Fact-checked by Grok 2 weeks ago

Completion

In , the completion of a (X, d) is defined as a pair consisting of a (Y, \rho) and an isometric embedding \iota: X \to Y such that the \iota(X) is dense in Y. This construction addresses the incompleteness of X by adjoining limits of all s from X, ensuring that every in X converges within Y. The existence of a completion is guaranteed for every , and any two completions of the same space are uniquely isomorphic via an that extends the identity on X. A prototypical example is the completion of the \mathbb{Q} under the standard metric, which produces the \mathbb{R} as the completed space, embedding \mathbb{Q} densely into \mathbb{R}. The standard construction identifies points in Y with equivalence classes of Cauchy sequences from X, where two sequences are equivalent if their distance approaches zero, and defines the metric on these classes via the of distances between sequences. This process is functorial, preserving metric space morphisms, and forms the foundation for developing on incomplete spaces like rational function fields or certain function spaces. Beyond metric spaces, completion appears in other mathematical domains with analogous goals of achieving completeness. In commutative algebra, the I-adic completion of a ring R with respect to an ideal I is the inverse limit \varprojlim R/I^n, yielding a complete topological ring in the I-adic topology where R embeds densely. This construction is essential in algebraic geometry for studying local properties and deformations, such as in the completion of the ring of integers at a prime ideal to obtain p-adic integers. In order theory, the Dedekind–MacNeille completion of a partially ordered set P is the smallest complete lattice L containing P as an order-dense subposet, constructed via cuts or ideals to preserve existing suprema and infima. These varied completions highlight a unifying theme: extending structures minimally to enable the existence of limits or meets/joins, facilitating proofs of convergence, compactness, and continuity in diverse contexts.

Mathematics

Metric completion

In mathematical analysis, a complete metric space is defined as a metric space (X, d) in which every Cauchy sequence converges to a point within the space. A Cauchy sequence (x_n) in (X, d) is one where the terms become arbitrarily close as n increases, formalized by the condition that for every \varepsilon > 0, there exists N \in \mathbb{N} such that d(x_m, x_n) < \varepsilon for all m, n > N. This property ensures that the space captures all possible limits of such sequences, making it "complete" in the sense that no points are "missing" to serve as limits. Not all metric spaces are complete; for instance, the rational numbers \mathbb{Q} equipped with the standard Euclidean metric d(p, q) = |p - q| form an incomplete space because sequences of rationals can converge to irrational limits outside \mathbb{Q}. To remedy this, the metric completion of an arbitrary metric space (X, d) is constructed by considering the set of all Cauchy sequences in X. Two such sequences (x_n) and (y_n) are declared equivalent, denoted (x_n) \sim (y_n), if \lim_{n \to \infty} d(x_n, y_n) = 0. The completion \hat{X} is then the quotient set of equivalence classes under this relation, denoted [(x_n)] for the class of (x_n). This space is endowed with a metric \hat{d}([(x_n)], [(y_n)]) = \lim_{n \to \infty} d(x_n, y_n), which is well-defined because equivalent sequences yield the same limit. The resulting (\hat{X}, \hat{d}) is a complete metric space, as every Cauchy sequence in \hat{X} converges within it. The original space X embeds into \hat{X} via the isometry \iota: X \to \hat{X} defined by \iota(x) = [(x, x, \dots)], the of the constant sequence at x. This embedding is , preserving distances since \hat{d}(\iota(x), \iota(y)) = d(x, y), and the \iota(X) is dense in \hat{X} because every element of \hat{X} is the from X. A canonical example is the completion of (\mathbb{Q}, |\cdot|), which yields the real numbers \mathbb{R} with the standard ; here, equivalence classes of Cauchy sequences of correspond to real numbers, with irrational limits arising from sequences like those approximating \sqrt{2}. Another illustrative example involves function spaces: the space of continuous functions C[0,1] on the compact interval [0,1] equipped with the supremum norm \|f\|_\infty = \sup_{x \in [0,1]} |f(x)| is already complete, forming a Banach space. However, subspaces like the polynomials on [0,1] under the same norm have C[0,1] as their completion, by the Weierstrass approximation theorem, which guarantees dense approximation of continuous functions by polynomials. In contrast, considering discontinuous extensions, the completion process highlights how sup-norm limits can include functions of bounded variation, though C[0,1] itself remains the core complete space for uniform convergence. The completion is unique up to isometry over the original space: if (\hat{X}_1, \hat{d}_1) and (\hat{X}_2, \hat{d}_2) are two completions of (X, d), then there exists a unique isometry \phi: \hat{X}_1 \to \hat{X}_2 such that \phi \circ \iota_1 = \iota_2, where \iota_1 and \iota_2 are the respective embeddings. This uniqueness ensures that the completion is intrinsically determined by the metric structure of X. The foundational formalization of metric spaces traces back to Maurice Fréchet's 1906 doctoral thesis, Sur quelques points du calcul fonctionnel, which laid the groundwork for modern . The explicit construction of completions was developed in the subsequent decades.

