IN A TRIANGLE
NAVNEEL SINGHAL
1. Notation
We use the notation (<) to be the circumcircle of figure <, (Q) to be the circle with
Q as center where the circle being discussed with Q as center is unique, and (M N ) to
be the circle with diameter M N . By G∗ we denote the inverse of G in the incircle.
2. Introduction
Firstly we describe the configuration.
ABC is a triangle with incenter I, circumcenter O, orthocenter H. The incircle touches
BC, CA, AB at D, E, F resp. The A−mixtilinear incircle touches the circumcircle of
(ABC), Γ at T , and the A−mixtilinear touchchord meets BC at Y . The midpoint of
arc BC of Γ not containing A is MA and its antipode in Γ is MA0 . EF intersects BC at
X1 , the circle Ω with AI as a diameter cuts Γ at KA and A. X is the foot of D on EF .
AI intersects BC at L1 . The midpoint of DX1 is L. K is on BC such that XK k AKA .
A0 is the antipode of A in Γ, EF cuts AKA at Z, M1 is the midpoint of IY , KA 0
is the
second intersection of (KA XD) with Γ. (IDY ) cuts (I) again at U . By ω we denote the
incircle, by ρ we denote (BIC), by α we denote (KA ZX1 ) and by β we denote (KA Y X1 )
3. The Properties
We begin by a few well known properties, whose proofs we will omit.
Lemma 3.1. I is the midpoint of the A-mixtilinear touch chord.
Lemma 3.2. Y , T , MA are collinear and MA0 , I, T are collinear with these lines being
perpendicular.
Lemma 3.3. D, L1 , MA , T are concyclic.
Lemma 3.4. A, KA , Y are collinear.
Proof. Note that Ω and ρ are tangent to each other because they meet at I and their
centers are the midpoint of AI and MA respectively, which are collinear with I. By
the radical axes theorem on Ω, ρ, Γ, the A−mixtilinear touch chord, BC and AKA are
collinear, implying the conclusion.
Lemma 3.5. KA , X, I, A0 are collinear and KA , D, MA are collinear.
1
2 NAVNEEL SINGHAL
Proof. Consider the inverse of KA under the inversion I about the incircle. As Γ is sent
to the ninepoint circle of DEF , and Ω is sent to EF , KA is sent to X, giving KA , X, I
collinear. Also since ∠AKA I = 90◦ , we have that KA I meets Γ at A0 .
For the second part, (B, C; D, X1 ) = −1 and DX ⊥ EF ⇒ DX bisects ∠BXC ⇒
FX
4BF X ∼ 4XEC ⇒ XE = F B BD
EC = DC . Since KA is the center of spiral similarity S
sending F E to BC, F E ∪ {X} is sent to BC ∪ {D} under S. Also, I is sent to MA under
S and so by the first part of this lemma, KA , D, MA are collinear.
Remark. The first part of the lemma appeared as a problem in NIMO 2014.
A Γ
KA Fe
J E
I′
U
X O
F I ρ
Z
M1
H
Y D
X1 L C
B L1
T
MA
Lemma 3.20. ZL ⊥ IO
Proof. ZKA · ZA = ZF · ZE ⇒ Z ∈ radical axis of ω and Γ ⇒ ZL is the radical axis of
ω and Γ, and so ZL ⊥ IO.
Lemma 3.21. Midpoint of the A−altitude of 4ABC, D, and the A−excenter IA are
collinear.
Proof. By the homothety at A sending the incircle to the A−excircle, D, A and the
endpoint of the diameter of (IA ) perpendicular to BC and not on BC are collinear. By
a stretch at BC with ratio 12 , we get our result.
Remark. The lemma is true even when D is replaced by the tangency point of the
A−excircle and BC and when IA is replaced by I.
Lemma 3.22. The radical axis of ω and ρ is the D-midline of DEF .
Proof. Clearly, the radical axis in question must be perpendicular to AI and thus be
parallel to EF . Since L is on the radical axis, and L is the midpoint of X1 D, we get our
result.
Lemma 3.23. IL and U D meet on ρ.
Proof. L is on the radical axis of ω and ρ so L∗ is on ρ. Also, L∗ ∈ IL. Also, as (L) is
orthogonal to ω, LU, LD are tangents to ω ⇒ L∗ ∈ U D and thus our result is proved.
Remark. As IL ⊥ U D, ∠IL∗ D = 90◦ . So the line U D meets ρ on the antipode of I in
ρ, which is IA . So U D bisects the A−altitude.
Lemma 3.24. IX1 , (L), Ω concur at a point.
Proof. As X1 is on EF , X1∗ is on Ω. As (L) and ω are orthogonal and X1 is on (L), X1∗ is
on (L). By the definition of inverse, X1∗ ∈ IX1 . Thus IX1 , (L), Ω concur at a point.
Lemma 3.25. U D, EF , KA X1∗ concur at a point.
Proof. The result follows from radical axes theorem applied to the triad of circles (L), ω, Ω.
Lemma 3.26. XD, ω, (M1 ) concur.
Proof. Let XD intersect ω again at J. Then, ∠JID = ∠JID = ∠XDI = ∠XKI =
∠Y KA X1 = θ = ∠DY I and thus J is on (M1 ).
Lemma 3.27. If Fe is the Feuerbach point of ABC, then Fe J is the reflection of the
line OI in EF .
Proof. It is well known that OI is the Euler line of DEF , and that Fe is the Euler
reflection point of DEF . Since J is the reflection of the orthocenter of DEF over EF ,
we get that Fe J is the reflection of OI over EF .
Observe that by the lemma just proved, the reflection of I over EF lies on JFe . the
inverse of JFe is the circumcircle of IJFe , which contains the inverse of the reflection of I
over EF , i.e., the center of Ω. Also since ∠MA0 KA A = θ, X1 , KA , MA0 are also collinear.
A consequence is that by Pascal’s theorem on T MA MA0 KA AA0 , OY, X1 MA0 , A0 T concur
at a point.
ON A RICH CONFIGURATION RELATED TO TANGENT CIRCLES IN A TRIANGLE 5
4. An application
Consider the following problem (points renamed) from the IMO Shortlist of 2002
(numbered G7):
Problem: ABC is an acute-angled triangle. The incircle touches BC at D. The altitude
AQ has midpoint M . The line KM meets the incircle again at U 0 . Show that the
circumcircle of BCU 0 touches the incircle of ABC at U 0 .
Solution:
By the remark accompanying lemma 3.23, we have U 0 ≡ U Apply radical axes theorem
on the circles (BU C), ω and Γ. The radical center is L by our previous discussion. But
LU 0 = LD. So in fact LU 02 = LD2 = LB · LC and thus LU 0 is tangent to ω. But as
LU 0 is the radical axis of ω and (BU 0 C), we get that LU 0 is tangent to (BU 0 C) as well
and so (BU 0 C) and ω are tangent to each other at U 0 as required.
References
[1] J. Vonk, The Feuerbach point and reflections of the Euler line, Forum Geom,
9(2009) 47-55
[2] A.V. Akopyan, Geometry in Figures, 2011