Unofficial archive — problems, solutions & results © IMC, reproduced with permission.

IMC / 2019 / Problems / Day 2, P10

IMC 2019 · Day 2 · P10

killer

2019 points are chosen at random, independently, and distributed uniformly in the unit disc {(x,y)R2:x2+y21}\{ (x, y) \in \mathbb{R}^2 : x^2 + y^2 \le 1 \}. Let CC be the convex hull of the chosen points. Which probability is larger: that CC is a polygon with three vertices, or a polygon with four vertices?

Proposed by Fedor Petrov, St. Petersburg State University

Solution (official)

We will show that the quadrilateral has larger probability.

Let D={(x,y)R2:x2+y21}D = \{ (x, y) \in \mathbb{R}^2 : x^2 + y^2 \le 1 \}. Denote the random points by X1,,X2019X_1, \dots, X_{2019} and let p=P(C is a triangle with vertices X1,X2,X3),q=P(C is a convex quadrilateral with vertices X1,X2,X3,X4).\begin{align*} p &= \mathsf{P} \bigl( C \text{ is a triangle with vertices } X_1, X_2, X_3 \bigr), \\ q &= \mathsf{P} \bigl( C \text{ is a convex quadrilateral with vertices } X_1, X_2, X_3, X_4 \bigr). \end{align*} By symmetry we have P(C is a triangle)=(20193)p\mathsf{P} \bigl( C \text{ is a triangle} \bigr) = \binom{2019}{3} p, P(C is a quadrilateral)=(20194)q\mathsf{P} \bigl( C \text{ is a quadrilateral} \bigr) = \binom{2019}{4} q and we need to prove that (20194)q>(20193)p\binom{2019}{4} q > \binom{2019}{3} p, or equivalently p<20164q=504qp < \frac{2016}{4} q = 504 q.

Note that pp is the average over X1,X2,X3X_1, X_2, X_3 of the following expression: u(X1,X2,X3)=P(X4X1X2X3)P(X5,X6,,X2019X1X2X3),u(X_1, X_2, X_3) = \mathsf{P} \bigl( X_4 \in \triangle X_1 X_2 X_3 \bigr) \cdot \mathsf{P} \bigl( X_5, X_6, \dots, X_{2019} \in \triangle X_1 X_2 X_3 \bigr), and qq is not less than the average over X1,X2,X3X_1, X_2, X_3 of v(X1,X2,X3)=P(X1,X2,X3,X4 form a convex quad.)P(X5,X6,,X2019X1X2X3).v(X_1, X_2, X_3) = \mathsf{P} \bigl( X_1, X_2, X_3, X_4 \text{ form a convex quad.} \bigr) \cdot \mathsf{P} \bigl( X_5, X_6, \dots, X_{2019} \in \triangle X_1 X_2 X_3 \bigr). Thus it suffices to prove that u(X1,X2,X3)500v(X1,X2,X3)u(X_1, X_2, X_3) \le 500\,v(X_1, X_2, X_3) for all X1,X2,X3X_1, X_2, X_3. It reads as area(X1X2X3)500area(Ω)\operatorname{area}(\triangle X_1 X_2 X_3) \le 500 \operatorname{area}(\Omega), where Ω={Y:X1,X2,X3,Y form a convex quadrilateral}\Omega = \{ Y : X_1, X_2, X_3, Y \text{ form a convex quadrilateral} \}. Assume the contrary, i.e., area(X1X2X3)>500area(Ω)\operatorname{area}(\triangle X_1 X_2 X_3) > 500 \operatorname{area}(\Omega).

Let the lines X1X2X_1 X_2, X1X3X_1 X_3, X2X3X_2 X_3 meet the boundary of DD at A1,A2,A3,B1,B2,B3A_1, A_2, A_3, B_1, B_2, B_3; these lines divide DD into 7 regions as shown in the picture; Ω=D4D5D6\Omega = D_4 \cup D_5 \cup D_6.

By our indirect assumption, area(D4)+area(D5)+area(D6)=area(Ω)<1500area(D0)<1500area(D)=π500.\operatorname{area}(D_4) + \operatorname{area}(D_5) + \operatorname{area}(D_6) = \operatorname{area}(\Omega) < \frac{1}{500} \operatorname{area}(D_0) < \frac{1}{500} \operatorname{area}(D) = \frac{\pi}{500}. From X1X3B3Ω\triangle X_1 X_3 B_3 \subset \Omega we get X3B3/X3X2=area(X1X3B3)/area(X1X2X3)<1/500X_3 B_3 / X_3 X_2 = \operatorname{area}(\triangle X_1 X_3 B_3) / \operatorname{area}(\triangle X_1 X_2 X_3) < 1/500, so X3B3<1500X2X3<1250X_3 B_3 < \frac{1}{500} X_2 X_3 < \frac{1}{250}. Similarly, the lengths segments A1X1,B1X1,A2X2,B2X2,A3X2A_1 X_1, B_1 X_1, A_2 X_2, B_2 X_2, A_3 X_2 are less than 1250\frac{1}{250}.

The regions D1,D2,D3D_1, D_2, D_3 can be covered by disks with radius 1250\frac{1}{250}, so area(D1)+area(D2)+area(D3)<3π2502.\operatorname{area}(D_1) + \operatorname{area}(D_2) + \operatorname{area}(D_3) < 3 \cdot \frac{\pi}{250^2}. Finally, it is well-known that the area of any triangle inside the unit disk is at most 334\frac{3\sqrt3}{4}, so area(D0)334\operatorname{area}(D_0) \le \frac{3\sqrt3}{4}.

But then i=06area(Di)<334+3π2502+π500<area(D),\sum_{i=0}^{6} \operatorname{area}(D_i) < \frac{3\sqrt3}{4} + 3 \cdot \frac{\pi}{250^2} + \frac{\pi}{500} < \operatorname{area}(D), contradiction.

How the field did

contestants scored
360
average (of 10)
0.62
solved (≥ 80%)
2.2%
near-0 (≤ 10%)
88.6%
discrimination
0.36

Score distribution (field cohort)

Computed on contestants with a meaningful total (field cohort); discrimination is the corrected item–total correlation.

Similar problems

IMC 2024 · Day 1 · P5killeravg 0.1/10 · solved 1% · near-0 97% · disc 0.25
IMC 2022 · Day 2 · P8very hardavg 1.3/10 · solved 8% · near-0 80% · disc 0.46
IMC 2017 · Day 2 · P10killeravg 0.2/10 · solved 2% · near-0 97% · disc 0.42
IMC 2018 · Day 1 · P5killeravg 0.5/10 · solved 3% · near-0 94% · disc 0.39