Literature
Display
← All papers

Lean Pool: An AI-Maintained Archive of Formalized Mathematics

Vasily Ilin

· arXiv (preprint) ↗

Contents

Overview

The author states that only the opening page was human-written; most of the remaining manuscript was produced with AI assistance. The author takes responsibility for its reported evidence and conclusions.

At a glance

Lean Pool collects completed mathematical formalizations and maintains them in a shared Lean/Mathlib environment. AI agents propose imports, dependency repairs, proof optimizations, and reviews; maintainers oversee acceptance. The September 21, 2026 census contains 211 projects and about 3.23 million source lines. Its operational history demonstrates maintenance and reuse, but does not independently establish the accuracy of its mathematical reviewers.

1 Lean Pool

A proof archive must remain usable as its language and libraries change. Lean checks formal proofs, while Lean Pool adds project admission rules, maintenance agents, and documentation around independently developed mathematics [57] [207]. Kernel acceptance checks the stated proposition under permitted axioms; it does not establish that the proposition faithfully expresses the advertised mathematical result.

Growth.

Agents discover eligible upstream projects, and people can submit their own. Pooling preserves attribution, updates dependencies, passes compilation and quality checks, and addresses costly compilation hotspots. Only completed formalizations with Apache-2.0 or MIT licenses qualify.

Maintenance.

Scheduled agents repair projects when Lean or Mathlib changes and propose ongoing reductions in proof length, compilation time, and memory use.

Documentation.

Readers can use the API index, the Exposition site, and attributed project announcements on Zulip.

Statistics.

The observed archive has 211 projects, 3,228,485 physical Lean lines, 18 commit contributors, and six dependency upgrades. These counts describe the paper’s dated snapshot.

Vision.

The proposed role is an arXiv-like home for maintained formalizations that accompany mathematical papers and can be imported into subsequent work.

Extended summary

The evidence combines source censuses, upgrade histories, accepted optimization changes, retained review reports, and an external reuse audit. Follow-up integration remained necessary after some automated repairs; reusable interfaces and shorter proofs did not always make compilation faster.

2 Motivation and scope

Preserving source files is insufficient for reuse: later researchers need compatible dependencies, discoverable statements, and visible assumptions. Lean Pool keeps independently attributed projects together so results can be inspected and imported in a common environment.

Admission rules.

Completed projects cannot contain sorry, admit, unchecked declarations, or extra axioms beyond Classical.choice, propext, and Quot.sound. Project cards record authors, source, proof provenance, and informal/formal main results. Open challenges are stored separately.

A formal counterpart to the mathematical literature.

The proposed relationship links a paper’s citations to imports between maintained formalizations. This paper evaluates existing maintenance operations and reuse evidence; universal adoption of that model remains a vision [15].

Relationship between papers on arXiv and maintained formalizations in Lean Pool.
Figure 1: The proposed relationship between arXiv and Lean Pool. New mathematics has both a paper and a maintained formalization. An author places the paper on arXiv and contributes its formalization to Lean Pool. Subsequent work cites the paper and imports the formalization, so formal dependencies can mirror the dependency graph of the mathematical literature. As formalization becomes easier and cheaper, we anticipate that most new mathematics papers will have accompanying formalizations. Lean Pool provides a home for these developments and maintains their connections as Lean and Mathlib evolve.

3 Contributing to and maintaining the archive

Contributions pass through attribution, automated checks, and maintainer decisions. The following subsections distinguish the workflow from the evidence about its past operation.

Community contributions.

A submitter need not be the original formalization author. Cards preserve upstream credit and distinguish human, AI, and mixed proof provenance. GitHub account counts do not measure how much work people versus agents performed.

Recurring work.

Daily jobs discover projects, inspect open PRs, address issues, optimize proofs, and announce accepted work. A separate dependency workflow tests new releases, assigns failures to repair agents, and assembles patches for review.

Continuous integration.

CI builds the whole archive and checks warnings, Mathlib linters, attribution, axioms, and resource rules. Lean Pool’s compilation profiling is advisory; the compared Tau Ceti workflow requires a performance check before merging [132] [205].

Evidence and scope.

Results use dated source snapshots and retained public records. Upgrade replays, production logs, accepted changes, and newly measured builds have different scopes; they are not a controlled estimate of overall agent productivity or operating cost.

4 The archive and evidence of reuse

The archive covers analysis, logic, algebra, geometry, probability, and computer science. Its 211 projects contain 837 registered main results; provenance labels identify 70 human, 102 AI, and 39 mixed projects. Source-line totals include comments and blank lines, and do not measure mathematical significance. The figures and tables retain the detailed census and growth history.

Table 1: Scale and participation at the source observation specified in Appendix A. Physical source lines include comments and blank lines. Declaration commands are counted in source, excluding examples and generated auxiliaries. Project-card provenance describes proof authorship. Contributor counts use GitHub user accounts and exclude bots; commit contributors and merged-PR authors are counted separately. Community PRs exclude the maintainer account. Open challenges are outside the source census.
Archive property Count
Completed projects 211
Lean source files 7,043
Physical source lines 3,228,485
Source declaration commands 193,862
Registered main results 837
Human / AI / mixed projects 70 / 102 / 39
Commit contributors 18
Contributors with merged PRs 17
Community PRs merged 63
Growth in completed projects and physical Lean source lines, with upgrades and optimizations marked.
Figure 2: Growth of the living archive. The curves show registered projects and physical Lean source lines, including comments and blank lines. Dotted markers indicate dependency upgrades. The project count continues to grow through periods of proof golfing; the annotated dips in source size mark library-wide compression and compilation optimizations. A separate project removal is labeled explicitly.

Reuse in research-level mathematics.

Lean Pool appears in more solutions than other repositories in the LeanEval declaration-matching comparison. Matching includes copied code and excludes ordinary package imports, so this is a specific measure of reuse, not a general ranking of library importance [15].

Build resources.

The clean-build comparison uses one shared Azure host with dependencies prepared before timing. The matching-release Mathlib comparison shares Lean Pool’s compiler and pinned Mathlib revision.

5 Maintaining compatibility as dependencies evolve

Six upgrades affected different snapshots of the archive. The stable 4.34.0 probe found compiler failures in 97 of 191 projects; 95 initial repair jobs succeeded and two failed. Additional integration produced a passing combined archive. Job success alone did not prove that an individual patch completed the migration, and some auxiliary hypotheses changed during repair.

Table 2: Projects encountering compiler failures under dependency upgrades. Earlier rows replay the original pre-upgrade source with the new environment. The final row uses the retained production probe before repairs. It measures the archive at that probe; subsequent imports and a separately accepted curation change altered membership before merge. The stable migration ultimately passed the combined archive build and repository checks. Warning-only cleanup is excluded from the failure count.
From Lean To Lean Projects at probe Compiler failures
4.30.0-rc2 4.31.0-rc1 59 44
4.31.0-rc1 4.32.0-rc1 91 58
4.32.0-rc1 4.33.0-rc1 143 100
4.33.0-rc1 4.33.0-rc2 145 19
4.33.0-rc2 4.34.0-rc1 148 3
4.34.0-rc1 4.34.0 191 97

6 Proof shortening and compilation speed

Shorter source does not reliably mean faster compilation. Five paired whole-library measurements ranged from a 5.4% slowdown after contributor proof golfing to a 5.8% speedup after elaboration-cost reduction. Each is one clean before/after pair at its own historical revision. Project-level measurements use different workloads and protocols; the reusable Navier–Stokes interfaces slightly increased project compilation time.

