{
  "id": "zorns-lemma",
  "title": "Zorn's Lemma",
  "lang": "en",
  "source_lang": "en",
  "summary": "Zorn's lemma is a maximal-element principle for partially ordered sets, equivalent to the axiom of choice and widely used in existence proofs.",
  "category": "mathematics",
  "subcategory": "discrete-logic",
  "type": "concept",
  "aliases": [
    "Zorn lemma",
    "Kuratowski–Zorn lemma"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/zorns-lemma/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/zorns-lemma/",
      "json": "https://aiwiki.page/zh/zorns-lemma.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Zorn's lemma is a principle of set theory asserting that a nonempty partially ordered set has a maximal element if every chain in it has an upper bound within the set. Over Zermelo–Fraenkel set theory without choice, it is equivalent to the axiom of choice. It is especially useful for proving that an object can be extended to one that admits no further extension, even when no explicit construction of such an object is available. Standard applications include the existence of bases for vector spaces, maximal ideals in rings, and extensions of linear functionals."
    },
    {
      "heading": "Statement and terminology",
      "text": "Let \\((P,\\leq)\\) be a nonempty set equipped with a partial order. A chain in \\(P\\) is a subset \\(C\\subseteq P\\) whose elements are pairwise comparable: for every \\(x,y\\in C\\), either \\(x\\leq y\\) or \\(y\\leq x\\).\n\nAn upper bound of \\(C\\) in \\(P\\) is an element \\(u\\in P\\) such that\n\n\\[c\\leq u\\qquad\\text{for every }c\\in C.\\]\n\nThe upper bound need not belong to \\(C\\), and it need not be the least possible upper bound.\n\nZorn's lemma states:\n\nIf every chain in a nonempty partially ordered set \\(P\\) has an upper bound in \\(P\\), then \\(P\\) contains a maximal element.\n\nAn element \\(m\\in P\\) is maximal if\n\n\\[m\\leq x\\quad\\Longrightarrow\\quad x=m\\]\n\nfor every \\(x\\in P\\). Thus no element of \\(P\\) lies strictly above \\(m\\).\n\nThe nonemptiness condition can instead be incorporated into the chain condition. If the empty set is counted as a chain and every chain must have an upper bound in \\(P\\), the hypothesis itself implies that \\(P\\) is nonempty. Another common formulation explicitly assumes \\(P\\neq\\varnothing\\) and requires upper bounds only for nonempty chains.\n\nMaximal is not greatest\n\nA greatest element \\(g\\) satisfies \\(x\\leq g\\) for every \\(x\\in P\\). A maximal element need only have no strictly larger element; it may be incomparable with many other elements.\n\nFor example, in the set\n\n\\[P=\\bigl\\{\\{a\\},\\{b\\}\\bigr\\}\\]\n\nordered by inclusion, both elements are maximal, but neither is greatest. Zorn's lemma guarantees a maximal element, not a greatest element or a unique one.\n\nReversing the order gives the dual statement: if every chain in a nonempty partially ordered set has a lower bound, then the set has a minimal element."
    },
    {
      "heading": "Why the chain condition matters",
      "text": "The chain condition expresses a form of stability under successive extensions. Individual objects may admit larger extensions, but a compatible collection of such extensions must still have a common extension belonging to the permitted class.\n\nThis condition is weaker than requiring an upper bound for every subset. It also does not require the upper bound to be a least upper bound.\n\nSeveral examples clarify its scope:\n\n- Every nonempty finite partially ordered set has a maximal element. Every nonempty chain in such a set has a greatest member, which supplies an upper bound.\n- The natural numbers with their usual order have no maximal element. The hypothesis fails because the chain consisting of all natural numbers has no upper bound among the natural numbers.\n- For a set \\(X\\), its power set ordered by inclusion satisfies the hypothesis: the union of any chain is an upper bound. Its greatest element is \\(X\\).\n- The finite subsets of an infinite set, ordered by inclusion, have no maximal element. A union of a chain of finite subsets need not be finite, so the same union argument does not establish the hypothesis.\n\nThe final example illustrates a common error: producing an upper bound in some larger collection is insufficient. The upper bound must belong to the particular partially ordered set to which the lemma is applied."
