We use tail bounds to bound the probability that a random variable X deviates from its expectation \mu = \mathbb{E}[X].
Throw n balls into n bins independently and uniformly at random (i.i.d). Let L_i be the load of bin i.
For the ith bin , L_i = \sum_{j=1}^n X_j is a sum of independent r.v.s.
Let’s choose \lambda = \frac{3 \ln n}{\ln \ln n}.
We want to bound the probability that any bin exceeds our threshold \lambda.
Throw m balls into n bins (e.g., n = 365 days in a year).
What is the probability of having no collisions (all balls land in distinct bins)?
We want to find m such that the collision probability is at least 1/2 (\Pr(\text{no collision}) \le 1/2).
Recall the classic Coupon Collector problem: we want to collect n unique coupons.
After throwing n balls into n bins, bound the probability that some bin has load > 6 \ln n / \ln\ln n.
Hint: per-bin Chernoff gives \Pr(L_i \ge \lambda) \le n^{-2} for \lambda = 6\ln n/\ln\ln n; union bound over n bins gives \le 1/n.
Coupon Collector & Random Sampling — concentration details, sample complexity, and comparing bounds.
CS G526: Advanced Algorithms & ComplexityTulasimohan Molli