Table 3: Clean builds of the complete Lean Pool library before and after accepted PRs. Dependencies and build settings are fixed within each pair on the same Azure VM. Time is elapsed build time; RAM is sampled peak combined memory of the build workers. Parentheses give time saved, computed from unrounded measurements; negative values mean longer builds. Source savings include refactoring and declaration removal. Proof shortening does not uniformly reduce build time: contributor golfing uses less memory but takes longer in this run. Each row reports one clean before/after pair; Appendix E gives the protocol.
Accepted change Lines removed Build min ↓\downarrow before →\to after RAM GiB ↓\downarrow before →\to after
117 Proof-search simplification 511 5.30 →\to 5.25 (+1.1%) 15.6 →\to 15.6
187 Library-wide compression 45,217 16.11 →\to 15.59 (+3.3%) 26.1 →\to 26.9
246 Contributor proof golfing 12,515 21.83 →\to 23.01 (-5.4%) 25.9 →\to 24.6
338 Certificate simplification 513 30.22 →\to 29.13 (+3.6%) 21.0 →\to 20.9
339 Elaboration-cost reduction 54,965 28.46 →\to 26.82 (+5.8%) 19.8 →\to 19.6
Table 4: Accepted project-level optimizations and API work. The proof optimizations reduce project compilation time, while the Navier–Stokes refactor adds reusable interfaces at a small compilation cost. Positive source reductions mean fewer lines; the API import has no comparable source-reduction baseline. Timings come from the linked PRs, with dependencies prepared before each measured build. Workloads and machines differ across rows. Import cleanup has unmatched cache preparation, so only its source reduction is shown. Measurement summaries appear in Appendix E.
Accepted change Net lines removed Project seconds ↓\downarrow
Tactic-import cleanup 735 —
Restricted-sum proofs 65 20.01 →\rightarrow 18.45
Navier–Stokes reusable APIs — 879.67 →\rightarrow 887.78
Infinite Connes rigidity -31 118.00 →\rightarrow 93.70
Burkholder majorant proofs 1,064 23.84 →\rightarrow 19.35
Quantum parallel repetition 32 141.85 →\rightarrow 86.08
Interior-point LP proofs 249 13.98 →\rightarrow 10.69

7 Mathematical review in operation

Mathematical review asks whether checked statements match advertised claims and whether attribution, novelty, and code quality are adequate. Retained reports mostly approved submissions, but no independent labels establish review accuracy. Historical API estimates total $308.52 across 285 reports; six newer quota-based reports total $752.38 in nominal API-equivalent estimates, not cash expenditure. The service was later manually disabled.

Retained review verdicts and subsequent pull-request merge or closure outcomes.
Figure 3: Retained mathematical review-service reports. Left: all retained structured reports. Right: PR outcomes at the observation date, grouped by each PR’s latest retained verdict; labels give counts and bar lengths give proportions. Most reports approve their submissions. Requests for changes and discussion verdicts are each followed by both closures and subsequent merges. All reviewed PRs have reached one of these outcomes by the observation date. Findings concern statement mismatches, attribution, incomplete results, duplicate definitions, and unnecessary code. Greptile’s advisory reviews and free-form daily-agent comments are outside this census. The recorded outcomes are merge and closure decisions; review accuracy was not independently labeled.
Table 5: Recorded review price estimates, separated by billing regime. Historical comments report API-based estimates. The newer Azure service consumes Codex account quota and reports a Standard-API-equivalent price using uncached-input rates. The equivalent prices are not cash payments. Totals cover retained reports with recorded estimates and exclude missing or overwritten executions.
Price basis Reports Total Median
Historical API estimates 285 $308.52 $0.20
Codex API-equivalent estimates 6 $752.38 $89.03

8 Exposition and the structure of the library

Exposition connects project descriptions and headline results to Lean statements, source links, and declaration graphs [133]. The deployed export covers 203 projects, 173,362 source-visible declarations, and 1,389,967 dependency links; it predates the latest imports, explaining its smaller project count. Graph size records formal structure rather than mathematical importance. Readers must inspect assumptions and definitions before reusing a result.

Table 6: Coverage of the deployed Exposition export. Nodes are source-visible declarations and edges are intra-project dependencies, with generated auxiliaries followed internally. The scaffold is excluded. The deployment precedes the latest import batch. Its covered projects and declaration and dependency counts refer to the export revision specified in the observation table.
Exposition coverage Count
Documented projects 203
Source declarations 173,362
Theorems and lemmas 124,412
Dependency links 1,389,967
Declarations used by multiple others 87,408
Table 7: Examples of the mathematical structures exposed by Exposition. Counts are recomputed from the retained project graphs. Large developments expose reusable intermediate results as well as their headline theorems; graph size measures formal dependency structure, not mathematical importance.
Project Declarations Dependencies
Gödel’s incompleteness theorems 7,522 137,739
Classification of compact surfaces 7,833 79,052
Navier–Stokes and Euler 48,626 530,547
Language generation in the limit 419 1,823
Exposition project overview and declaration graph for the incompleteness theorems.
Figure 4: The Incompleteness project in Exposition. The project card identifies its formalization author and upstream source and lists the headline results. The graph places declarations within the supporting development, making its structure visible alongside the informal account. The headline results include Gödel’s incompleteness theorems, while the surrounding graph exposes the formal infrastructure on which they depend. Selecting a declaration links its position in the graph to its statement, source, and API documentation. This makes the project accessible both from its main mathematical claims and from the lemmas used to establish them. The screenshot and coverage tables use the deployment identified in Appendix A.

9 Finding and building on archived results

Use Octo for semantic queries, Exposition for browsing subjects and projects, and project cards for a known paper. Follow candidates into source and API documentation to check assumptions and import paths.

Table 8: Finding a result and following it into a development. A mathematical query, subject, or known paper provides an entry point. The reader can then inspect the candidate’s assumptions and definitions before importing it alongside other archived work in the shared environment.
Starting point Preferred tool What the reader obtains
A mathematical question Octo semantic search Candidate declarations across the archive
A subject or project Exposition Project descriptions, statements, and dependency graphs
A known paper or result Project card Attribution, source references, and formal headline results
A candidate to reuse Source and API documentation Full assumptions, supporting definitions, and import paths

10 Related work

Related systems address complementary tasks: maintaining shared libraries, discovering results, constructing proofs, and repairing or assessing formalizations.

Archives and shared libraries.

AFP combines reviewed submissions with maintenance; Mathlib integrates general-purpose mathematics; Reservoir indexes packages; Software Heritage preserves source [16] [207] [130] [3]. Lean Pool emphasizes independently attributed projects kept compatible together. Tau Ceti, Palomar, and Mathlib use different integration and review policies.

Table 9: Contribution and maintenance models in the Lean ecosystem, as documented by the projects [132; 206; 177; 131; 207]. Lean Pool maintains independent developments together. Tau Ceti and Mathlib integrate contributions into a shared mathematical API, with different roles for human and AI contributors. Palomar registers checked versions of separate repositories; updated versions require resubmission, verification, and review.
Project Contributions Maintenance Acceptance criteria Documentation
Lean Pool Human, AI, and mixed projects; direct submissions and attributed imports Agents propose upgrades and optimizations; maintainers oversee merges Completed proofs; permitted axioms; attributed cards; permissive license; CI and mathematical review Project cards, informal results, dependency graphs, source-linked API
Tau Ceti AI-authored mathematics following human roadmaps AI workers prioritize maintenance and review, update Mathlib pins, and repair the library CI and adversarial AI review; roadmap fit, reuse, API coherence, and compatibility with Mathlib Human roadmaps, module documentation, declaration docstrings
Palomar Registry Human, AI, or mixed formalizations submitted by an author or authorized maintainer Verified versions persist; authors resubmit updates Lean and NanoDa proof checks; automated review of statement faithfulness, disclosure, and research interest Recorded statements, provenance, dependencies, and public verification and review reports
Mathlib Human contributors; disclosed AI use with contributor understanding and justification Community maintenance across Lean releases Human review, generality, integration, maintainability, style, and CI Module and declaration documentation, API reference, library overviews

Discovery, formalization, and reuse.

TheoremSearch and TheoremGraph support discovery; LeanDojo and LeanAgent assist proof construction. Repository-level benchmarks and the LeanEval audit examine working within existing developments and reusing formalized mathematics [7] [126] [220] [121] [15].

Repair and mathematical review.

Proof-repair systems address compatibility, while statement-evaluation methods test correspondence between natural language and formal expressions. Expert audits underline why filling proof gaps is insufficient without faithful definitions and statements [216] [183] [109] [154].

11 Conclusion

The archive shows that agent-assisted maintenance can keep many independently developed formalizations compatible and reusable. Its operational evidence supports that practical role, while review reliability, ongoing coverage, and the costs of human oversight remain incompletely measured.

AI use statement

AI assisted evidence collection, analysis code, interpretation, literature discovery, figures, and writing. The author states that numerical results were computed from retained records rather than synthesized and accepts responsibility for the manuscript.

Reproducibility statement

The paper distinguishes newly executed measurements from historical reports and identifies source revisions and observation periods. Raw records and analysis programs are retained separately from the submission.

