{
  "id": "borel-cantelli-lemmas",
  "title": "Borel–Cantelli Lemmas",
  "lang": "en",
  "source_lang": "en",
  "summary": "The Borel–Cantelli lemmas relate the sum of event probabilities to whether infinitely many of those events occur, with independence needed for the standard divergence conclusion.",
  "category": "mathematics",
  "subcategory": "probability-statistics",
  "type": "concept",
  "aliases": [
    "Borel–Cantelli lemma",
    "Borel–Cantelli theorems"
  ],
  "generated_at": "2026-11-06",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/borel-cantelli-lemmas/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/borel-cantelli-lemmas/",
      "json": "https://aiwiki.page/zh/borel-cantelli-lemmas.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The Borel–Cantelli lemmas are results in probability theory that determine whether events in an infinite sequence occur infinitely often. The first lemma states that if the sum of their probabilities is finite, then only finitely many occur with probability one. The second states that, for independent events, a divergent sum implies that infinitely many occur with probability one. Named after Émile Borel and Francesco Paolo Cantelli, the lemmas are fundamental tools for proving almost-sure convergence and describing the long-term occurrence of rare events."
    },
    {
      "heading": "Events occurring infinitely often",
      "text": "Let \\((\\Omega,\\mathcal F,\\mathbb P)\\) be a probability space, and let \\(A_1,A_2,\\ldots\\) be events. The event that infinitely many \\(A_n\\) occur is\n\n\\[\\{A_n\\text{ infinitely often}\\} = \\bigcap_{m=1}^{\\infty}\\bigcup_{n=m}^{\\infty}A_n.\\]\n\nIt is commonly abbreviated as \\(\\{A_n\\text{ i.o.}\\}\\) and called the limit superior of the sequence of events:\n\n\\[\\limsup_{n\\to\\infty}A_n.\\]\n\nAn outcome belongs to this event if, however far along the sequence one starts, at least one later event occurs. Its complement is\n\n\\[\\bigcup_{m=1}^{\\infty}\\bigcap_{n=m}^{\\infty}A_n^c,\\]\n\nthe event that only finitely many \\(A_n\\) occur.\n\nThe statements concern the number of occurrences along an entire infinite sequence, not whether any particular event occurs. “Only finitely many almost surely” does not impose a deterministic upper bound on that number: different outcomes may have arbitrarily large, but finite, numbers of occurrences."
    },
    {
      "heading": "First Borel–Cantelli lemma",
      "text": "For any sequence of events,\n\n\\[\\sum_{n=1}^{\\infty}\\mathbb P(A_n)<\\infty \\quad\\Longrightarrow\\quad \\mathbb P\\!\\left(\\limsup_{n\\to\\infty}A_n\\right)=0.\\]\n\nNo independence assumption is required. Thus a summable sequence of event probabilities guarantees that the events eventually stop occurring almost surely.\n\nProof by a tail bound\n\nFor every positive integer \\(m\\), countable subadditivity gives\n\n\\[\\mathbb P\\!\\left(\\bigcup_{n=m}^{\\infty}A_n\\right) \\leq \\sum_{n=m}^{\\infty}\\mathbb P(A_n).\\]\n\nSince \\(\\limsup_n A_n\\) is contained in every such union,\n\n\\[0\\leq \\mathbb P(\\limsup_n A_n) \\leq \\sum_{n=m}^{\\infty}\\mathbb P(A_n).\\]\n\nThe tails of a convergent series tend to zero, proving the result.\n\nThe same argument provides a useful finite-tail estimate:\n\n\\[\\mathbb P(\\text{at least one }A_n\\text{ occurs for }n\\geq m) \\leq \\sum_{n=m}^{\\infty}\\mathbb P(A_n).\\]\n\nInterpretation through the number of occurrences\n\nDefine the extended nonnegative random variable\n\n\\[N=\\sum_{n=1}^{\\infty}\\mathbf 1_{A_n},\\]\n\nwhere \\(\\mathbf 1_{A_n}\\) is the indicator function of \\(A_n\\). By monotone convergence,\n\n\\[\\mathbb E[N]=\\sum_{n=1}^{\\infty}\\mathbb P(A_n).\\]\n\nA nonnegative random variable with finite expectation cannot be infinite on a set of positive probability. Consequently, summability implies \\(N<\\infty\\) almost surely."