    },
    {
      "heading": "Equivalence with the axiom of choice",
      "text": "Zorn's lemma is not merely a consequence of choice that happens to be convenient. Within Zermelo–Fraenkel set theory, it has exactly the same logical strength as the axiom of choice. It is also equivalent to the well-ordering theorem, which states that every set admits a well-ordering.\n\nFrom Zorn's lemma to choice\n\nLet \\((A_i)_{i\\in I}\\) be an indexed family of nonempty sets. A partial choice function selects an element of \\(A_i\\) for each index in some subset \\(J\\subseteq I\\).\n\nLet \\(P\\) be the set of all such partial choice functions, ordered by extension. It is nonempty because it contains the empty function.\n\nIf \\(C\\) is a chain of partial choice functions, their union\n\n\\[f_C=\\bigcup_{f\\in C}f\\]\n\nis again a partial choice function. The chain condition ensures agreement on overlapping domains: of any two functions in the chain, one extends the other. Thus \\(f_C\\) is an upper bound in \\(P\\).\n\nZorn's lemma supplies a maximal partial choice function \\(f\\). If an index \\(i\\in I\\) were absent from its domain, an element \\(a\\in A_i\\) would give a proper extension\n\n\\[f\\cup\\{(i,a)\\},\\]\n\ncontradicting maximality. Hence \\(f\\) is defined on all of \\(I\\), establishing the axiom of choice.\n\nFrom choice to Zorn's lemma\n\nAssume the axiom of choice, and let \\(P\\) satisfy the hypotheses of Zorn's lemma. Suppose, for a contradiction, that \\(P\\) has no maximal element.\n\nFor every chain \\(C\\subseteq P\\), choose an upper bound \\(u_C\\). Since \\(u_C\\) is not maximal, there is an element strictly above it. Choice permits these selections to be made simultaneously for all chains.\n\nStarting with the empty chain, transfinite recursion can then repeatedly select an element strictly above every previously selected element. At each stage the earlier elements form a chain, so the selected extension exists.\n\nThis would produce a strictly increasing sequence indexed by arbitrarily large ordinals. But for any set \\(P\\), there is an ordinal that cannot be injected into \\(P\\); this fact is provable without choice. A strictly increasing sequence of that length would supply such an injection. The contradiction shows that \\(P\\) must have a maximal element.\n\nThe need for transfinite rather than merely countable extension is important. An ordinary infinite sequence of extensions may have an upper bound that can itself be extended further."
    },
    {
      "heading": "A standard proof pattern",
      "text": "Many applications follow the same structure:\n\n- Specify the objects. Form a set \\(P\\) of partial constructions or admissible extensions.\n- Order by extension. Usually the order is inclusion of sets, functions, or algebraic structures.\n- Establish nonemptiness. Provide an initial admissible object.\n- Bound every chain. Usually take a union and prove that it remains admissible.\n- Apply Zorn's lemma. Obtain a maximal admissible object.\n- Use maximality. Show that failure of the desired property would permit a proper admissible extension.\n\nThe chain-union step often works because the defining conditions involve only finitely many elements at a time. Any finite selection of elements from the union of a chain already lies in one member of that chain. Properties such as closure under binary operations and linear independence are therefore often preserved.\n\nThe final step is equally essential: maximality in the chosen order does not automatically imply the mathematical property sought. That implication must be proved separately."
    },
    {
      "heading": "Applications",
      "text": "Bases of vector spaces\n\nLet \\(V\\) be a vector space over a field \\(F\\). Consider all linearly independent subsets of \\(V\\), ordered by inclusion. This collection contains the empty set.\n\nFor any chain \\(C\\), let\n\n\\[U=\\bigcup_{A\\in C}A.\\]\n\nThe set \\(U\\) is linearly independent. Indeed, a linear dependence would involve only finitely many vectors. Because \\(C\\) is a chain, those vectors would all lie in a single member of \\(C\\), contradicting its independence.\n\nZorn's lemma therefore gives a maximal linearly independent set \\(B\\). If a vector \\(v\\in V\\) lay outside the linear span of \\(B\\), then \\(B\\cup\\{v\\}\\) would remain linearly independent. Hence \\(B\\) spans \\(V\\) and is a basis.\n\nThe same argument, restricted to independent sets containing a specified independent set, proves that every linearly independent set extends to a basis.\n\nFor infinite-dimensional spaces, this means an algebraic or Hamel basis: each vector is a finite linear combination of basis vectors. It should not be confused with an orthonormal basis in a Hilbert space, where expansions may involve convergent infinite sums.\n\nMaximal ideals in rings\n\nLet \\(R\\) be a nonzero commutative ring with identity. Consider its proper ideals, ordered by inclusion. The zero ideal supplies an initial element.