159900674418,181.64822758767713)--(40.55195611766193,-7.432519872606985), linewidth(0.7) + dotted + qqqqcc); draw((-207.04159900674418,181.64822758767713)--(10.449861635083561,-40.95463641443643), linewidth(0.7) + dotted + qqqqcc); draw((-116.194992146551,101.009082617485)--(55.33808465897373,-190.07483487934255), linewidth(0.7) + dotted + qqwwzz); draw((-116.194992146551,101.009082617485)--(219.00880851615148,-38.07662582869403), linewidth(0.7) + dotted + qqwwzz); /* dots and labels */ dot((-29.98348329057844,24.484239246786483),linewidth(4pt) + dotstyle); label("$A$", (-47.54160628328455,22.478747874960526), NE * labelscalefactor); dot((-47.42866961804265,-41.75333159450325),linewidth(4pt) + dotstyle); label("$D$", (-59.922091385791546,-36.90782406893955), NE * labelscalefactor); dot((43.22917793503198,-40.50229801939051),linewidth(4pt) + dotstyle); label("$C$", (46.923006807082444,-37.29684924871013), NE * labelscalefactor); dot((35.66773566157664,52.89871075414429),linewidth(4pt) + dotstyle); label("$B$", (38.60593588832661,58.136056714495865), NE * labelscalefactor); dot((-11.638625889740442,32.42406787818682),linewidth(4pt) + dotstyle); label("$P$", (-11.368443883887988,38.501914884613484), NE * labelscalefactor); dot((10.449861635083561,-40.95463641443643),linewidth(4pt) + dotstyle); label("$R$", (-1.6306406948211165,-51.86757689560427), NE * labelscalefactor); dot((-34.97379312581012,5.536548174507169),linewidth(4pt) + dotstyle); label("$S$", (-49.12087401297351,-3.3456365038865656), NE * labelscalefactor); dot((40.55195611766193,-7.432519872606985),linewidth(4pt) + dotstyle); label("$Q$", (44.34373907739349,-3.7980758666453875), NE * labelscalefactor); dot((-20.30758245566413,61.222635442471756),linewidth(4pt) + dotstyle); label("$H$", (-22.264782532332774,67.87385990070176), NE * labelscalefactor); dot((219.00880851615148,-38.07662582869403),linewidth(4pt) + dotstyle); label("$G$", (226.3766266096827,-40.51879888916896), NE * labelscalefactor); dot((55.33808465897373,-190.07483487934255),linewidth(4pt) + dotstyle); label("$F$", (53.04983517458477,-203.01266579949268), NE * labelscalefactor); dot((-62.744011092931494,10.305171460027486),linewidth(4pt) + dotstyle); label("$E$", (-75.3976977639253,7.971434411054623), NE * labelscalefactor); dot((-116.194992146551,101.009082617485),linewidth(4pt) + dotstyle); label("$K$", (-114.08671370925964,104.9519010115018), NE * labelscalefactor); dot((-1.8318123804182076,-0.1544990220922823),linewidth(4pt) + dotstyle); label("$O$", (-1.042832889203694,5.129969868525217), NE * labelscalefactor); dot((168.20097673778162,-151.43235197670788),linewidth(4pt) + dotstyle); label("$N$", (174.27541846996576,-162.26023568547498), NE * labelscalefactor); dot((-53.26139988871275,45.1466652419549),linewidth(4pt) + dotstyle); label("$M$", (-54.76355592641363,49.818985799554675), NE * labelscalefactor); dot((-207.04159900674418,181.64822758767713),linewidth(4pt) + dotstyle); label("$T$", (-212.09888743744003,186.4567612395372), NE * labelscalefactor); clip((xmin,ymin)--(xmin,ymax)--(xmax,ymax)--(xmax,ymin)--cycle); /* end of picture */ [/asy][/asy] Claim: $ABCD$ has an incircle. Proof. Let the tangents from $\omega_1$ and $\omega_2$ meet $\overline{AB}$ at $W_1$ and $W_2$. Consider clockwise and label the tangents from $\overline{BC}$, $\overline{CD}$, $\overline{DA}$ by $X_i$, $Y_i$ and $Z_i$. Similarly denote the tangency points of $\omega_1$ and $\omega_2$ to $\overline{PR}$ by $T_1$ and $T_2$. Continuing, label the tangents at $\overline{QS}$ from $\omega_2$ and $\omega_3$ by $T_3$ and $T_4$, and so on. Then observe that, \begin{align*} AB + CD - AD - BC &= W_1W_2 + Y_1Y_2 - X_1X_2 - Z_1Z_2\\ &= \frac{1}{2} \left(2W_1W_2 + 2Y_1Y_2 - 2X_1X_2 - 2Z_1Z_2 \right) \end{align*}Note then that $W_1W_2 = T_3T_7$ due to the fact that they are both common external