Acknowledgments

The acknowledgments credit Justin Asher for initial discovery tooling and CI, Austin Letson for proof golfing, and the archive’s contributors and upstream authors.

Appendix A Observation scope and measurement methods

Most archive and build observations are from September 21, 2026; Exposition is a September 20 export, and the last retained structured review is September 13. Source commands, exported declarations, contributor accounts, and review reports are counted separately. These distinct populations should not be combined as if they described one simultaneous snapshot.

Table 10: Observation periods and populations. The archive, its automation repository, and the exposition site continue to change; these dates specify the source observations and measurements used here.
Evidence Observation
Archive source census September 21, 2026, 06:07 UTC
Public PR records Retrieved September 21, 2026
Structured review-service reports Last retained report: September 13, 2026
Exposition export September 20, 2026, 14:24 UTC
LeanEval reuse analysis Analysis retrieved September 12, 2026
Automation job definitions Unchanged revision rechecked September 21, 2026
Build resources Clean builds measured September 21, 2026 (UTC)
Optimization comparison Historical revisions rebuilt September 13–14, 2026 (UTC)
Exposition screenshots Live interface captured September 21, 2026 (UTC)

Research records.

Retained inputs include source hashes, build logs, memory samples, review reports, and publication records. Validation regenerates displays without new compiler or model calls; model-generated assessments remain distinct from source attribution.

A.1 Composition of the collection

The largest projects vary widely in size and provenance. Two projects were separately retired for size-policy and textbook-scope reasons; those removals were curation decisions, not failed compiler repairs.

Table 11: Largest archived projects by physical source size. Counts include comments, blank lines, and integration changes within each project. Provenance comes from the attributed project card. The examples span analysis, geometry, logic, number theory, and computer science; source size describes the formal development rather than the mathematical significance of its headline result. The full catalogue links every project to its upstream source and publications.
Project Provenance Lean lines
Finite-time breakdown for Navier–Stokes and Euler AI 641,073
A complex structure on the six-sphere AI 260,646
Classification of Compact Surfaces AI 143,423
Polynomial-factor hardness of the closest vector problem mix 135,818
Improved asymptotic bounds for binary and spherical codes mix 112,078
The Convex-Octagon Case of Erdős Problem 97 AI 101,724
Jordan–Schönflies theorem AI 77,482
Quantum parallel repetition mix 71,975
Modular forms and the generalized residue theorem mix 67,709
Ehrhart’s sharp volume inequality mix 56,357
Sharp asymptotic upper bounds for sphere packing mix 55,713
Gödel’s First and Second Incompleteness Theorems human 55,430

A.2 Build-resource comparison

The full Lean Pool build takes 60.28 minutes and peaks at 20.03 GiB of proportional process memory, versus 37.44 minutes and 7.35 GiB for matching-release Mathlib. Each row is a single build on a shared host, with dependency preparation, documentation, and separate CI audits outside timing. Different source sizes and cache roles limit direct efficiency comparisons; background activity was recorded separately.

Table 12: Clean builds of the named libraries on the same Azure VM. Lean Pool has more source lines and a longer elapsed build than the matching Mathlib release in these runs. Lean Pool retains its fetched Mathlib cache; Mathlib rebuilds its own sources with its external dependencies prepared. The matching-release comparison uses the same compiler and Mathlib revision. Current Mathlib and Tau Ceti use their supported compiler versions, listed in Appendix A.2. Source size excludes dependency code. CPU time sums user and system time; RAM is sampled peak combined proportional resident memory, accounting for shared pages. Each row is one complete build on a shared VM.
Library Source MLOC Build min ↓\downarrow CPU h ↓\downarrow RAM GiB ↓\downarrow
Lean Pool 3.23 60.28 14.29 20.03
Mathlib (matching release) 2.33 37.44 9.38 7.35
Mathlib (current) 2.33 35.83 8.99 7.00
Tau Ceti 1.46 30.89 7.13 10.67
Table 13: Shared measurement configuration. Builds run sequentially with the same CPU affinity and thread settings. The observer verifies that every source module compiles, its output exists, and the source checkout remains unchanged. Memory peaks are sampled between process scans. Cache downloads, documentation, and separate CI audits are outside the timed command. Normal in-build lint options remain enabled. The VM ran background tasks during these measurements; their activity is recorded separately and is outside the reported CPU and process-memory totals.
Setting Value
Host Azure Linux VM; AMD EPYC 9V45 processor
Available CPUs / Lean threads 16 / 16
Physical RAM 125.8 GiB
Dependency preparation Prebuilt; outside timing
Root-library outputs Absent before each build
Filesystem cache Compilation inputs preloaded before timing
Artifact restoration Disabled for the measured root library
Run order Lean Pool, matching Mathlib, current Mathlib, Tau Ceti
Time instrumentation GNU time; elapsed and aggregate user + system CPU
Memory instrumentation Summed proportional resident memory of the build process group
Table 14: Measured source revisions and compiler versions. Links identify the exact source for each completed build. The matching Mathlib release is the dependency pinned by Lean Pool. The current Mathlib observation uses a newer compiler; Tau Ceti uses its own supported release candidate.
Library Lean Source revision
Lean Pool 4.34.0 bc6f18b2
Mathlib (matching release) 4.34.0 5ed29652
Mathlib (current) 4.35.0-rc2 0a6c8e03
Tau Ceti 4.34.0-rc2 8aa2ad02

Appendix B CI and profiling

Checks combine full-library compilation, warning rejection, source and environment linters, generated-index validation, and archive admission rules. Sharded compilation is followed by a combined build. The comparison table distinguishes shared checks from each repository’s governance and performance requirements.

Table 15: Mechanical checks and performance tooling in the observed repositories, supported by the retained workflow definitions and benchmark documentation [132; 151; 205]. All use library compilation and Mathlib linters. Lean Pool adds project-level admission checks, while Tau Ceti requires a performance comparison for merging; Pool profiling is advisory. Mathlib maintains whole-build benchmark and telemetry infrastructure.
Responsibility Lean Pool Mathlib Tau Ceti
Compilation Combined archive build; warning rejection Library build, tests, warning checks Complete source-tree build; warning and output checks
Declaration lint Mathlib environment linters Mathlib lint driver Mathlib environment linters with a recorded baseline
Source style Mathlib style checks; no waivers Source and bibliography style checks Style and header checks; bounded files
Logical assumptions Allowed-axiom check and compiled-environment audit Standard trusted Lean foundations; library review and linting Allowed-axiom audit of compiled sources
Project governance Attributed cards, completed results, permissive license, separated content PRs Review of shared mathematical API AI-contributed mathematics under trusted governance checks
Resource policy No content option overrides; source and proof size limits Library-configured options and reviewed exceptions No content option overrides; per-process watchdog
Performance Advisory per-file profiling Build benchmark and telemetry tooling Required comparison using hardware instructions or CPU time; advisory heartbeats

Archive-specific rules.

Cards must resolve registered results and retain provenance, licensing, subject tags, and corresponding informal statements. Completed projects cannot bypass proof, axiom, linter, or resource checks. The separate challenge board permits open statements; proposed solutions undergo an independent kernel comparison against the registered statement.

Profiling.

Per-file profiling reports source changes, declarations, elapsed elaboration, and Lean heartbeats. Heartbeats are allocation-based computation counts, not substitutes for whole-build timings. Pool’s reports are advisory; Tau Ceti requires a base/head performance comparison.

Appendix C Daily jobs and contributor participation

The automation runner refreshes source, prevents duplicate instances of a job, and can use a fallback backend after failure. Job definitions describe intended responsibilities, not a census of which model performed historical work. Dependency repairs and optimization proposals still require checks and maintainer decisions.

Table 16: Daily responsibilities in the maintained automation repository. Discovery jobs prepare attributed imports; review and issue jobs act on existing submissions; optimization jobs improve archived code; announcements connect accepted results to the community. These are job responsibilities, not a census of completed executions. Dependency bumping runs in a separate scheduled repository workflow.
Job Work performed
Recent discovery Find recently completed formalizations; check attribution, license, fit, dependencies, and cost; prepare an import PR.
Older discovery Search the preceding literature and repositories for completed projects missed by recent discovery; apply the same admission checks.
Open-PR review Inspect current submissions, mathematical references, CI, reviews, and profiling; repair eligible branches or record unresolved issues.
Project optimization Profile an existing project, shorten proofs or reduce compilation cost, validate the change, and open a PR.
Maintainer issues Inspect submissions awaiting a maintainer decision; resolve eligible source or metadata problems and retain unresolved decisions.
Announcements Identify newly accepted projects, check their advertised results, and announce them with attribution while suppressing duplicate posts.