Order completion

Order completion extends partially ordered sets to structures where suprema and infima exist for all subsets, typically resulting in complete lattices or chains. A , or poset, is an (X, \leq) consisting of a set X and a binary relation \leq that is reflexive, antisymmetric, and transitive. A is a poset (L, \leq) in which every subset M \subseteq L has both a least upper bound, called the supremum \sup M, and a greatest lower bound, called the infimum \inf M. These completions preserve the original order while adjoining new elements to ensure completeness. This approach was pioneered by in 1872 for linear orders using cuts. For dense linear orders, the Dedekind completion adjoins elements via Dedekind cuts to form a complete chain. A Dedekind cut in a linear order is a partition of the set into a lower set A and an upper set B such that A has no maximum and B has no minimum. The completion is the set of all such cuts, ordered by inclusion of lower sets, yielding a complete linear order. A canonical example is the completion of the rational numbers \mathbb{Q} under the usual order, where cuts correspond to real numbers \mathbb{R}, filling gaps like those at \sqrt{2}. This construction ensures every nonempty bounded subset has a supremum and infimum. The MacNeille completion, introduced by Henry M. MacNeille in , provides an embedding of any poset into a using Dedekind-MacNeille cuts. For a poset P, the completion consists of all subsets A \subseteq P such that A = \downarrow (\uparrow A), where \uparrow A is the set of upper bounds of A and \downarrow B is the set of lower bounds of B, ordered by inclusion. This yields the smallest containing P as an order-embedded subset, preserving all existing suprema and infima of P. Viewing posets as categories (with elements as objects and \leq as unique morphisms), the Yoneda completion is the free cocompletion, given by the category of presheaves on the poset. This embeds the poset into the presheaf category \hat{P} = [P^{op}, \mathbf{Set}], where every object is a colimit of representables, ensuring all small colimits (suprema) exist. The embedding is full and faithful, reflecting the Yoneda lemma's principle that the poset is determined by its hom-functors. Examples illustrate these completions. The poset of natural numbers \mathbb{N} under divisibility (m \leq n if m divides n) completes via order ideals to the lattice of divisor-closed subsets of \mathbb{N}, ordered by inclusion, which is a distributive complete lattice. Similarly, the poset of all subsets of a set X under inclusion is incomplete if restricted to finite subsets, but its Dedekind-MacNeille completion is the full power set lattice \mathcal{P}(X), a complete Boolean lattice where every subset has arbitrary unions and intersections as suprema and infima. Key properties hold across these completions. The embedding into the completion is an order-embedding, meaning it is injective and preserves and reflects the order: p \leq q in the poset if and only if the images satisfy the same in the completion. If the original poset is conditionally complete—meaning every nonempty subset with an upper bound has a supremum—the image is dense in the completion, so every element of the completion is a supremum of elements from the original poset.

