{
  "id": "exponential-time-hypothesis",
  "title": "Exponential Time Hypothesis",
  "lang": "en",
  "source_lang": "en",
  "summary": "The Exponential Time Hypothesis conjectures that deciding satisfiability of 3-CNF Boolean formulas requires exponential time in the number of variables in the worst case.",
  "category": "computer-science",
  "subcategory": "algorithms-theory",
  "type": "concept",
  "aliases": [
    "ETH"
  ],
  "generated_at": "2026-11-05",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/exponential-time-hypothesis/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/exponential-time-hypothesis/",
      "json": "https://aiwiki.page/zh/exponential-time-hypothesis.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "The Exponential Time Hypothesis (ETH) is a conjecture in computational complexity asserting that satisfiability of Boolean formulas in 3-conjunctive normal form cannot be decided in subexponential time in the number of variables. More precisely, its standard formulation asserts a positive lower bound on the achievable exponential rate for solving 3-SAT. ETH strengthens the conjecture that \\(\\mathrm{P}\\ne\\mathrm{NP}\\): rather than merely excluding polynomial-time algorithms, it excludes a much wider range of running times. It is widely used to establish conditional lower bounds for exact algorithms and parameterized problems."
    },
    {
      "heading": "Formal definition",
      "text": "The Boolean satisfiability problem asks whether a Boolean formula has an assignment of truth values that makes it true. In \\(k\\)-SAT, the formula is in conjunctive normal form: it is an AND of clauses, each of which is an OR of at most \\(k\\) literals. A literal is a variable or its negation.\n\nLet \\(n\\) denote the number of variables and \\(L\\) the input length. For each fixed \\(k\\geq 3\\), define the optimal exponential rate by\n\n\\[s_k=\\inf\\left\\{s\\geq 0: k\\text{-SAT can be decided deterministically in time } 2^{sn}L^{O(1)} \\right\\}.\\]\n\nETH states that\n\n\\[s_3>0.\\]\n\nThus, there is a constant \\(\\delta>0\\) such that no deterministic algorithm decides every 3-SAT instance in time \\(2^{\\delta n}L^{O(1)}\\).\n\nETH is commonly summarized as the assertion that 3-SAT has no \\(2^{o(n)}\\)-time algorithm, with polynomial factors in the input length understood. Its positive-rate formulation makes the quantifiers explicit: it excludes not only a single subexponential algorithm, but also algorithms achieving arbitrarily small positive exponential rates.\n\nA running time such as \\(2^{\\sqrt n}\\) or \\(2^{n/\\log n}\\) is subexponential in \\(n\\), and would contradict ETH if achieved for unrestricted 3-SAT. By contrast, \\(1.1^n\\) remains exponential and does not, by itself, contradict the hypothesis.\n\nThe relevant time complexity is measured in the worst case. ETH says nothing directly about the difficulty of a typical formula, a particular application’s instances, or formulas drawn from a specified probability distribution."
    },
    {
      "heading": "Historical development",
      "text": "Russell Impagliazzo and Ramamohan Paturi introduced ETH in work published in journal form in 2001, amid research on the complexity of \\(k\\)-SAT and improved exact exponential-time algorithms.\n\nThe motivation was that NP-completeness provides only a coarse account of computational difficulty. Assuming \\(\\mathrm{P}\\ne\\mathrm{NP}\\), an NP-complete problem cannot have a polynomial-time algorithm, but that assumption alone does not distinguish between running times such as \\(2^{\\sqrt n}\\), \\(1.01^n\\), and \\(2^n\\).\n\nETH provides a framework for studying these distinctions. A central supporting result is the sparsification lemma, developed by Impagliazzo, Paturi, and Francis Zane. It makes it possible to transfer lower bounds from satisfiability while controlling both the number of variables and the number of clauses."
    },
    {
      "heading": "Sparsification and the choice of size measure",
      "text": "An exponential lower bound depends critically on what counts as input size. A 3-SAT formula with \\(n\\) variables can have substantially more than \\(n\\) clauses. A reduction whose output size depends on the clause count may therefore fail to preserve a lower bound expressed in terms of variables.\n\nThe sparsification lemma addresses this issue. For every fixed \\(k\\) and every \\(\\varepsilon>0\\), a \\(k\\)-CNF formula on \\(n\\) variables can be expressed as a disjunction of at most\n\n\\[2^{\\varepsilon n}\\]\n\n\\(k\\)-CNF formulas, each having at most \\(C_{k,\\varepsilon}n\\) clauses, where \\(C_{k,\\varepsilon}\\) depends only on \\(k\\) and \\(\\varepsilon\\). The decomposition can be computed in \\(2^{\\varepsilon n}\\) times a polynomial in the input length.\n\nThe original formula is satisfiable exactly when at least one formula in the decomposition is satisfiable. Consequently, sufficiently fast algorithms for sparse formulas would yield sufficiently fast algorithms for general formulas.\n\nIn particular, ETH implies that 3-SAT remains resistant to subexponential-time algorithms even when restricted to formulas with a linear number of clauses, with the sparsity constant chosen appropriately. This allows many lower-bound arguments to use \\(n+m\\), where \\(m\\) is the clause count, rather than \\(n\\) alone.\n\nSparsification is not simply the deletion of redundant clauses. It produces multiple formulas whose disjunction preserves the original formula’s satisfiability."