C.1 What other contributors have added

The table attributes merged community PRs covering mathematics, corrections, and tooling. Submitter counts remain separate from the original authorship recorded in project cards.

Table 17: Contributions from community submitters through the current observation. Counts cover merged PRs; descriptions summarize the mathematics and tooling in their accepted changes. Project cards separately preserve upstream formalization authorship.
Contributor Merged PRs Contributions
lyfar 13 Combinatorics, geometry, and project corrections
jtraverso 7 Convex duality, graph decompositions, matching, flows, and concentration
Lemmy00 2 Pythagorean parametrization; Cramer–Wold
ldct 2 Commuting probability; Chudnovsky formula
AlexeyMilovanov 1 Harper’s isoperimetric theorem
ElNando888 1 Krafft sieve
Hydrodynamical 1 Vlasov mean-field limit
JonCYeh 1 Pentagonal number theorem and analytic Jacobi triple product
dillon-11 1 Lehmer polynomial and Coxeter element
gersh 1 Three-body nonintegrability
jarfo 1 Unique multiset sums
mccorvie 1 Classification of surfaces
rkirov 1 Jacobian of a Riemann surface
vaguiarl 1 Odlyzko bound solution

Appendix D Stable-version migration

The stable migration needed further repairs after two initial jobs failed. Accepted integration changed 639 files with 5,951 added or removed lines; final membership also reflected unrelated imports and retirements. Adjustments included stronger measurability, sigma-finiteness, and maximality hypotheses. A successful build verifies the accepted statements, not equivalence with every pre-upgrade declaration. Churn is an effort proxy; historical charges and human labor were not logged.

Table 18: Production record of the stable-version migration. The initial agent jobs produced repair patches; additional agent-assisted integration addressed remaining build errors, warning changes, and API compatibility. Accepted membership reflects intervening imports and the separately approved project retirements. Source churn compares the merge with its first parent and therefore excludes those unrelated changes. Job success is the workflow status, not a guarantee that its patch completes the combined migration.
Recorded stage Count
Probe projects 191
Compiler failures 97
Warning only projects 32
Repair jobs 97
Repair jobs succeeded 95
Repair jobs failed 2
Accepted projects 198
Changed files 639
Line churn 5,951
Table 19: Source changes in the accepted dependency upgrades. Churn is added plus removed physical Lean lines. It includes cleanup in projects that already compiled and therefore describes the work that landed rather than the minimum repair. These records provide an observable proxy for maintenance effort; historical monetary charges and human labor were not logged.
Target Lean Files edited Lines changed
4.31.0-rc1 825 25,583
4.32.0-rc1 402 5,653
4.33.0-rc1 576 11,690
4.33.0-rc2 585 5,221
4.34.0-rc1 448 4,676
4.34.0 639 5,951

Appendix E Accepted optimization changes

Whole-library optimization pairs fix dependencies, configuration, CPU affinity, and threads within each pair, remove root outputs, and preload dependency inputs outside timing. Memory is sampled proportionally across the process group. Project-level results include medians, single runs, and unmatched cache preparation, so they are reported separately instead of pooled into one speedup.

Table 20: Complete-library builds underlying the optimization comparison. Each row reports one clean run, with the parent before the accepted change. The VM uses the same CPU affinity and Lean thread setting as the library comparison. Dependencies are prebuilt and their compilation inputs preloaded outside timing. CPU hours sum user and system time across the build; RAM is combined proportional resident memory sampled throughout the run. Paired measurements use each PR’s own source and toolchain, so comparisons are within PRs rather than between different archive sizes.
PR Revision Lean Wall min ↓\downarrow CPU hours ↓\downarrow RAM GiB ↓\downarrow
117 Before 4.30.0-rc2 5.30 0.96 15.60
After 4.30.0-rc2 5.25 0.94 15.57
187 Before 4.32.0-rc1 16.11 4.04 26.12
After 4.32.0-rc1 15.59 3.90 26.85
246 Before 4.32.0-rc1 21.83 5.51 25.95
After 4.32.0-rc1 23.01 5.81 24.58
338 Before 4.33.0-rc2 30.22 7.66 21.02
After 4.33.0-rc2 29.13 7.38 20.89
339 Before 4.34.0-rc1 28.46 7.20 19.81
After 4.34.0-rc1 26.82 6.78 19.55

Library-wide module and proof optimization.

A separate historical module/proof benchmark reduced retired instructions from 124.499 to 103.172 trillion and sampled anonymous memory from 13.222 to 5.594 GiB. It used a fixed workload on a shared WSL host before later imports. Its individual-process and anonymous-memory metrics differ from the newer Azure proportional-memory measurements.

Table 21: Fixed-workload benchmark accompanying the accepted module and proof optimization. The merged PR compares the same original archive workload with pinned dependencies and cold project caches on a shared WSL host. Its measured candidate precedes later imports and final integration. Retired instructions decrease alongside memory use in this historical comparison. Process RSS is the largest individual process; anonymous memory sums sampled concurrent private allocations and excludes file-backed mappings. These differ from the proportional-memory metric in the fresh Azure comparison. The current source census and Azure build include the merged optimization.
Workload Instructions T ↓\downarrow Process GiB ↓\downarrow Anonymous GiB ↓\downarrow
Original source 124.499 12.605 13.222
Optimized source 103.172 2.891 5.594

Appendix F Production reviews and agreement

Repeated reviews agree on 37 of 69 same-PR pairs, but code and models can change between reports. No pair has an explicitly matched reviewed revision. These counts therefore measure observed verdict consistency rather than mathematical correctness. Fourteen reports explicitly identify partial coverage.

Table 22: Agreement between retained reports of the same review type. Same-PR comparisons can span changed code and review models. Same-revision comparisons require an explicit matching reviewed commit. Report pairs are dependent when a PR has repeated reviews; agreement describes verdict consistency rather than mathematical correctness.
Comparison Pairs Same verdict
Same PR 69 37
Same reviewed revision 0 0
Table 23: Composition of the retained review-service reports. Project, challenge, solution, and refactor reviews follow distinct rubrics. Explicit partial-coverage notices are counted separately and may occur in any category.
Report category Reports
Project 278
Challenge 7
Solution 2
Refactor 9
Explicitly partial reports 14

Service evolution and state.

The later Azure service split large changes for review and integrated findings with a source-coverage manifest. Coverage records what the model saw, not its accuracy. The workflow was manually disabled before the current census, so retained reports do not establish continuing review of recent imports.

Price accounting.

Historical API estimates and newer Standard-API-equivalent estimates use different accounting regimes. The newer jobs consume account quota, value input at uncached rates, and omit VM costs. Missing or overwritten executions are unrecoverable, so report totals are neither invoices nor complete operating costs.

Appendix G Exposition, discovery, and reuse

The preserved panel shows how a headline result opens its informal description, formal Lean statement, dependency information, and source links.

Exposition statement panel for Gödel’s second incompleteness theorem.
Figure 5: The theorem panel for Gödel’s second incompleteness theorem in Exposition. Selecting a headline result opens its informal description, Lean statement preview, and links to source and API documentation. This close-up shows how the site connects a result in the overview graph to its formal account. Capture dates appear in Table 10.

Maintaining the publication pipeline.

Graph caches depend on project source, transitive imports, toolchain configuration, and extractor version. Documentation reuses CI artifacts only when source revisions and verified trees match; cached exports still undergo data validation.

G.1 Recorded reuse

Reuse records distinguish direct theorem application, attributed copying, explicit package dependencies, and later maintainer-controlled projects. LeanEval matches normalized declarations in 20 solutions to Lean Pool, compared with 8 for the next repository shown. Matches can include both an archived project and its upstream source; ordinary package dependencies are excluded [15].

