Let $ A, B$ and $ C$ be non-collinear points. Prove that there is a unique point $ X$ in the plane of $ ABC$ such that \[ XA^2 + XB^2 + AB^2 = XB^2 + XC^2 + BC^2 = XC^2 + XA^2 + CA^2.\]
Problem
Source: IMO Shortlist 1995, G
Tags: geometry, circumcircle, reflection, complex numbers, perpendicular bisector, IMO Shortlist
27.03.2009 18:14
thaithuan_GC wrote: Let $ A, B$ and $ C$ be non-collinear points. Prove that there is a unique point $ X$ in the plane of $ ABC$ such that \[ XA^2 + XB^2 + AB^2 = XB^2 + XC^2 + BC^2 = XC^2 + XA^2 + CA^2. \] It's easily killed with complex numbers. Consider the complex plane with center at the circumcenter of $ \triangle ABC$, and let $ a,b,c,x$ be the corresponding complex numbers to $ A,B,C,X$ respectively. (Then $ |a| = |b| = |c|$) Besides let $ X'$ be a $ 180 ^\circ$ rotation of $ X$, so $ x' = - x$. Let $ \cdot$ be the real product, then we have: $ XA^2 + XB^2 + AB^2 = XB^2 + XC^2 + BC^2 \iff$ $ XA^2 + AB^2 - XC^2 - BC^2 = 0 \iff$ $ (x - a) \cdot (x - a) + (a - b) \cdot (a - b) - (x - c) \cdot (x - c) - (b - c) \cdot (b - c) = 0\iff$ (Using that $ |a| = |b| = |c|$) $ (a - c) \cdot (b - ( - x)) = (a - c) \cdot (b - x') = 0 \iff$ $ AC \perp BX'$ In the same way we get, $ AB \perp CX'$, $ BC \perp AX'$, so there is equality iff $ X'$ is the orthocenter. So the equation is fulfilled if and only if $ X$ is the orthocenter rotated $ 180 ^\circ$ around the circumcenter
27.03.2009 19:33
The first condition $ XA^2 + XB^2 + AB^2 = XB^2 + XC^2 + BC^2$ is equivalent to $ XA^2 - XC^2 = BC^2 - AB^2.$ Which clearly means that $ X$ belows to the reflection of the B-altitude with respect to the perpendicular bisector of $AC.$ By similar reasoning, we conclude that $ X$ is the reflection of the orthocenter $H$ with respect to the circuncenter $O.$ In other words, $ X$ is the De Longchamps point $X_{20}$ of $\triangle ABC.$
12.07.2009 04:17
Draw the following circles: Center $ C$, radius segment AB Center $ B$, radius segment AC Center $ A$, radius segment BC. From the equations, it is clear that the unique point is the radical center of these circles.
12.07.2009 06:34
Dear kyyuanmathcount, this is the way that de Longchamps quoted by Luis has taken... Sincerely Jean-Louis
26.03.2016 05:05
I think maybe a problem which can be so easily done with coordinates should not be on ISL. WLOG $B(-1,0),C(1,0),A(a,b),X(x,y)$. Since $A,B,C$ not collinear we have $b \neq 0$. We have $BC^2=4$ $AB^2=(a+1)^2+b^2$ $AC^2=(a-1)^2+b^2$ $XB^2=(x+1)^2+y^2$ $XC^2=(x-1)^2+y^2$ $XA^2=(x-a)^2+(y-b)^2$ The given condition is there exists a real number $K$ with: I. $K=(x-a)^2+(y-b)^2+(x+1)^2+y^2+(a+1)^2+b^2$ II. $K=(x+1)^2+y^2+(x-1)^2+y^2+4$ III. $K=(x-1)^2+y^2+(x-a)^2+(y-b)^2+(a-1)^2+b^2$ Define $N=K-2(x^2+y^2)$. From (II), we have $N=6$. Then (I) and (III) are of the form: I. $\alpha_1x+\beta_1y+\gamma_1=6$ III. $\alpha_3x+\beta_3y+\gamma_3=6$ Note that $\beta_3=-2b \neq 0$. Multiply (I) by $\beta_3$ and (III) by $\beta_1$, and subtract: $x(\alpha_1\beta_3-\alpha_3\beta_1)+(\gamma_1\beta_3-\gamma_3\beta_1)=6(\beta_3-\beta_1)$ Therefore there exists a unique solution to this system iff $\alpha_1\beta_3 \neq \alpha_3\beta_1$. But $\alpha_1=2-2a$,$\beta_1=-2b$,$\alpha_3=-2a-2$, $\beta_3=-2b$, so this is obviously the case.
11.09.2023 18:29
cant we just assume X is origin, and all that's left is to prove that there exist points A,B,C that satisfy the conditions(also that form a triangle) because can't you just translate point x to the origin for any valid formation, also transforming ABC