    },
    {
      "heading": "Second Borel–Cantelli lemma",
      "text": "If \\(A_1,A_2,\\ldots\\) are mutually independent, then\n\n\\[\\sum_{n=1}^{\\infty}\\mathbb P(A_n)=\\infty \\quad\\Longrightarrow\\quad \\mathbb P\\!\\left(\\limsup_{n\\to\\infty}A_n\\right)=1.\\]\n\nTogether, the two lemmas give the dichotomy\n\n\\[\\mathbb P(\\limsup_n A_n)= \\begin{cases} 0,&\\displaystyle\\sum_n\\mathbb P(A_n)<\\infty,\\\\[4pt] 1,&\\displaystyle\\sum_n\\mathbb P(A_n)=\\infty, \\end{cases}\\]\n\nwhen the events are independent.\n\nProof using independence\n\nFix \\(m\\). For \\(N\\geq m\\), independence gives\n\n\\[\\mathbb P\\!\\left(\\bigcap_{n=m}^{N}A_n^c\\right) = \\prod_{n=m}^{N}\\bigl(1-\\mathbb P(A_n)\\bigr).\\]\n\nUsing \\(1-x\\leq e^{-x}\\) for \\(0\\leq x\\leq1\\),\n\n\\[\\mathbb P\\!\\left(\\bigcap_{n=m}^{N}A_n^c\\right) \\leq \\exp\\!\\left(-\\sum_{n=m}^{N}\\mathbb P(A_n)\\right).\\]\n\nEvery tail of the divergent sum also diverges, so the right-hand side tends to zero. Continuity of probability for decreasing events yields\n\n\\[\\mathbb P\\!\\left(\\bigcap_{n=m}^{\\infty}A_n^c\\right)=0.\\]\n\nTaking the countable union over \\(m\\) shows that the probability of only finitely many occurrences is zero.\n\nPairwise independence is sufficient\n\nAlthough mutual independence is standard in the statement, pairwise independence already suffices for the divergence conclusion.\n\nFor fixed \\(m\\), write\n\n\\[S_{m,N}=\\sum_{n=m}^{N}\\mathbf 1_{A_n}, \\qquad \\mu_{m,N}=\\sum_{n=m}^{N}\\mathbb P(A_n).\\]\n\nPairwise independence makes the covariances between distinct indicators vanish. Therefore,\n\n\\[\\operatorname{Var}(S_{m,N}) = \\sum_{n=m}^{N}\\mathbb P(A_n)\\bigl(1-\\mathbb P(A_n)\\bigr) \\leq \\mu_{m,N}.\\]\n\nBy Chebyshev’s inequality, whenever \\(\\mu_{m,N}>0\\),\n\n\\[\\mathbb P(S_{m,N}=0) \\leq \\frac{\\operatorname{Var}(S_{m,N})}{\\mu_{m,N}^{\\,2}} \\leq \\frac{1}{\\mu_{m,N}}.\\]\n\nIf the probability sum diverges, this bound tends to zero. As in the preceding proof, every tail contains an occurrence almost surely, establishing infinitely many occurrences."
    },
    {
      "heading": "Why the dependence assumption matters",
      "text": "A divergent sum of probabilities alone does not determine the probability of infinitely many occurrences.\n\nRepeating a single event\n\nLet \\(A_n=A\\) for every \\(n\\), with \\(\\mathbb P(A)=p>0\\). Then\n\n\\[\\sum_n\\mathbb P(A_n)=\\infty, \\qquad \\limsup_n A_n=A.\\]\n\nThus\n\n\\[\\mathbb P(A_n\\text{ i.o.})=p,\\]\n\nwhich can be any value in \\((0,1]\\). Repeating the same event produces no new independent opportunities for occurrence.\n\nDecreasing events with divergent probability sum\n\nLet \\(U\\) have the uniform distribution on \\((0,1)\\), and set\n\n\\[A_n=\\{U<1/n\\}.\\]\n\nThen \\(\\mathbb P(A_n)=1/n\\), so the probability sum diverges. Nevertheless, for every \\(U>0\\), the inequality \\(U<1/n\\) eventually fails. Hence\n\n\\[\\mathbb P(A_n\\text{ i.o.})=0.\\]\n\nThis example also shows why infinite expected total count does not imply infinitely many occurrences with positive probability. Here the count is finite almost surely but has infinite expectation."