Ring completion

In commutative algebra, the completion of a ring R with respect to a proper ideal I is defined as the inverse limit \hat{R} = \lim_{\leftarrow} R/I^n, where the elements are coherent sequences (x_n \mod I^n)_{n \geq 1} in the product \prod_{n=1}^\infty R/I^n satisfying the compatibility condition x_{n+1} \equiv x_n \pmod{I^n} for all n. This construction equips \hat{R} with a structure via componentwise and . The on R underlying this completion is the I-adic topology, in which the neighborhoods of 0 are the subgroups I^n for n \geq 1. The completion \hat{R} inherits a topology making it a complete topological , and it is Hausdorff if and only if \bigcap_{n=1}^\infty I^n = \{0\}. For a module M over R, the completion with respect to I is similarly given by \hat{M} = \lim_{\leftarrow} M/I^n M, with elements as coherent sequences (m_n \mod I^n M)_{n \geq 1} satisfying m_{n+1} \equiv m_n \pmod{I^n M}. There is a canonical R-module homomorphism M \to \hat{M}, and under the Hausdorff condition, this map is injective. A representative example is the of p-adic integers \mathbb{Z}_p, which is the completion of \mathbb{Z} with respect to the (p) for a prime p. Another is the of k[] over a k, obtained as the completion of the k with respect to the (x). The completion satisfies a : for any I-adically complete S and continuous \phi: R \to S, there exists a unique continuous extension \hat{\phi}: \hat{R} \to S such that \phi = \hat{\phi} \circ \iota, where \iota: R \to \hat{R} is the . Moreover, completions preserve exact sequences under suitable conditions, such as when the modules are finitely presented. Applications include , which leverages the completeness of \hat{R} to lift solutions of equations from R/I to \hat{R} under certain non-singularity conditions. In , completions of rings at prime ideals yield local rings that facilitate the study of local-global principles and ramification in number fields. This construction was developed by Wolfgang Krull in the 1930s as part of advancing ideal theory in .

Computer science

Autocompletion

Autocompletion, also known as or autosuggest, is a feature that algorithmically generates and displays completions for partial inputs in text fields, such as words, phrases, or commands, to accelerate and reduce errors. This functionality relies on analyzing user input prefixes against predefined dictionaries or learned patterns to offer relevant suggestions in , often via dropdown menus or inline predictions. By anticipating , autocompletion enhances efficiency in applications like search engines, composition, and mobile messaging, where incomplete queries like "best pizz" might expand to "best pizza near me". The origins of autocompletion trace back to the early in desktop email clients, such as Eudora version 5.1, which introduced address suggestion based on a local history file to streamline recipient entry during composition. The feature gained widespread adoption with the launch of Suggest in August 2004, which used server-side processing to provide dynamic query completions drawn from aggregated search logs, transforming search interfaces by reducing keystrokes and aiding query formulation. On mobile devices, autocompletion evolved further with the introduction of intelligent keyboards; for instance, SwiftKey, released in 2010, popularized context-aware predictions by leveraging user typing patterns across apps. As of 2025, autocompletion has further advanced with large language models for more sophisticated, context-aware predictions in search and messaging applications. Core mechanisms for autocompletion include (prefix tree) data structures, which enable efficient storage and retrieval of strings by organizing them into a tree where each node represents a character, facilitating rapid prefix matching for suggestions. Probabilistic approaches, such as n-gram models that estimate word sequence likelihoods or Markov chains that model transitions between states (e.g., words), generate predictions based on historical frequency data. enhances these by incorporating user-specific data, like search histories or typing behaviors, to prioritize relevant options; for example, integrates past queries while offering incognito modes to bypass this. In practice, autocompletion appears in dropdowns, where partial input like "best pizza in" triggers suggestions such as "best pizza in " based on popular completions. Mobile keyboards exemplify advanced implementations; SwiftKey employs to deliver context-sensitive predictions, adapting to slang, emojis, or multilingual inputs from user data. Algorithms vary from simple exact prefix searches in tries, which return all matching entries, to sophisticated enhancements like TF-IDF scoring for ranking suggestions by term importance in query logs, or recurrent neural networks (RNNs) that capture sequential dependencies for more accurate next-word forecasting in extended text. Privacy issues stem from the collection of input for and personalizing models, which can reveal sensitive patterns like location interests or habits if aggregated across sessions. To address this, providers implement safeguards such as options and browsing, where suggestions revert to generic, non-personalized defaults without storing or using history. Quality assessment of autocompletion systems employs metrics like mean reciprocal rank (MRR), which averages the reciprocal of the rank position of the first relevant suggestion across queries, yielding values from 0 (poor) to 1 (ideal) to quantify how promptly useful options appear. High MRR indicates effective systems that place desired completions near the top, balancing speed and relevance in user evaluations.

