Show that for n events A₁, A₂, ⋯, Aₙ
P(⋂ᵢ₌₁ⁿ Aᵢ) ≥ ∑ᵢ₌₁ⁿ P(Aᵢ) − (n − 1)
7 marks · directive: Show
The core of a good answer
Pass to complements: P(⋂Aᵢ) = 1 − P(⋃Aᵢᶜ) ≥ 1 − ∑P(Aᵢᶜ) = 1 − ∑[1 − P(Aᵢ)] = ∑P(Aᵢ) − (n − 1), using Boole's inequality for the union of the complements. This is Bonferroni's lower bound for an intersection.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
A short proof is asked for. The cleanest route is De Morgan's law plus Boole's inequality (the union bound), which itself is proved by induction from P(A ∪ B) = P(A) + P(B) − P(A ∩ B) ≤ P(A) + P(B). The examiner wants each step justified, not just the final line.
Answer outline
Introduction
Let Aᵢᶜ denote the complement of Aᵢ. The idea is that the intersection fails only if at least one Aᵢ fails, and the probability of 'at least one failure' is at most the sum of the individual failure probabilities.
Step 1: Boole's inequality for unions
- For two events: P(B₁ ∪ B₂) = P(B₁) + P(B₂) − P(B₁ ∩ B₂) ≤ P(B₁) + P(B₂).
- Induction: if P(⋃ᵢ₌₁ᵏ Bᵢ) ≤ ∑ᵢ₌₁ᵏ P(Bᵢ), then P(⋃ᵢ₌₁ᵏ⁺¹ Bᵢ) ≤ P(⋃ᵢ₌₁ᵏ Bᵢ) + P(Bₖ₊₁) ≤ ∑ᵢ₌₁ᵏ⁺¹ P(Bᵢ).
- Hence P(⋃ᵢ₌₁ⁿ Bᵢ) ≤ ∑ᵢ₌₁ⁿ P(Bᵢ) for every n.
Step 2: De Morgan's law
- (⋂ᵢ₌₁ⁿ Aᵢ)ᶜ = ⋃ᵢ₌₁ⁿ Aᵢᶜ, so P(⋂Aᵢ) = 1 − P(⋃Aᵢᶜ).
Step 3: Apply Boole to the complements
- P(⋃Aᵢᶜ) ≤ ∑ P(Aᵢᶜ) = ∑ [1 − P(Aᵢ)] = n − ∑ P(Aᵢ).
- So P(⋂Aᵢ) ≥ 1 − n + ∑ P(Aᵢ) = ∑ᵢ₌₁ⁿ P(Aᵢ) − (n − 1).
Step 4: Alternative direct induction (optional)
- n = 2: P(A₁ ∩ A₂) = P(A₁) + P(A₂) − P(A₁ ∪ A₂) ≥ P(A₁) + P(A₂) − 1.
- Assume true for n; put B = ⋂ᵢ₌₁ⁿ Aᵢ: P(B ∩ Aₙ₊₁) ≥ P(B) + P(Aₙ₊₁) − 1 ≥ ∑ᵢ₌₁ⁿ⁺¹ P(Aᵢ) − n.
Conclusion
The bound P(⋂Aᵢ) ≥ ∑P(Aᵢ) − (n − 1) holds for any events, independent or not. Check: if every P(Aᵢ) = 1 the bound gives 1, as it must.
Examples and data to cite
- Simultaneous confidence intervals: if each of n intervals has coverage 1 − α/n, the bound gives joint coverage at least 1 − α (the Bonferroni method).
- With n = 3 events each of probability 0.9, the bound gives P(all three) ≥ 2.7 − 2 = 0.7.
Facts to remember
- Boole's inequality: P(⋃Aᵢ) ≤ ∑P(Aᵢ).
- De Morgan: (⋂Aᵢ)ᶜ = ⋃Aᵢᶜ.
- The result P(⋂Aᵢ) ≥ ∑P(Aᵢ) − (n − 1) is one of the Bonferroni inequalities.
- The bound can be negative and then is trivially true; it is useful when the P(Aᵢ) are close to 1.