oai-math-revised beta

Browse › Theoretical computer science › 128 A factor-two approximation for shortest common superstring

A Polynomial-Time 2-Approximation for Shortest Common Superstring

OpenAI · September 24, 2026

Lean

Open PDFCite Report a problem Add a reference or comment Interested in working on this Start a discussion

Edit NOTES.md · Existing issues · Source on GitHub

Abstract

We give a deterministic algorithm that, for every finite family of explicitly represented ordinary strings, outputs a common superstring of length at most twice the optimum in time polynomial in the total encoded input size, including symbol labels. The guarantee applies to the algorithm constructed here, not the classical maximum-overlap Greedy procedure.

If the PDF does not display here, open it directly: Open PDF.

Notes

No notes yet. Add a reference, a comment or your name with the buttons above, or edit NOTES.md on GitHub.

Automatically collected notes (unreviewed)

work in progress

Collected automatically by AI tools (Claude Opus 5.5 and GPT-6 Astra, 2026-10-21) and not checked by a person. May be wrong or incomplete. Report an error.

Follow-up works

  • rob-p (GitHub), practical-scs-2opt (GitHub repository with Typst note, Python reference and Rust implementation). Says the base graph needs only the at most 2L prefixes and suffixes rather than all O(L²) substrings, and benchmarks the algorithm against greedy and the lower bound W on repeat-rich and real data.
  • Krushna Dere and Svyatoblood (nurric-ai), lazy-scs: Scalable factor-2 shortest common superstring (GitHub repository with Rust/Python implementation and Lean 4 file, October 2026). Implements the algorithm with periodic rules applied lazily, reports runs on a 4.64 Mb E. coli genome, and formalizes the lower-bound argument in Lean, taking the paper's count facts as hypotheses; the README says some construction branches "never fired in any test". [AI-generated per source]
  • bkarak (GitHub), shortest-superstring-js (GitHub repository, JavaScript). Implements the paper's algorithm as twoApprox next to Greedy, MGREEDY, TGREEDY and an exact solver, and reports benchmarks including the family {c(ab)^k, (ba)^k, (ab)^k c}, where it reports greedy outputs of 4k+2 against the algorithm's 2k+6.
  • rongou (GitHub), superstring-rs (GitHub repository, Rust). Provides preprocessing, an exact solver, a greedy baseline and the occurrence-count lower bound W as groundwork for the paper's algorithm; the README states the approximation algorithm itself is not implemented yet.

Formalization

  • OpenAI, A factor-two approximation for shortest common superstring — Lean scope note (OpenAI Math Release, lean/docs/128.md, 2026-10-06). The release's scope note describes a Lean formalization of a deterministic polynomial-time factor-two algorithm for every finite list of explicitly encoded strings, with time measured in input bits and approximation in symbol length, and links the comparator statement ComparatorChallenges/Superstring.lean.
  • buberlo (GitHub), VeriSCS (GitHub repository, example run dated 2026-10-07). Packages the release's Lean function OAI.Superstring.answer (openai/math commit adc7f12) as an executable, reports axiom checks and running times, and leaves a human review of IsCommonSuperstring and opt against classical SCS as an open item (issue list).

Possibly relevant literature (found by the agent, not checked)

Digests and expositions

  • 梁植华, 最短公共超串的多项式时间 2-近似算法 (Eksmu, 2026-10-11). This Chinese-language guide outlines the manuscript's occurrence-count lower bound, balanced hierarchical graph, periodic layers and connection budgets, and distinguishes its algorithm from maximum-overlap Greedy.
  • Claude Opus 5.5 (Anthropic), Shortest common superstrings (Open Mathematics Courses, October 2026). Four lessons develop the manuscript's hierarchical graph, forced occurrence counts, periodic layers and connection argument with examples and exercises; the course identifies its checking as AI self-checking and says independent review is incomplete. [AI-generated per source]

Elsewhere

Independent sites; their content is not reviewed here.

Cite

@misc{OAI:A-Polynomial-Time-2-Approximation-for-Shortest-Common-Superstring-September-24-2026,
  author = {{OpenAI}},
  title = {{A Polynomial-Time 2-Approximation for Shortest Common Superstring}},
  howpublished = {OpenAI Math Release preprint
                  \href{https://github.com/openai/math/blob/main/preprints/A-Polynomial-Time-2-Approximation-for-Shortest-Common-Superstring-September-24-2026/paper.pdf}{OAI:A-Polynomial-Time-2-Approximation-for-Shortest-Common-Superstring-September-24-2026}},
  year = {2026}
}

Lean formalization

OpenAI provides Lean files for this family: lean/docs/128.md. The Lean statements have not been checked by people against the paper's claims.

Family 128: A factor-two approximation for shortest common superstring

Gives a deterministic polynomial-time algorithm constructing a common superstring of length at most twice the optimum for every finite family of explicitly represented strings. The running time is polynomial in the full encoded input length, including symbol labels.

This is the only paper in this family.