Code completion

Code completion is an automated feature in integrated development environments (IDEs) that provides context-aware suggestions for code elements, such as method names, parameters, variables, and snippets, to streamline software development. It analyzes the surrounding code to predict and offer relevant completions, reducing manual typing and aiding in syntax recall. This functionality enhances developer efficiency by integrating seamlessly into the coding workflow, often triggered by partial input or specific keystrokes. The feature emerged in early during the late 1980s, with foundational integrated environments like Borland's introducing basic assistance tools that evolved into modern . By the early 2000s, provided advanced content assist capabilities, including keyword and member suggestions based on semantics. , released in 2015, popularized IntelliSense for cross-language support, leveraging language servers for real-time suggestions. As of 2025, has integrated advanced large language models, enhancing multi-line predictions and semantic understanding in tools like . Core components include syntax parsing for static code analysis, which identifies structural elements; semantic understanding via abstract syntax trees (ASTs) to represent program hierarchy; and dynamic to deduce variable types without explicit declarations. These elements enable precise suggestions by processing the codebase's context, such as resolving function signatures or import requirements. Techniques range from rule-based systems, which rely on keyword matching and predefined heuristics for simple predictions, to (ML)-based approaches using transformer models for probabilistic next-token prediction trained on vast code repositories. Seminal ML advancements, such as those combining semantic engines with transformers, have improved completion accuracy by re-ranking suggestions and generating multi-line code. Integration with the (LSP), introduced by in 2016, standardizes these techniques across IDEs, allowing language-specific servers to handle completion requests via for interoperability. Practical examples illustrate its utility: in , typing a list variable followed by a dot might suggest append() based on type context; in , it can automatically generate import statements for referenced classes; and AI-driven tools like use large language models to predict entire lines or functions from partial code. These suggestions draw from trained models like variants, fine-tuned on public repositories. Benefits include reduced syntax errors through validated suggestions and boosted , with studies showing 6% reduction in time in internal evaluations. However, challenges arise from over-suggestion of irrelevant options, which can disrupt , or inaccurate completions in ambiguous or novel contexts, potentially increasing frustration for experienced developers. extends general text autocompletion principles but emphasizes programming-specific semantics like types and scopes.

Logic program completion

Logic program completion, commonly referred to as Clark's completion, is a formal transformation that converts a definite logic program into equivalent sentences, capturing its declarative meaning under the . Introduced by Keith Clark in 1977, this method equates the procedural interpretation of the program—defined by its success set of computable queries—with its declarative semantics in terms of the least Herbrand model, thereby enabling and analysis of logic programs. For a logic program P consisting of Horn clauses of the form \forall \mathbf{v} (Head \leftarrow Body), where Head is an atomic formula and Body is a conjunction of literals, Clark's completion Comp(P) is constructed by defining, for each predicate p with arity n, the completion formula: Comp(p) = \forall x_1 \dots x_n \left( p(x_1, \dots, x_n) \leftrightarrow \bigvee_{i=1}^k \exists \mathbf{y}_i \, Body_i(x_1, \dots, x_n, \mathbf{y}_i) \right), where the disjunction ranges over all clauses defining p, and \mathbf{y}_i are the existentially quantified variables in the body of the i-th clause not appearing in the head. The full completion Comp(P) is the conjunction of Comp(p) over all predicates p in P, augmented with the unique names assumption (distinguishing constants) and the domain closure assumption (restricting the Herbrand universe to program constants). This transformation assumes function-free programs and ensures that models of Comp(P) correspond to supported models of P. A representative example is the transitive closure program for ancestry:
  • ancestor(X,Y) \leftarrow parent(X,Y).
  • ancestor(X,Y) \leftarrow parent(X,Z), ancestor(Z,Y).
The completion yields: \forall X Y \left( ancestor(X,Y) \leftrightarrow parent(X,Y) \vee \exists Z \left( parent(X,Z) \wedge ancestor(Z,Y) \right) \right). This biconditional asserts that X is an of Y if and only if there is a direct relation or a chain through some intermediary Z, precisely capturing the least fixed point semantics without extraneous models. Despite its foundational role, Clark's completion has limitations: it assumes stratified, function-free programs and struggles with negation as failure in non-stratified cases, where models of Comp(P) may not align with stable models, leading to inconsistencies or unsupported facts. For instance, programs with even loops involving , such as p \leftarrow \neg q. \quad q \leftarrow \neg p., yield completions with no classical models, though stable models exist. Extensions address these issues, including three-valued completions for disjunctive logic programs, which incorporate partial truth values to handle unsupported atoms and enable semantics. These have applications in program verification, debugging, and knowledge representation.