Table 24: Recorded reuse within and beyond the archive. The Kurosh development applies the existing finite-graph formalization’s spanning-tree basis in its Schreier index formula. Other entries distinguish attributed source copying, an explicit package dependency, and use within later projects controlled by the maintainer. These source histories complement the LeanEval declaration-matching analysis.
Consumer Recorded reuse
Kurosh subgroup development Applies the archived finite-graph spanning-tree basis
Feige development Copies attributed Pool isoperimetry; applies its theorem
Poincaré development Pins Pool and reexports its nonintegrability theorem
Maintainer homotopy development Contains attributed homotopy modules imported through Pool
Maintainer Mazur development Uses the archived Grothendieck-vanishing result
LeanEval solutions with matching declarations from Lean Pool and comparison repositories.
Figure 6: Reuse of public repositories in the LeanEval audit [15]. Lean Pool has matching declarations in more solutions than any other repository in the comparison. Counts are solutions with matching normalized source declarations, including copied code; ordinary package imports such as Mathlib dependencies are excluded. A solution can match both an archived project and its upstream repository. The audit’s agent traces document how archived mathematics entered subsequent research-level formalizations. Further source-level reuse appears in Appendix G.1.

Appendix H Original upstream Lean versions

The complete 211-project table records upstream Lean versions and links the supporting source. A dagger identifies the latest pre-registration default-branch revision used when an import pin was unavailable; “Unpinned” means the retained evidence supplied no toolchain pin.

