EGMO 2018 Problem 1: solution, and the circle behind a translation
22 November 2026
EGMO 2018 Problem 1 hands you a moving point, a ratio and a perpendicular, and asks for a circle nobody names. The work is mostly finding that circle. Once you notice what the angle does to the circumcentre, turns out to be slid along a fixed vector, and a slid circle is still a circle. Here is the statement, a full solution, a coordinate check, and what in the wording points the way.
The signal in the statement
Three details carry the whole problem.
- A two-to-one ratio next to a midpoint. That is the shape of a median and its centroid: the centroid sits a third of the way along a median, measured from the midpoint. So look for a triangle whose median runs through and whose centroid is .
- An isosceles triangle with a wide apex. The apex is , so the base angles are , and is the angle that puts the circumcentre at the mirror image of the apex. Write for the circumcentre. Then is the midpoint of , and is perpendicular to .
- A perpendicular through the moving point. By the last point, that line is parallel to . A fixed direction from a moving point on a circle suggests a translation, and a translation carries the circumcircle onto another circle of the same size.
Put those together and the guess is: , so lies on the circle of radius centred at .
Solution
Write for the circumcentre of and for its circumradius. Treat points as vectors.
- Since and , the base angles are . The line is the axis of symmetry, so and .
- The circumcentre lies on that axis. The central angle on chord is twice the inscribed angle at a point of the major arc, and makes sit on the minor arc, so with on the other side of from . Then triangle is isosceles with base angles , so . Hence is the midpoint of , and .
- The chord of the circumcircle subtends , so . Likewise , and because is the centre. So the circle centred at with radius passes through , and . Call it .
- with on segment means .
- Define . Step 2 gives , so
So is the centroid of triangle . The median from goes through the midpoint of , and the centroid lies on it. Hence is on line .
- , which is perpendicular to by step 2. So is also on the line through perpendicular to .
- The problem says those two lines meet at a unique point . is on both, so .
- . So lies on , the circle centred at through and , for every allowed .
In words: is a parallelogram, because and are parallel and equal. Its opposite sides give .
An independent check
Coordinates reach the same circle without the centroid. Put at the origin with on the -axis and . Then , , and . For on the circumcircle, . The line is the set of multiples of , and the vertical line through meets it at . That is moved up by , the length of , and its distance from is the distance from to , which is .
This page also runs the construction straight from the definitions, on two thousand triangles placed at random positions, sizes and angles in the plane: build from the angle, compute as the circumcentre, from the ratio, and as a line intersection. It confirms every time, confirms is the centroid of , and confirms that only and the point of the circle opposite make the two lines coincide. It also changes one given at a time. With instead, or with , the distance wanders, so both givens are doing work.
Where marks go missing
- Finding the circle from three sketches and stopping. Drawing three positions of is how you guess the answer, and it is fine as a guess. The proof is the identity in step 5, and it has to cover every .
- Asserting the midpoint without the angle. That is the midpoint of is true only because the apex angle is . Show the base angles in triangles and , or the computation , so a grader can see where the angle was used.
- Checking one line only. is defined as a crossing point, so must be shown on both lines. Step 7 then uses the uniqueness the statement gives you. Without it, is a real problem: there is the perpendicular itself.
- Naming the wrong circle. The circumcircle of is not the answer, and neither is a circle centred at . Say which circle: centre , radius , through , and .
Practise the idea
The proof is a translation: the circumcircle slides along onto , one of the transformations that turn a locus problem into one line of algebra. If the centroid does not jump out, the coordinates in the check are the fallback from complex numbers and coordinates, and they take five lines here. The fact that a centroid splits each median is olympiad problem o3-gm-5, and the vector proof there is exactly the one used in step 5. The daily problem posts a new one every day, no account needed.
Related guides
- When to reach for a rotation or a reflection in a geometry proofTransformations look like magic until you know what triggers them. Five signals in a figure that tell you which rotation, reflection or homothety to use.
- Complex numbers for olympiad geometry: when to reach for themComplex coordinates turn a geometry problem into algebra you can finish. The configurations where that trade pays, the six-line dictionary, and where it stalls.
- Angle chasing: when to stop and what to reach for insteadAngle chasing rarely fails outright. It finishes, and you keep going. Four signals that the chase is over, and what the conclusion says to reach for next.
- EGMO 2013 Problem 1: solution, and how to spot the hidden centroidA full solution to EGMO 2013 Problem 1, the short centroid proof and a vector check, plus the signal in the statement that tells you where to look.
Train it on Lemma
Every technique here is taught explicitly, in order. 128 lessons, 1237 curated problems and unlimited generated practice at six difficulties. Free to start, no card, and the paid plan opens with 3 free days.
Find your level