References

  1. [1]
    [PDF] completion.pdf - Mathematics
    Definition 2.2. Let X be a metric space. A completion of X is a pair (X/,ιX) where X/ is a complete metric space and ι : X → X/ is an isometry with dense image.
  2. [2]
    [PDF] Completing a Metric Space - Rose-Hulman
    Recall that a metric space M is said to be complete if every Cauchy se- quence in M converges to a limit in M. Not all metric spaces are complete,.
  3. [3]
    [PDF] Completion of a Metric Space - IISER Pune
    A completion of a metric space (X, d) is a pair consisting of a complete metric space (X∗,d∗) and an isometry ϕ: X → X∗ such that ϕ[X] is dense in X∗. Theorem 1 ...
  4. [4]
    [PDF] The Completion of a Metric Space - Northwestern Math Department
    The goal of these notes is to construct a complete metric space which contains X as a subspace and which is the “smallest” space with respect to these two.<|control11|><|separator|>
  5. [5]
    Complete Metric Spaces - Advanced Analysis
    Jan 17, 2024 · A metric space ( X , d ) is called complete when all Cauchy sequences in the metric space are convergent. Examples. The real numbers with the ...
  6. [6]
    Section 10.96 (00M9): Completion—The Stacks project
    We say R is I-adically complete if R is I-adically complete as an R-module. It is not true that the completion of an R-module M with respect to I is I-adically ...
  7. [7]
    [PDF] Dedekind-MacNeille and Related Completions: Subfitness ...
    Apr 9, 2025 · We start by recalling that a completion of a poset P is a pair (C,e) where C is a complete lattice and e : P → C is a poset embedding. As ...<|control11|><|separator|>
  8. [8]
    [PDF] 10. Completion - Brandeis
    Completion. The real numbers are the completion of the rational numbers with respect to the usual absolute value norm. This means that any Cauchy.
  9. [9]
    [PDF] Cauchy sequences, Complete metric spaces - Purdue Math
    A Cauchy sequence {} in. IRK is always convergent. Page 10. Def: A metric space in which every Cauchy sequence converges is said be complete. to be. Ex: From ...
  10. [10]
    [PDF] [6pt] @let@token Linear Analysis Lecture 6
    Elements of Cc(Rn) are continuous functions with compact support. • C0(Rn) is complete in the sup-norm. This can either be shown directly, or by showing ...
  11. [11]
    [PDF] Chapter 2: Continuous Functions - UC Davis Math
    In this chapter, we study linear spaces of continuous functions on a compact set equipped with the uniform norm. These function spaces are our first examples of ...
  12. [12]
    Maurice Fréchet (1878 - 1973) - Biography - MacTutor
    In it he introduced the concept of a metric space, although he did not invent the name 'metric space' which is due to Hausdorff. The thesis concerns 'functional ...
  13. [13]
    Partially Ordered Set -- from Wolfram MathWorld
    A partially ordered set (or poset) is a set taken together with a partial order on it. Formally, a partially ordered set is defined as an ordered pair.
  14. [14]
    Complete Lattice -- from Wolfram MathWorld
    A partially ordered set (or ordered set or poset for short) (L,<=) is called a complete lattice if every subset M of L has a least upper bound.
  15. [15]
    None
    Summary of each segment:
  16. [16]
    [PDF] On lattices and their ideal lattices, and posets ... - UC Berkeley math
    Abstract. For P a poset or lattice, let Id(P) denote the poset, respectively, lattice, of upward directed downsets in P, including the empty set,.Missing: divisibility | Show results with:divisibility
  17. [17]
    COMPLETIONS OF ORDERED SETS - Cambridge University Press
    A completion of a. q.o. set A is a pair (C, e) where C is a complete p.o. set (i.e. every subset of C has a supremum) and e:A->C is an embedding. A subset B of ...
  18. [18]
    [PDF] FILTERED RINGS AND MODULES. GRADINGS AND COMPLETIONS.
    One important example of I-adic completion is A = Z, I = (p), where p is a prime. One obtains the p-adic integers Zp = lim. ←− Z/p n. Z, which can be regarded.
  19. [19]
    Section 10.153 (04GE): Henselian local rings—The Stacks project
    A henselian local ring (R, m, kappa) has the property that for every monic f in R[T] and root a_0 of \overline{f} where \overline{f'}(a_0) \not= 0, there ...<|control11|><|separator|>
  20. [20]
    Completion (Algebra) | Encyclopedia MDPI
    Dec 1, 2022 · In commutative algebra, the filtration on a commutative ring R by the powers of a proper ideal I determines the Krull topology (after Wolfgang ...<|control11|><|separator|>