\n\nThe union of a nonempty chain of proper ideals is an ideal. It remains proper because, if it contained \\(1\\), some ideal in the chain would contain \\(1\\) and therefore equal \\(R\\). The empty chain has the zero ideal as an upper bound.\n\nZorn's lemma gives a maximal ideal of \\(R\\). More generally, applying the argument to proper ideals containing a fixed proper ideal \\(I\\) shows that every proper ideal is contained in a maximal ideal.\n\nThis is a basic existence result in abstract algebra and algebraic geometry. For a commutative ring with identity, the quotient by a maximal ideal is a field.\n\nExtending linear functionals\n\nA standard proof of the real Hahn–Banach theorem uses Zorn's lemma. Suppose a linear functional is defined on a subspace of a real vector space and is dominated by a sublinear function.\n\nConsider all linear extensions that preserve this domination, ordered by extension. The union of a chain is again a dominated linear functional on the union of its domains. Zorn's lemma gives a maximal extension.\n\nA separate one-dimensional extension argument shows that its domain cannot be a proper subspace of the original vector space. Consequently, it is defined everywhere.\n\nThis illustrates the division of work in a Zorn argument: the lemma produces a maximal object, while a specific mathematical extension theorem establishes that the maximal object has the required domain. Hahn–Banach is a central tool of functional analysis, but its usual real form does not require the full strength of the axiom of choice.\n\nMaximal chains and compactness\n\nThe Hausdorff maximal principle states that every chain in a partially ordered set is contained in a maximal chain. Applying Zorn's lemma to the set of chains containing a given chain proves this principle: the union of a chain of such chains is itself a chain.\n\nConversely, a maximal chain in a partially ordered set satisfying Zorn's upper-bound hypothesis has an upper bound, and that upper bound must be maximal. Thus the two principles are equivalent over Zermelo–Fraenkel set theory.\n\nIn topology, choice-based maximality arguments also occur in proofs of Tychonoff's theorem, which states that a product of compact spaces is compact in the product topology. The general theorem for arbitrary compact spaces is equivalent to the axiom of choice. The restriction to compact Hausdorff spaces is weaker, so the exact formulation matters when assessing logical strength."
    },
    {
      "heading": "Logical scope and limitations",
      "text": "Zorn's lemma is an existence principle, not a general algorithm. Its conclusion does not identify a preferred maximal element, prove uniqueness, or give a procedure for finding one. Particular applications may nevertheless admit explicit constructions or require less than full choice.\n\nSeveral distinctions are important:\n\n- Set-sized domains: The usual lemma applies to partially ordered sets, not arbitrary proper classes. A collection informally described as “all structures” must be restricted to an appropriate set before the ordinary lemma applies.\n- All chains, not just sequences: Bounding every countable increasing sequence does not establish the hypothesis. Chains can be uncountable.\n- Upper bounds must be admissible: A union may be an upper bound as a set while failing the required finiteness, closure, continuity, or other condition.\n- A use of Zorn does not prove equivalence to choice: A theorem proved using Zorn's lemma may have a weaker proof. Establishing that it implies choice requires a separate argument.\n\nBecause Zorn's lemma implies the axiom of choice, it cannot be proved in choice-free Zermelo–Fraenkel set theory if that theory is consistent. Its acceptance in ordinary classical mathematics reflects the widespread use of Zermelo–Fraenkel set theory with choice as a foundational framework."
    },
    {
      "heading": "History",
      "text": "The principle is named after Max Zorn, who published a maximality principle in 1935 and emphasized its usefulness in algebra. Kazimierz Kuratowski had published an equivalent form in 1922, which accounts for the alternative name Kuratowski–Zorn lemma.\n\nRelated maximality principles appeared earlier, notably Hausdorff's maximal principle in 1909. Zorn's formulation became particularly influential because its hypotheses fit extension arguments directly: instead of explicitly well-ordering a collection and carrying out a long transfinite construction, a proof can verify upper bounds for chains and then work with a maximal object."
    }
  ],
  "links": [
    "set-theory",
    "zermelo-fraenkel-set-theory",
    "axiom-of-choice",
    "partial-order",
    "empty-set",
    "natural-number",
    "power-set",
    "well-ordering-theorem",
    "ordinal-number",
    "linear-independence",
    "vector-space",
    "field-mathematics",
    "linear-span",
    "basis-linear-algebra",
    "orthonormal-basis",
    "ring-mathematics",
    "ideal-ring-theory",
    "abstract-algebra",
    "algebraic-geometry",
    "hahn-banach-theorem",
    "linear-functional",
    "functional-analysis",
    "topology"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}