    },
    {
      "heading": "Examples and applications",
      "text": "Independent trials with decreasing success probabilities\n\nSuppose independent trials have success events \\(A_n\\) with\n\n\\[\\mathbb P(A_n)=n^{-\\alpha}, \\qquad \\alpha>0.\\]\n\nThe convergence criterion for this series gives\n\n\\[\\begin{aligned} \\alpha>1&:\\quad\\text{only finitely many successes almost surely},\\\\ 0<\\alpha\\leq1&:\\quad\\text{infinitely many successes almost surely}. \\end{aligned}\\]\n\nIn the second case, individual success probabilities approach zero, but the cumulative supply of independent opportunities is still sufficient for infinitely many successes.\n\nProving almost-sure convergence\n\nLet \\(X_n\\) and \\(X\\) be real-valued random variables. If, for every \\(\\varepsilon>0\\),\n\n\\[\\sum_{n=1}^{\\infty} \\mathbb P(|X_n-X|>\\varepsilon)<\\infty,\\]\n\nthen \\(X_n\\to X\\) almost surely.\n\nIndeed, the first lemma says that deviations exceeding any fixed tolerance occur only finitely often. It is enough to check the countable family \\(\\varepsilon=1/k\\), \\(k\\geq1\\). Intersecting the resulting probability-one events ensures that every such tolerance is eventually satisfied.\n\nThis is a standard bridge from quantitative probability bounds to pathwise convergence. Independence of the \\(X_n\\) is unnecessary when the required summability bounds hold.\n\nThe converse is false. With \\(U\\) uniform on \\((0,1)\\), let\n\n\\[X_n=\\mathbf 1_{\\{U<1/n\\}}.\\]\n\nThen \\(X_n\\to0\\) almost surely, but for \\(0<\\varepsilon<1\\),\n\n\\[\\sum_n\\mathbb P(|X_n|>\\varepsilon) = \\sum_n\\frac1n = \\infty.\\]\n\nExtracting an almost-surely convergent subsequence\n\nIf \\(X_n\\to X\\) in probability, a subsequence converges almost surely.\n\nChoose increasing indices \\(n_k\\) such that\n\n\\[\\mathbb P(|X_{n_k}-X|>2^{-k})\\leq2^{-k}.\\]\n\nSuch indices exist by convergence in probability. Since the upper bounds form a convergent geometric series, the first lemma implies that\n\n\\[|X_{n_k}-X|\\leq2^{-k}\\]\n\neventually almost surely. Thus \\(X_{n_k}\\to X\\) almost surely.\n\nA route to a strong law of large numbers\n\nThe lemmas often supply the final step in proofs of the strong law of large numbers. For example, let \\(X_1,X_2,\\ldots\\) be independent and identically distributed, with mean \\(\\mu\\) and finite fourth central moment. Set \\(Y_i=X_i-\\mu\\). Independence and centering give\n\n\\[\\mathbb E\\!\\left[\\left(\\sum_{i=1}^{n}Y_i\\right)^4\\right] = n\\mathbb E[Y_1^4] + 3n(n-1)\\bigl(\\mathbb E[Y_1^2]\\bigr)^2.\\]\n\nConsequently, for each \\(\\varepsilon>0\\), there is a finite constant \\(C\\), independent of \\(n\\), such that\n\n\\[\\mathbb P\\!\\left( \\left|\\frac1n\\sum_{i=1}^{n}X_i-\\mu\\right|>\\varepsilon \\right) \\leq \\frac{C}{\\varepsilon^4 n^2}.\\]\n\nThese bounds are summable, so the first lemma proves almost-sure convergence of the sample averages.\n\nThe fourth-moment assumption is sufficient for this short proof, not necessary for the general strong law for independent, identically distributed variables."
    },
    {
      "heading": "Measure-theoretic form and scope",
      "text": "The first lemma extends directly to measure theory. If \\((S,\\mathcal A,\\mu)\\) is any measure space and measurable sets \\(A_n\\) satisfy\n\n\\[\\sum_n\\mu(A_n)<\\infty,\\]\n\nthen\n\n\\[\\mu(\\limsup_n A_n)=0.\\]\n\nThe proof uses the same tail-union estimate and does not require \\(\\mu(S)\\) to be finite.\n\nThe second lemma is more sensitive to dependence. There are extensions based on conditional probabilities or controlled intersections, but divergence of the unconditional probability sum alone is insufficient. The practical distinction is therefore between a summability conclusion valid for arbitrary events and a divergence conclusion requiring additional structure."
    }
  ],
  "links": [
    "probability",
    "probability-space",
    "event-probability",
    "almost-surely",
    "infinite-series",
    "random-variable",
    "indicator-function",
    "expected-value",
    "statistical-independence",
    "chebyshevs-inequality",
    "uniform-distribution",
    "convergence-in-probability",
    "geometric-series",
    "law-of-large-numbers",
    "measure-theory"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}