  21. [21]
    How Google autocomplete predictions work - Google Search Help
    Autocomplete is a feature within Google Search that makes it faster to complete searches that you start to type. Our automated systems generate predictions ...<|control11|><|separator|>
  22. [22]
    [PDF] Eudora® Email 7.0 User Guide for Windows - Index of /
    History File—Select to make available the names from your history file to auto-complete in the To:, Cc:, and Bcc: fields in the message header. Turn off to ...
  23. [23]
    Google Suggest Inventor Kevin Gibbs on the Origin of Autocomplete
    Aug 23, 2013 · Google Suggest was built by Kevin Gibbs, a recent Stanford grad who joined Google just a couple of months before it went public.Missing: Eudora | Show results with:Eudora
  24. [24]
    About Us | Microsoft SwiftKey
    In 2016, SwiftKey released SwiftKey Symbols, a symbol-based assistive communication app targeted to non-verbal people with special needs.Missing: history | Show results with:history
  25. [25]
    Auto-complete feature using Trie - GeeksforGeeks
    Feb 20, 2023 · Search for the given query using the standard Trie search algorithm. · If the query prefix itself is not present, return -1 to indicate the same.
  26. [26]
    [PDF] Deep Learning Methods for Query Auto Completion - Microsoft
    • Utilizing spelling correction pairs, we train a. Markov n-gram transformation model that captures user spelling behavior in an unsupervised fashion ...
  27. [27]
    Neural Networks for Text Correction and Completion in Keyboard ...
    Sep 19, 2017 · This paper proposes a sequence-to-sequence neural attention network system for automatic text correction and completion.
  28. [28]
    Data Privacy Settings, Controls & Tools - Google Safety Center
    When you turn on Incognito mode in Maps and YouTube, your activity, like the places you search for or the videos you watch, won't be saved to your Google ...
  29. [29]
    How Chrome Incognito keeps your browsing private - Google Help
    Browsing in Incognito mode means your activity data isn't saved on your device, or to a Google Account you're not signed into.
  30. [30]
    Autocompletion for Search Enginees - Yuan Meng
    Jan 6, 2024 · Mean Reciprocal Rank (MRR): On average, how early does the first relevant completion appear? The formula is 1 P ∑ p = 1 P 1 r a n k p \frac ...
  31. [31]
    ML-Enhanced Code Completion Improves Developer Productivity
    Jul 26, 2022 · We discuss how ML and SEs can be combined by (1) re-ranking SE single token suggestions using ML, (2) applying single and multi-line completions ...
  32. [32]
    IntelliSense - Visual Studio Code
    IntelliSense is a general term for various code editing features including: code completion, parameter info, quick info, and member lists.IntelliSense for your... · IntelliSense features · Customizing IntelliSense
  33. [33]
    Eclipse 2.1 - New and Noteworthy
    Code assist in the Java editor now provides completion on keywords. Improved Java outline view. You can adjust the Java outline to show just the members of the ...<|control11|><|separator|>
  34. [34]
    [2211.04673] Syntax-Aware On-the-Fly Code Completion - arXiv
    Nov 9, 2022 · Abstract:Code completion aims to help improve developers' productivity by suggesting the next code tokens from a given context.
  35. [35]
    A Common Protocol for Languages - Visual Studio Code
    Jun 27, 2016 · An open, JSON-based protocol that anyone can use to add support for a new programming language to VS Code by implementing a language server.
  36. [36]
    GitHub Copilot code suggestions in your IDE
    Copilot code completion uses the GPT-4.1 Copilot model. The GPT-4.1 Copilot model was trained on a wide range of high quality public GitHub repositories, ...
  37. [37]
    Negation as Failure | SpringerLink
    The negation as failure rule only allows us to conclude negated facts that could be inferred from the axioms of the completed data base.
  38. [38]
    [PDF] Program Completion
    A logic program is a finite set of rules. The completion formula for an n-ary predicate constant P relative to a logic program Π is the sentence obtained as ...
  39. [39]
    [PDF] Completion of Disjunctive Logic Programs - IJCAI
    Clark's completion plays an important role in ASP computation: it discards unsupported models via unit resolution; hence, it improves the performance of ASP ...