tangents to $\omega_1$ and $\omega_2$. Then we have, \begin{align*} \frac{1}{2} \left(2W_1W_2 + 2Y_1Y_2 - 2X_1X_2 - 2Z_1Z_2 \right) &= \frac{1}{2} \left(2T_3T_8 + 2T_4T_7 - 2T_2T_5 - 2T_6T_1 \right)\\ &= \frac{1}{2}(OT_1 + OT_2 + OT_5 + OT_6 - OT_3 - OT_4 - OT_7 - OT_8)\\ &= 0 \end{align*}whence the last step follows by equal tangents. Hence $AB + CD - AD - CB = 0 \iff AB + CD = AD + CB$ and hence $ABCD$ is tangential. $\square$ Let $\omega$ be the incircle of $ABCD$. Now define the following points. $E = \overline{AB} \cap \overline{QS}$. Similarly define the points $F$, $G$, and $H$ with regards to $\overline{BC}$, $\overline{CD}$ and $\overline{DA}$. $T = \overline{PQ} \cap \overline{RS}$. $K$ as the exsimilicenter of $\omega_1$ and $\omega_3$. $M = \overline{FS} \cap \overline{GP}$ and similarly $N = \overline{ER} \cap \overline{HQ}$. Claim: The points $K$, $A$ and $C$ are collinear. Proof. Note from Monge on $\omega_1$, $\omega_3$ and $\omega$ our claim follows. $\square$ Claim: The points $F$, $E$ and $K$ are collinear. Similarly the points $G$, $H$, and $K$ are collinear. Proof. From Monge on $\omega_1$, $\omega_2$ and $\omega_3$ our claim follows. $\square$ Now to finish it off. Claim: The points $K$, $M$, $N$, $A$, $C$ and $T$ are all collinear. Proof. From Pappus Theorem on $\overline{FPH}$ and $\overline{GSE}$ we find $\overline{MKA}$ collinear. Similarly by Pappus Theorem on $\overline{EQG}$ and $\overline{HRF}$ we find $\overline{NKC}$ collinear. Also recall the points $\overline{KAC}$ are collinear. Finally by Pappus Theorem on $\overline{PRF}$ and $\overline{SQG}$ we find $\overline{TMC}$ collinear. Thus as of right now we have, $\overline{MKA}$ $\overline{NKC}$ $\overline{KAC}$ $\overline{TMC}$ Combining the first and third we have $\overline{MKAC}$ collinear. Combining this with the second we have $\overline{MKANC}$ collinear. Finally combining this with the fourth we have $\overline{TMKANC}$ all collinear as desired. Specifically, $T$, $A$ and $C$ are collinear, so we are done. $\square$
22.07.2024 16:29
We will show $AC$, $PQ$, and $RS$ are concurrent in the projective plane. Claim: $PR=QS$ Proof: Let $O_A$ be the incenter of $APOS$. Define $O_B$, $O_C$, and $O_D$ similarly. Let $O_C'$ and $O_D'$ be the reflection of $O_C$ about $O_BO_D$ and the reflection of $O_D$ about $O_AO_C$, respectively. Let $Q'$, $R'$, and $S'$ be the reflection of $Q$ about $O_BO_D$, the reflection of $R$ about $O$, and the reflection of $S$ about $O_AO_C$. Then, $P$, $Q'$, $R'$, and $S'$ are the intersections of $OP$ with $(OO_AO_B)$, $(OO_BO_C')$, $(OO_C'O_D')$, and $(OO_D'O_A)$, respectively. Let $*$ denote the image of a point under inversion about $O$. Then, $P*$, $Q'*$, $R'*$, and $S'*$ are the intersections of $OP$ with $O_A*O_B*$, $O_B*O_C'*$, $O_C'*O_D'*$, and $O_D'*O_A*$. By DIT, an involution on line $OP$ swaps $P*$ with $R'*$, $Q'*$ with $S'*$, and $O$ with itself. This is an inversion. Assume it has radius $1$. Then, for some $r$, $OP*=1+r$ and $OR'*=1+\frac1r$. Thus, $PR=\frac1{OP*}+\frac1{OR'*}=1$. Similarly, $QS=1$. Then, using Pitot's theorem and adding, we get $AB+CD-AD-BC = 2PR-2QS=0$, so $ABCD$ is tangential. Let $X$ be the exsimilicenter of $\omega_A$ and $\omega_C$. By Monge's theorem on $\omega_A$, $\omega_C$, and the incircle of $ABCD$, $X$ must lie on $AC$. Let $\ell_P$ be the line tangent to $\omega_C$ that passes through $P$ and doesn't pass through $O$. Define $\ell_S$ similarly. Let $Y$ be the intersection of $\ell_P$ and $\ell_S$. By Pitot's theorem, $YP-YS=OP-OS=AP-AS$. Thus, $APYS$ has an incircle. By Monge's theorem on the incircle of $APYS$, $Y$ lies on $AX$. Thus, $Y$ lies on $AC$. Note that $PYSQCR$ has incircle $\omega_C$. Thus, by Brianchon's theorem, $PQ$, $RS$, and $CY$ are concurrent. Line $CY$ is line $AC$, so $PQ$, $RS$, and $AC$ are concurrent.