Table 25: Original upstream Lean environments for every current project, ordered as in the publication concordance. Unmarked entries use an import pin or explicit original import report. A dagger marks the latest upstream default-branch revision before registration, used when the import pin is unavailable. Unpinned means the retained source or original import report records no upstream toolchain pin. Links lead to the source evidence.
No. Project Upstream Lean
1 2-Coloring Cycles in One Round 4.28.0†
2 A 4AP-free permutation of the positive integers 4.33.1
3 A complex structure on the six-sphere 4.33.0
4 A Conditional Fourteen-Point Case of Erdős Problem 132 4.32.0-rc1†
5 A Conditional Sieve Criterion for Twin Primes via Krafft Geometry 4.32.0-rc1†
6 A Detailed Proof of the Chudnovsky Formula 4.32.0-rc1†
7 A formalization of Borel determinacy in Lean 4.28.0-rc1†
8 A sharp 5/8 bound for Erdős Problem 865 4.28.0
9 ABC implies that Ramanujan’s tau function misses almost all primes 4.26.0†
10 Accelerated Nesterov convergence under local Polyak-Lojasiewicz conditions 4.28.0†
11 Almost all primes are partially regular 4.26.0†
12 Analysis of Boolean functions in Lean 4.16.0-rc2†
13 Apportionmentlib 4.30.0†
14 Archon-FirstProof-Results 4.28.0†
15 Artin-Wedderburn Theorem 4.14.0-rc2†
16 Asymptotically attaining the Moore bound 4.32.2
17 Aumann’s Agreement Theorem 4.28.0†
18 Axiomatic projective geometry (Faure–Frölicher) Unpinned
19 Bennett–Bernstein, Freedman, and Hoeffding concentration inequalities 4.28.0
20 BKLO simultaneous perfect matchings and spread matchings 4.28.0
21 Boolean Isoperimetry and Conway–Guy Coherent Gaps 4.28.0†
22 Bounds for the centered Hardy-Littlewood maximal constant 4.32.0
23 Brauer Group Core 4.26.0-rc2†
24 Burkholder Martingale Transform Inequality 4.30.0-rc2
25 Carlet’s Kasami cyclic-additive conjecture 4.33.0
26 Certified Runge-Kutta Order Conditions 4.30.0-rc2
27 Chaos and the Julia set of the complex exponential 4.34.0
28 Chebyshev Quotients and Demazure Multiplicities 4.28.0
29 Chordal graphs: minimal separators and simplicial vertices 4.34.0-rc1†
30 Chvátal’s conjecture and sharp Boolean correlation 4.33.1
31 Circuit Complexity in Lean 4 4.29.0-rc4†
32 circuitlib: a circuit verification library for Lean 4 4.30.0-rc1
33 Classification of Compact Surfaces 4.32.2†
34 Classification of groups of order p * q 4.15.0†
35 Classification of low-dimensional solvable Lie algebras 4.19.0†
36 Clawristotle: Vlasov-Maxwell-Landau steady-state classification 4.24.0
37 Coinductive Interaction Trees using QPFs 4.29.0†
38 Common-neighbour counterexamples for Saxl graphs 4.34.0-rc2
39 Conditional formalization of Zhou’s Connes-rigidity counterexample 4.34.0-rc2
40 Connes-Kreimer Hopf algebra of rooted trees 4.30.0-rc2†
41 Convex Three-Distance Degree-Six Theorem and Exceptional-Word Closures 4.34.0-rc1†
42 Counterexamples to graph compactness and two-degenerate extremal bounds 4.32.0
43 Counting Critical Portraits 4.31.0†
44 Craig interpolation and Beth definability for propositional dynamic logic 4.33.1
45 Craig Interpolation for Gödel-Löb logic via coalgebraic proofs 4.28.0†
46 Dead Ends in Square-Free Digit Walks 4.26.0
47 Demazure Operators and Lean 4.14.0-rc3†
48 Density-One GKP Divisibility and Its Carry-Language Characterization 4.32.0-rc1†
49 Dilatations of categories and commutative rings 4.24.0
50 Directed Topology in Lean 4 4.6.0-rc1
51 Disproof of the Aharoni-Korman Conjecture 4.16.0-rc2†
52 Disproof of the Köthe conjecture (Krempa’s matrix form) 4.28.0
53 DomainTheory 4.30.0
54 Duality theory in linear optimization and its extensions 4.18.0†
55 Ehrhart’s sharp volume inequality 4.32.0
56 Erdős Problem #137: powerful products of consecutive integers 4.28.0
57 Erdős Problem #367 4.28.0
58 Erdős Problem 346: Ratio Limit Forces the Golden Ratio 4.31.0†
59 Euclidean Distance Geometry 4.32.0-rc1†
60 Euler’s pentagonal number theorem 4.26.0-rc2
61 Euler’s pentagonal number theorem, two independent proofs, and the Jacobi triple product 4.34.0-rc1
62 Even graphs are edge-disjoint unions of cycles 4.28.0
63 Event Structures and Causal-Consistent Reversibility 4.28.1†
64 Exact odd-prime distributions of central-binomial valuations 4.32.0-rc1†
65 Exceptional Set in the abc Conjecture 4.21.0-rc3†
66 Existence of a finitely presented non-sofic group 4.32.0
67 Existence of Nash equilibria via Brouwer’s fixed-point theorem 4.30.0†
68 Explicit type A_​n\_n and BC_​n\_n root pairings 4.30.0-rc2†
69 Extended Demazure Product on ASP Permutations 4.31.0-rc2
70 Factorization Systems 4.14.0-rc2†
71 Feige’s sharp unit-slack inequality 4.31.0
72 Fel’s Conjecture on Syzygies of Numerical Semigroups 4.26.0
73 Fermat’s Last Theorem for regular primes 4.34.0-rc2
74 FinEqs - reducing equations defining a subset of n-space over a finite field 4.24.0†
75 Finite max-flow / min-cut 4.28.0
76 Finite witnesses and the width hierarchy for language generation 4.24.0
77 Finite Čencov-Petz Uniqueness 4.29.1
78 Finite-N Kuramoto Synchronization 4.31.0-rc1†
79 Finite-time breakdown for Navier–Stokes and Euler 4.34.0-rc2
80 Finitely generated cones and finite LP duality 4.28.0†
81 First Order Language of ZF Set Theory 4.22.0-rc3†
82 Flean: Floating-Point Numbers in Lean 4.27.0-rc1†
83 Formal Learning Theory Kernel 4.29.0-rc6†
84 Formalisation of the Bruhat-Tits Tree 4.19.0†
85 Formalizations of theorems related to model checking 4.30.0†
86 Formalized Complex Analysis in Lean 4.28.0-rc1†
87 Foundational local complex-analytic geometry 4.32.0
88 FrontierMath Ramsey Hypergraphs 4.28.0
89 Fundamental groups of finite connected graphs 4.32.0
90 Galois groups of quadratic polynomial iterates 4.33.1
91 Gaussian Moments Conjecture counterexamples 4.34.0
92 Genus-Zero Pairings and Catalan Numbers 4.29.0-rc6†
93 Geometric Four-Colorings of the Moser Lattice and Ring 4.32.0-rc1†
94 Global completeness of one-sorted definedness-free matching logic 4.33.0
95 Grothendieck’s Vanishing Theorem 4.28.0
96 Gödel’s First and Second Incompleteness Theorems 4.16.0-rc2
97 Halving the Kalton-Roberts upper bound 4.28.0†
98 Homogeneous self-dual interior-point method for linear programming 4.31.0-rc1†
99 Hurwitz’s Classification of Euclidean Composition Algebras 4.33.0
100 Improved asymptotic bounds for binary and spherical codes 4.32.0
101 Improved long gaps between consecutive primes 4.33.0
102 Improved Lower Bounds for Strong n-Conjectures 4.30.0-rc2†
103 Infinitary logic and countable model theory 4.34.0-rc1
