Maths Olympiad Prep

Library / /697 of 1394

, 2023

Number theory Difficulty 5.3 AIME, harder Find the answer United States

Problem:

The number 20272027 is prime. For i=1,2,,2026i = 1, 2, \ldots, 2026, let pip_{i} be the smallest prime number such that pii(mod2027)p_{i} \equiv i \pmod{2027}. Estimate max(p1,,p2026)\max \left(p_{1}, \ldots, p_{2026}\right).

Submit a positive integer EE. If the correct answer is AA, you will receive 25min((E/A)8,(A/E)8)\left\lfloor 25 \min \left((E / A)^{8},(A / E)^{8}\right)\right\rfloor points. (If you do not submit a positive integer, you will receive zero points for this question.)

A number or a short expression. Fractions can be typed as 3/2, and spacing doesn't matter.

Solution

Solution:

In this solution, all logs are in base ee. Let p1,p2,p_{1}, p_{2}, \ldots be the primes in sorted order. Let qi=pimod2027q_{i} = p_{i} \bmod 2027. Since the residues of primes modulo 20272027 should be uniformly distributed, we can make the probabilistic approximation that the qiq_{i} are random variables uniformly distributed among 1,,20261, \ldots, 2026. This becomes the famous "coupon collector" problem: the random variables qiq_{i} are coupons with 20262026 different types, and we keep collecting coupons until we have encountered one of each type. In other words, we seek to find the smallest kk such that {q1,,qk}={1,,2026}\{q_{1}, \ldots, q_{k}\} = \{1, \ldots, 2026\}, and then the answer to the problem is pkp_{k}.

It is known that the expected value of kk is 2026(11+12++12026)2026log20262026\left(\frac{1}{1} + \frac{1}{2} + \cdots + \frac{1}{2026}\right) \approx 2026 \log 2026. This is because we must draw an expected 20262026\frac{2026}{2026} coupons until we get our first distinct coupon type, then an expected 20262025\frac{2026}{2025} coupons until we get our second new coupon type, and so on. The standard deviation of kk is a small fraction of its expectation, so we can safely assume that kk is approximately 2026log20262026 \log 2026. Since the nn-th prime is approximately nlognn \log n, our estimate is
E2026log2026log(2026log2026)2026log22026117448 \begin{aligned} E & \approx 2026 \log 2026 \log (2026 \log 2026) \\ & \approx 2026 \log^{2} 2026 \\ & \approx 117448 \end{aligned}
This achieves A/E0.969A / E \approx 0.969, which scores 19 out of 25 points.

Want a route through all this instead of an archive? The track puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement reproduced verbatim; metadata (topic, difficulty) added by this project.