18.08.2024 15:10
2015 G7 We start off by showing the following claim. Claim: $ABCD$ is tangential quadrilateral. Proof: Apply pitot theorem multiple times to show that $AB+CD= AD+CB$. $\blacksquare$ Now we introduce the following points. let $H$ be the exsimilicenter of $\omega_A$ and $\omega_C$ and $G$ be the exsimilicenter of $\omega_B$ and $\omega_D$. $Z= \overline{DA}\cap\overline{PR}$, $Y = \overline{AB}\cap \overline{QS}$, $X = \overline{BC}\cap \overline{PR}$, and $W= \overline{CD}\cap\overline{QS}$. Now by Monge’s theorem we have that; $HAC, GBD, YHX, ZWH, YZG, XGW$ are collinear. Hence we can finish the problem by applying Pappus theorem on; $SQW, PRX$ and $XZP, YWS$.
28.09.2024 14:21
Solved with ohiorizzler1434. This is going to be pretty similar to other solutions, and indeed this problem is an absolute projective menace . Also wow why does this feel like ISL 2022/G8. Work over the projective plane for this question. We first start off by showing $ABCD$ has an incircle $\omega$. However this is just ELMO 2011/1. So let $\omega_1$, $\omega_2$, $\omega_3$, $\omega_4$ be incircles of $APOS$, $BQOP$, $CROQ$ and $DSOR$ resp. We now constuct $K_1=\overline{PR}\cap\overline{AD}$ (the exsimilicentre of $ \omega_1$ and $\omega_4$), $K_2=\overline{PR}\cap\overline{BC}$ (the exsimilicentre of $ \omega_2$ and $\omega_3$), $L_1=\overline{CD}\cap\overline{QS}$ (the exsimilicentre of $ \omega_3$ and $\omega_4$), $L_2=\overline{AB}\cap\overline{QS}$ (the exsimilicentre of $ \omega_1$ and $\omega_2$), $M_1$ the exsimilicentre of $\omega_1$ and $\omega_3$, $M_2$ the exsimilicentre of $\omega_2$ and $\omega_4$. By repeated applications of Monge we have that the following are collinear: $K_1$, $L_2$, $M_2$ ($\omega_1$, $\omega_2$,$\omega_4$) $K_2$, $L_1$, $M_2$ ($\omega_3$, $\omega_2$,$\omega_4$) $K_1$, $L_1$, $M_1$ ($\omega_1$, $\omega_4$,$\omega_3$) $K_2$, $L_2$, $M_1$ ($\omega_1$, $\omega_2$,$\omega_3$) $A$, $C$, $M_1$ ($\omega$, $\omega_1$,$\omega_3$) $B$, $D$, $M_2$ ($\omega$, $\omega_2$,$\omega_4$) So now delete the incircles. We are now left with the following problem: Incircle-less ISL 2015/G7 wrote: We are given a quadrilateral $ABCD$ with points $P$, $Q$, $R$, $S$ on $\overline{AB}$, $\overline{BC}$, $\overline{CD}$, $\overline{DA}$ such that the following are collinear: $K_1$, $L_2$, $M_2$ $K_2$, $L_1$, $M_2$ $K_1$, $L_1$, $M_1$ $K_2$, $L_2$, $M_1$ $A$, $C$, $M_1$ $B$, $D$, $M_2$ where $K_1=\overline{PR}\cap\overline{AD}$, $K_2=\overline{PR}\cap\overline{BC}$, $L_1=\overline{CD}\cap\overline{QS}$, $L_2=\overline{AB}\cap\overline{QS}$, $M_1=\overline{AC}\cap\overline{K_1L_1}$ and $M_2=\overline{BD}\cap\overline{K_1L_2}$. Show that $\overline{AC}$, $\overline{PQ}$, $\overline{RS}$ concur. By Pappus on $\overline{K_1PK_2}$ and $\overline{L_2SL_1}$, $A=\overline{L_2P}\cap\overline{K_1S}$, $M_1=\overline{K_1L_1}\cap\overline{K_2L_2}$, and $U=\overline{L_1P}\cap\overline{K_2S}$ are collinear. Similarly, using Pappus on $\overline{K_1RK_2}$ and $\overline{L_2SL_1}$, $C=\overline{L_1R}\cap\overline{K_2Q}$, $M_1=\overline{K_1L_1}\cap\overline{K_2L_2}$, and $V=\overline{L_2R}\cap\overline{K_1Q}$ are collinear. Thus both $U$ and $V$ lie on $\overline{AC}$, and so $\overline{UV}=\overline{AC}$. Now we finish off by using Pappus on $\overline{K_1PQ}$ and $\overline{L_2SR}$; then $A=\overline{L_2P}\cap\overline{K_1S}$, $V=\overline{L_2R}\cap\overline{K_1Q}$ and $X=\overline{PQ}\cap\overline{RS}$ are collinear. Since $\overline{AV}$ is just $\overline{AC}$ however, we have $\overline{PQ}\cap\overline{RS}\cap\overline{AC}=X$ which completes the proof.