104 Infinite counterexamples to Connes’s rigidity conjecture 4.32.0
105 Irrationality of ζ\zeta(3) 4.18.0†
106 Jordan–Schönflies theorem 4.32.2
107 Kahn–Kalai expectation-threshold theorem 4.32.0
108 Known Bounds for the Hadwiger-Nelson Problem 4.32.0-rc1†
109 Kolokolnikov’s ACMAX conjecture 4.32.0-rc1
110 Kurosh and Schreier subgroup theorems via covering graphs 4.32.0
111 Lean-QuantumAlg 4.30.0
112 Lehmer’s Polynomial and the E10 Coxeter Element 4.32.0-rc1†
113 Lentil: Temporal Logic of Actions in Lean 4 4.28.0†
114 Leo Moser’s finite distinct-subset-sums inequality 4.32.0-rc1†
115 Maxima of Coxeter frieze patterns are Fibonacci numbers 4.10.0-rc2†
116 Mean-field derivation and well-posedness of the Vlasov equation 4.33.0-rc1
117 Minimum modulus for the unique multiset-sum problem 4.32.0†
118 Misere combinatorial games 4.29.0-rc1
119 Modular forms and the generalized residue theorem 4.29.0-rc8
120 Monlib4 Operator-Algebra and Quantum-Set Core 4.21.0-rc3†
121 Monotonicity Formula for Stationary Harmonic Maps 4.30.0-rc2
122 Monsky’s Theorem 4.16.0-rc2†
123 Moreira’s version of Sard’s theorem 4.27.0-rc1
124 MRiscX 4.29.0†
125 Nash-Williams fronts and 2-better-quasi-orders 4.28.0
126 Near-perfect triangle packings from sum-zero triples 4.28.0†
127 Neukirch’s Algebraic Number Theory: Hilbert ramification theory 4.5.0-rc1†
128 Nivat’s conjecture 4.33.1
129 On the paucity of lattice triangles 4.26.0†
130 Optimal Pebbling Number of the Hypercube 4.30.0-rc2†
131 Oracle Computability and Turing Degrees 4.24.0†
132 Osterwalder-Schrader Axioms for the Gaussian Free Field 4.29.0†
133 Partial Combinatory Algebras 4.15.0-rc1†
134 PCF Theory 4.18.0-rc1†
135 Period Lengths of Rational Cut-and-Project Gap Sequences 4.29.1
136 Planar Point Sets Whose Non-Diameter Distances Form a Geometric 3-Chain 4.31.0-rc1†
137 Poincaré’s Nonintegrability Theorem for the Restricted Three-Body Problem 4.33.0-rc1†
138 Pointwise Birkhoff Ergodic Theorem 4.20.0-rc5†
139 Poitou’s Explicit Odlyzko Bound for Root Discriminants 4.33.0-rc1†
140 Polylean Unit Conjecture Counterexample 4.19.0-rc3†
141 Polynomial ABC (Mason–Stothers) and its corollaries 4.9.0-rc2
142 Polynomial parametrizations of Pythagorean triples 4.30.0-rc2†
143 Polynomial-factor hardness of the closest vector problem 4.32.0
144 Prekopa-Leindler, Brunn-Minkowski, and the isoperimetric inequality 4.26.0-rc2†
145 Primitive Sets Above x (Erdos Problem 1196) 4.30.0-rc1†
146 Pumping Lemma for Context-Free Grammars 4.15.0-rc1†
147 Pólya’s enumeration theorem 4.14.0-rc2
148 Quadratic-order prime classification 4.28.0†
149 Quantum parallel repetition 4.32.0
150 Quartic-over-logarithmic lower bound for rational permanent formulas 4.32.0
151 Quasi-Borel Spaces 4.28.0-rc1†
152 Radó’s theorem for Riemann surfaces 4.33.0
153 Rellich–Kondrachov compact embedding theorem 4.29.1†
154 Riemann Mapping Theorem 4.26.0
155 Riemann–Roch for algebraic function fields 4.31.0
156 RL Theory in Lean 4.28.0-rc1†
157 Sabidussi’s compatibility conjecture for Eulerian multigraphs 4.31.0
158 Selberg Sieve 4.7.0-rc2†
159 Sensitivity Conjecture: sqrt(deg) <= sensitivity 4.28.0†
160 Serre’s mass formula for totally ramified local-field extensions 4.32.2
161 Shannon Entropy Characterization 4.28.0†
162 Sharp asymptotic upper bounds for sphere packing 4.32.0
163 Sharp Five-Distance and Sup-Norm Gap Theorems 4.30.0†
164 Simple zeros of the Riemann zeta function 4.34.0-rc2
165 Spectral positivity 4.30.0
166 Spectral theorem for compact self-adjoint operators 4.29.1
167 Square-difference-free polynomials beyond Naslund’s conjectured bound 4.33.0
168 Stable phase retrieval for Hermite-Fock expansions 4.29.0-rc6†
169 Strengthened V0 Bounded Arithmetic Interfaces 4.22.0†
170 Sums of Distinct Factorials That Are Powers of Two (Erdos Problem 403) 4.31.0†
171 Sums of Three Squares 4.29.1†
172 Sundog certificates 4.30.0
173 Superexponential lower bounds for multicolor triangle Ramsey numbers 4.32.0
174 Synthetic Euclidean geometry: Euclid’s Elements Book I Unpinned†
175 Thakur’s hypotheses on power sums of F_q[t] 4.28.0†
176 The 5/8 theorem 4.28.0†
177 The Anderson Conjecture: A Weakly Quasi-Complete Ring Need Not Be Quasi-Complete 4.29.0-rc8†
178 The Bannai-Bannai-Stanton Bound on Distance Sets 4.31.0†
179 The Bollobás–Nikiforov conjecture 4.33.0-rc1
180 The Convex-Octagon Case of Erdős Problem 97 4.32.0-rc1†
181 The Cramer-Wold theorem 4.32.0-rc1†
182 The Erdős–Graham–Ruzsa–Straus two-prime theorem 4.29.1†
183 The Erdős–Sós conjecture (Erdős Problem 548) 4.28.0
184 The Erdős–Tuza–Valtr conjecture 4.13.0-rc3
185 The Fundamental Inequality of Valued Fields 4.30.0†
186 The Hanson-Wright inequality for sub-Gaussian quadratic forms 4.33.0
187 The Jacobian of a Compact Riemann Surface 4.33.0-rc2
188 The Johnson–Lindenstrauss Lemma and Quantized JL 4.31.0†
189 The Komlós and Beck–Fiala bounds with constant 36 4.34.0
190 The Kunen inconsistency theorem 4.29.0-rc4†
191 The main theorem of polytopes 4.7.0-rc2
192 The optimal exponent relating sumsets and difference sets 4.32.1
193 The polynomial Freiman–Ruzsa conjecture 4.34.0-rc2
194 The polynomial method and restricted sums of congruence classes 4.27.0-rc1
195 The Ramanujan-Nagell theorem 4.26.0-rc2
196 The Rupert Problem for convex polyhedra 4.28.0
197 The Three-Gap (Steinhaus) Theorem 4.29.1
198 The transcendence of π\pi 4.30.0-rc2†
199 The Wallace problem in ZFC 4.30.0
200 The Zhang-Yeung non-Shannon information inequality 4.28.0-rc1
201 Turán’s theorem (the "Book" weighting proof) 4.24.0-rc1†
202 Ulm’s theorem for countable reduced abelian p-groups 4.30.0-rc1
203 Unconditional Schauder Bases 4.30.0-rc2
204 Uniqueness of Shannon Capacity-Achieving Priors 4.29.1†
205 Verified canonical labelling of finite graphs 4.33.1
206 Verified interval-Cauchy real arithmetic 4.17.0-rc1†
207 Virasoro Project 4.27.0-rc1†
208 Weierstrass models and singular points for Tate’s algorithm nightly 2023-08-19 †
209 Whitehead’s theorem for CW-complexes 4.21.0-rc3
210 Wigner Semicircle Distribution 4.24.0†
211 ZFLean 4.27.0†

Appendix I Imported projects and their publications

The complete 211-project concordance links each imported project to upstream source and identified publications. It distinguishes companion papers, mathematical source material, software releases, and records without an identified publication. These are provenance records, not independent endorsements of every advertised mathematical claim.