    },
    {
      "heading": "Relationship to other complexity assumptions",
      "text": "P versus NP\n\nETH implies a negative answer to the P versus NP problem:\n\n\\[\\mathrm{ETH}\\implies \\mathrm{P}\\ne\\mathrm{NP}.\\]\n\nIf \\(\\mathrm{P}=\\mathrm{NP}\\), then 3-SAT would have a polynomial-time algorithm and its optimal exponential rate would be zero.\n\nThe converse is not known. The assertion \\(\\mathrm{P}\\ne\\mathrm{NP}\\) does not rule out a subexponential but superpolynomial algorithm for 3-SAT.\n\nStrong Exponential Time Hypothesis\n\nThe Strong Exponential Time Hypothesis (SETH) makes a stronger claim about \\(k\\)-SAT as clause width increases:\n\n\\[\\lim_{k\\to\\infty}s_k=1.\\]\n\nEquivalently, for every \\(\\varepsilon>0\\), there is a fixed \\(k\\) for which \\(k\\)-SAT cannot be solved in time\n\n\\[2^{(1-\\varepsilon)n}L^{O(1)}.\\]\n\nSETH does not assert that 3-SAT requires essentially \\(2^n\\) time. Algorithms for fixed small clause widths can improve substantially on exhaustive enumeration without contradicting it.\n\nSETH implies ETH, using the sparsification lemma together with reductions between bounded-width satisfiability problems. The reverse implication is not known. ETH generally supports lower bounds on the order of growth of an exponent, whereas SETH can support sharper claims about the base of an exponential running time or the exponent of a polynomial running time.\n\nRandomized and counting variants\n\nStandard ETH concerns deterministic decision algorithms. Excluding bounded-error randomized algorithms requires a randomized version of the hypothesis, often called randomized ETH or rETH.\n\nThe counting Exponential Time Hypothesis, usually written \\(\\#\\)ETH, concerns computing the exact number of satisfying assignments. It asserts a positive exponential lower bound for counting satisfying assignments of 3-CNF formulas. A fast exact counting algorithm would also decide satisfiability, but hardness of counting alone does not establish the corresponding hardness of decision."
    },
    {
      "heading": "Conditional lower bounds through reductions",
      "text": "The main use of ETH is to transfer conjectured hardness from 3-SAT to other problems through a computational reduction.\n\nSuppose a reduction maps a sparse 3-SAT formula with \\(n\\) variables to an equivalent instance of a problem \\(Q\\), with size \\(N=O(n)\\). A \\(2^{o(N)}\\)-time algorithm for \\(Q\\) would then give a \\(2^{o(n)}\\)-time algorithm for 3-SAT, contradicting ETH.\n\nThe reduction’s size increase is essential. If it produces \\(N=O(n^2)\\), the directly supported lower bound is instead against running time\n\n\\[2^{o(\\sqrt N)}.\\]\n\nOrdinary proofs of NP-hardness often allow polynomial increases in instance size. Such proofs do not automatically establish tight exponential-time lower bounds.\n\nExact exponential-time algorithms\n\nETH is used to show that many graph and combinatorial problems have no subexponential-time exact algorithms. For example, standard linear-size reductions from sparse 3-SAT imply that the general Vertex Cover problem has no \\(2^{o(N)}\\)-time algorithm under ETH, where \\(N\\) is the number of graph vertices.\n\nSuch a result does not exclude improvements from \\(2^N\\) to \\(1.3^N\\), nor does it exclude faster algorithms on restricted graph classes. The lower bound must be read with its precise instance restrictions and size parameter.\n\nParameterized complexity\n\nIn parameterized complexity, running times are expressed using both input size \\(N\\) and a parameter \\(k\\). A typical fixed-parameter running time has the form\n\n\\[f(k)N^{O(1)},\\]\n\nwhere \\(f\\) is a computable function.\n\nETH can establish stronger quantitative barriers than a classification as parameterized hard. A standard example is the Clique problem: under ETH, finding whether a graph contains a clique of size \\(k\\) cannot be done in time\n\n\\[f(k)N^{o(k)}\\]\n\nfor any computable function \\(f\\). This rules out replacing the familiar linear dependence on \\(k\\) in the exponent of \\(N\\) by a sublinear one.\n\nETH also supports lower bounds in terms of structural graph parameters such as treewidth. These bounds help determine whether the exponential parameter dependence of a dynamic programming algorithm can be substantially improved."
    },
    {
      "heading": "Scope and limitations",
      "text": "ETH remains an unproved conjecture. A statement established “under ETH” is a conditional theorem: an algorithm violating the stated bound would refute ETH, rather than contradict a proven unconditional lower bound.\n\nSeveral distinctions are important when interpreting such results:\n\n- Worst case versus typical performance. ETH is compatible with efficient performance on many practical instances.\n- General versus restricted inputs. Special structural restrictions can permit subexponential or polynomial-time algorithms.\n- Decision versus approximation. A lower bound for exact decision does not automatically apply to an approximation algorithm.\n- Computational model. Standard ETH does not by itself exclude randomized or quantum algorithms with the same running time.\n- Parameter choice. A lower bound in the number of variables, clauses, vertices, edges, or a structural parameter is not interchangeable with a bound in another measure.\n\nThese qualifications are part of the mathematical content of an ETH-based lower bound: the reduction, computational model, input restrictions, and size measure determine exactly which algorithms the result excludes."
    }
  ],
  "links": [
    "computational-complexity",
    "boolean-satisfiability-problem",
    "algorithm",
    "time-complexity",
    "np-completeness",
    "p-versus-np-problem",
    "strong-exponential-time-hypothesis",
    "randomized-algorithm",
    "np-hardness",
    "dynamic-programming",
    "approximation-algorithm"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}