Table 26: Complete concordance of the current archive and its publications. Each registered project has a row and a link to its upstream repository. Companion papers and manuscripts describe the formalized results or development; mathematical sources provide underlying results. Source records distinguish papers, books, manuscripts, and repository documentation. Projects removed before the observation are excluded.
No. Imported project / upstream Publications and mathematical sources
1 2-Coloring Cycles in One Round Companion paper: 2-Coloring Cycles in One Round [68].
2 A 4AP-free permutation of the positive integers Companion paper: A 4AP-free permutation of the positive integers [97].
3 A complex structure on the six-sphere Mathematical manuscript: A compact complex threefold fibred by tori over the projective line, and the six-sphere [10].
4 A Conditional Fourteen-Point Case of Erdős Problem 132 Mathematical source: Constructions of Maximum Few-Distance Sets in Euclidean Spaces [204].
5 A Conditional Sieve Criterion for Twin Primes via Krafft Geometry Companion preprint: Variational Optimisation of Spectral Sieve Quotients for Primes in Bounded Symmetric Intervals [185].
6 A Detailed Proof of the Chudnovsky Formula Mathematical source: A detailed proof of the Chudnovsky formula with means of basic complex analysis – Ein ausführlicher Beweis der Chudnovsky-Formel mit elementarer Funktionentheorie [155].
7 A formalization of Borel determinacy in Lean Companion paper: A formalization of Borel determinacy in Lean [149].
8 A sharp 5/8 bound for Erdős Problem 865 No publication identified; see upstream documentation.
9 ABC implies that Ramanujan’s tau function misses almost all primes Companion paper: ABC implies that Ramanujan’s tau function misses almost all primes [12].
10 Accelerated Nesterov convergence under local Polyak-Lojasiewicz conditions Companion paper: Optimal local linear convergence of Nesterov’s accelerated gradient method for C2C^{2} functions under the Polyak–Łojasiewicz inequality [67].
11 Almost all primes are partially regular Companion paper: Almost all primes are partially regular [45].
12 Analysis of Boolean functions in Lean Mathematical source (book): Analysis of Boolean Functions [168].
13 Apportionmentlib Mathematical source (book): Fair Representation: Meeting the Ideal of One Man, One Vote [22].
14 Archon-FirstProof-Results Source challenge paper: First Proof [1].
15 Artin-Wedderburn Theorem Mathematical source: The Wedderburn-Artin Theorem [35]. Mathematical source (book): A First Course in Noncommutative Rings [128].
16 Asymptotically attaining the Moore bound Companion paper: Asymptotically attaining the Moore bound [212].
17 Aumann’s Agreement Theorem Companion paper: We Can’t Agree to Disagree, Formally: Aumann’s Theorem and Assumption Accounting in Lean [47].
18 Axiomatic projective geometry (Faure–Frölicher) Mathematical source (book): Modern Projective Geometry [66].
19 Bennett–Bernstein, Freedman, and Hoeffding concentration inequalities Companion manuscript for reusable components: Linear-Error Clique Partitions of Split Graphs via Structured Triangle Packing [78].
20 BKLO simultaneous perfect matchings and spread matchings Companion manuscript for reusable components: Linear-Error Clique Partitions of Split Graphs via Structured Triangle Packing [78]. Mathematical source: Edge-decompositions of graphs with high minimum degree [25].
21 Boolean Isoperimetry and Conway–Guy Coherent Gaps Mathematical source: A sum packing problem of Erdös and the Conway-Guy sequence [32]. Companion manuscript: Robust Harper Stability at the Exponential Scale [157]. Mathematical source: Optimal numberings and isoperimetric problems on graphs [90].
22 Bounds for the centered Hardy-Littlewood maximal constant Companion proof note: A lower bound 1.6855 for the planar centred maximal constant over squares [139].
23 Brauer Group Core Mathematical source (book): Central Simple Algebras and Galois Cohomology [80].
24 Burkholder Martingale Transform Inequality Mathematical source: Boundary Value Problems and Sharp Inequalities for Martingale Transforms [37].
25 Carlet’s Kasami cyclic-additive conjecture Companion paper: On a conjecture on the Kasami APN function: reductions, structure theorems, a proof for kmodn∈{1,2,n−2,n−1}k\bmod n\in\{1,2,n{-}2,n{-}1\}, and exhaustive verification for n≤13n\leq 13 [162].
26 Certified Runge-Kutta Order Conditions Companion preprint: What Order a Method Knows: certified Runge–Kutta order conditions over rooted trees, machine-checked in Lean 4 [159].
27 Chaos and the Julia set of the complex exponential Mathematical source: The exponential map is chaotic: An invitation to transcendental dynamics [199].
28 Chebyshev Quotients and Demazure Multiplicities Companion paper: Chebyshev quotients, Demazure multiplicities, and Dyck-path models [31].
29 Chordal graphs: minimal separators and simplicial vertices Companion preprint: Complete-Split Extremizers for a Fractional Triangle-Cover Functional on Chordal Graphs [77].
30 Chvátal’s conjecture and sharp Boolean correlation Companion paper: A proof of Chvátal’s conjecture via a sharp correlation inequality [40].
31 Circuit Complexity in Lean 4 No publication identified; see upstream documentation.
32 circuitlib: a circuit verification library for Lean 4 Mathematical source: A Complete Theory of Sequential Digital Circuits: Denotational, Operational and Algebraic Semantics [75].
33 Classification of Compact Surfaces No publication identified; see upstream documentation.
34 Classification of groups of order p * q Companion paper: Classifying the groups of order p​qpq in Lean [91].
35 Classification of low-dimensional solvable Lie algebras Mathematical source (book): Classification and Identification of Lie Algebras [201].
36 Clawristotle: Vlasov-Maxwell-Landau steady-state classification Companion paper: Semi-Autonomous Formalization of the Vlasov-Maxwell-Landau Equilibrium [108].
37 Coinductive Interaction Trees using QPFs Prior Coq formalization paper: Interaction trees: representing recursive and impure programs in Coq [221].
38 Common-neighbour counterexamples for Saxl graphs Companion paper: Common neighbour conjectures for Saxl graphs fail at every base size [188]. Software release: Computational source code and Lean formalization for Common neighbour conjectures for Saxl graphs fail at every base size [189]. The registered Zenodo DOI identifies the accompanying software.
39 Conditional formalization of Zhou’s Connes-rigidity counterexample Companion paper: ICC property (T) groups without W∗-superrigidity [230].
40 Connes-Kreimer Hopf algebra of rooted trees Companion preprint: How a Tree Forgets Its Order: the Eulerian idempotent and the Adams spectrum on the commutative Connes–Kreimer–Butcher Hopf algebra of rooted trees, machine-checked in Lean 4 [161]. Companion preprint: How a Tree Remembers Its Cuts: the Connes–Kreimer–Foissy Hopf Algebra of Rooted Trees, Machine-Checked in Lean 4 [160].
41 Convex Three-Distance Degree-Six Theorem and Exceptional-Word Closures Mathematical source: On the graph of large distances [65].
42 Counterexamples to graph compactness and two-degenerate extremal bounds Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
43 Counting Critical Portraits Mathematical source: Degree-d-invariant laminations [208]. Software release: no-way-labs/lean-critical-portraits: lean-critical-portraits [118]. The registered Zenodo DOI identifies software.
44 Craig interpolation and Beth definability for propositional dynamic logic Companion paper: Propositional Dynamic Logic has Craig Interpolation: a tableau-based proof [34].
45 Craig Interpolation for Gödel-Löb logic via coalgebraic proofs Companion thesis: Proofs as Coalgebras [79]. Mathematical source: Interpolation properties for provability logics GL and GLP [197].
46 Dead Ends in Square-Free Digit Walks Companion paper: Dead ends in square-free digit walks [44].
47 Demazure Operators and Lean Companion thesis: Demazure operators and Lean [192]. Mathematical source: SCHUBERT CELLS AND COHOMOLOGY OF THE SPACESG/P [28].
48 Density-One GKP Divisibility and Its Carry-Language Characterization Mathematical source (book): Concrete Mathematics: A Foundation for Computer Science [86].
49 Dilatations of categories and commutative rings Companion paper: Dilatations of categories, via their Lean formalization [152]. Prior formalization: Formalizing multi-graded Brenner–Schröer Proj schemes and dilatations of rings in Lean4 [153].
50 Directed Topology in Lean 4 Companion paper: The Directed Van Kampen Theorem in Lean [26].
51 Disproof of the Aharoni-Korman Conjecture Companion paper: The Aharoni–Korman conjecture is false [99]. Mathematical source: Greene-Kleitman’s theorem for infinite posets [5].
52 Disproof of the Köthe conjecture (Krempa’s matrix form) Companion paper: A counterexample to Köthe’s conjecture and a question of Rowen [4].
53 DomainTheory Companion manuscript: Scott’s 3 Successively Less Topological, Simpler, and More Constructive Presentations of Domain Theory and Their Equivalence [58]. Mathematical source: Domains for denotational semantics [194].
54 Duality theory in linear optimization and its extensions Companion paper: Duality theory in linear optimization and its extensions – formally verified [61].
55 Ehrhart’s sharp volume inequality Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
56 Erdős Problem #137: powerful products of consecutive integers No publication identified; see upstream documentation.
57 Erdős Problem #367 Companion manuscript: Powerful parts of consecutive integers and Davenport–Zannier polynomials [103].
58 Erdős Problem 346: Ratio Limit Forces the Golden Ratio No publication identified; see upstream documentation.
59 Euclidean Distance Geometry Mathematical source: Remarks to Maurice Frechet’s Article “Sur La Definition Axiomatique D’Une Classe D’Espace Distances Vectoriellement Applicable Sur L’Espace De Hilbert [193].
60 Euler’s pentagonal number theorem No publication identified; see upstream documentation.
61 Euler’s pentagonal number theorem, two independent proofs, and the Jacobi triple product Formalization blueprint: Pentagonal Number Theorem [51].
62 Even graphs are edge-disjoint unions of cycles Companion manuscript for reusable components: Linear-Error Clique Partitions of Split Graphs via Structured Triangle Packing [78].
63 Event Structures and Causal-Consistent Reversibility Mathematical source: Event structures [217].
64 Exact odd-prime distributions of central-binomial valuations Mathematical source (working paper): Carry-Run Theorem and Sakib Index for the Exact Distribution of νp​(2​nn)\nu_{p}\binom{2n}{n} over nn mod pkp^{k} [191]. Mathematical source: Über die Ergänzungssätze zu den allgemeinen Reciprocitätsgesetzen. [122].
65 Exceptional Set in the abc Conjecture Companion paper: Bounds on the exceptional set in the a​b​cabc conjecture [27].
66 Existence of a finitely presented non-sofic group Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
67 Existence of Nash equilibria via Brouwer’s fixed-point theorem Mathematical source: Equilibrium points in n -person games [163].
68 Explicit type A_​n\_n and BC_​n\_n root pairings Mathematical source (book): Introduction to Lie Algebras and Representation Theory [104].
69 Extended Demazure Product on ASP Permutations Companion paper: An extended Demazure product on integer permutations via min-plus matrix multiplication [182].
70 Factorization Systems Mathematical source (book): Category Theory in Context [186].
71 Feige’s sharp unit-slack inequality Companion paper: Sharp small-deviation inequalities for sums of independent nonnegative random variables [72].
72 Fel’s Conjecture on Syzygies of Numerical Semigroups Companion paper: Fel’s Conjecture on Syzygies of Numerical Semigroups [46].
73 Fermat’s Last Theorem for regular primes Companion paper: A complete formalization of Fermat’s Last Theorem for regular primes in Lean [29].
74 FinEqs - reducing equations defining a subset of n-space over a finite field Companion paper: Reducing the number of equations defining a subset of the nn-space over a finite field [24].
75 Finite max-flow / min-cut Companion manuscript for reusable components: Linear-Error Clique Partitions of Split Graphs via Structured Triangle Packing [78].
76 Finite witnesses and the width hierarchy for language generation Companion paper: Characterizing Language Generation in the Limit: Finite Witnesses and a Separation-Width Hierarchy [136].
77 Finite Čencov-Petz Uniqueness Mathematical source (book): Statistical Decision Rules and Optimal Inference [49].
78 Finite-N Kuramoto Synchronization Companion preprint: Kuramoto-lean: Lean 4 Library for Finite-N Kuramoto Synchronisation Dynamics [39].
79 Finite-time breakdown for Navier–Stokes and Euler Companion paper: Finite Time Blowup for Navier–Stokes [171]. Companion paper: Finite Time Blowup for the Euler Equation [170].
80 Finitely generated cones and finite LP duality Companion preprint: Affine Profile Reduction for Fractional Triangle Packings in Split Graphs [76].
81 First Order Language of ZF Set Theory Mathematical source (book): Set Theory [111].
82 Flean: Floating-Point Numbers in Lean Technical standard: IEEE Standard for Floating-Point Arithmetic [107].
83 Formal Learning Theory Kernel Companion textbook: A Textbook of Formal Learning Theory [87].
84 Formalisation of the Bruhat-Tits Tree Companion paper: Formalising the Bruhat-Tits Tree [146]. Mathematical source (book): Trees [196].
85 Formalizations of theorems related to model checking Mathematical source: An automata-theoretic approach to linear temporal logic [214].
86 Formalized Complex Analysis in Lean No publication identified; see upstream documentation.
87 Foundational local complex-analytic geometry No publication identified; see upstream documentation.
88 FrontierMath Ramsey Hypergraphs Companion manuscript: A constant-factor lower bound for H⁡(n)H(n) [84]. Mathematical source: Choosing between incompatible ideals [36].
89 Fundamental groups of finite connected graphs Mathematical source (book): Algebraic Topology [93].
90 Galois groups of quadratic polynomial iterates Mathematical source: On Stoll’s criterion for the maximality of quadratic arboreal Galois representations [135]. Mathematical source: Arboreal Galois representation for a certain type of quadratic polynomials [134]. Mathematical source: Galois groups over ℚ\mathbb{Q} of some iterated polynomials [202].
91 Gaussian Moments Conjecture counterexamples Companion paper: Small Counterexamples to the Gaussian Moments Conjecture [143].
92 Genus-Zero Pairings and Catalan Numbers No publication identified; see upstream documentation.
93 Geometric Four-Colorings of the Moser Lattice and Ring Companion paper: A note on geometric colorings of the Moser lattice [60].
94 Global completeness of one-sorted definedness-free matching logic Companion paper: Completeness and incompleteness of basic matching logic [48].
95 Grothendieck’s Vanishing Theorem Companion expert-review paper: Sorries Are Not the Hard Part: An Expert-Review Case Study of a Semi-Autonomous Formalization [109]. Mathematical source (book): Algebraic Geometry [92].
96 Gödel’s First and Second Incompleteness Theorems Mathematical source: Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I [81].
97 Halving the Kalton-Roberts upper bound Companion paper: Halving the original Kalton–Roberts upper bound for nearly additive set functions [98].
98 Homogeneous self-dual interior-point method for linear programming Mathematical source: An O(\sqrt{}nL)-Iteration Homogeneous and Self-Dual Linear Programming Algorithm [223].
99 Hurwitz’s Classification of Euclidean Composition Algebras Mathematical source (document): Über die Composition der quadratischen Formen von beliebig vielen Variablen [106].
100 Improved asymptotic bounds for binary and spherical codes Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
101 Improved long gaps between consecutive primes Companion paper: Improved Long Gaps Between Primes [174].
102 Improved Lower Bounds for Strong n-Conjectures Companion paper: IMPROVED LOWER BOUNDS FOR STRONG n-CONJECTURES [100].
103 Infinitary logic and countable model theory Mathematical source (book): Lectures on Infinitary Model Theory [150].
104 Infinite counterexamples to Connes’s rigidity conjecture Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
105 Irrationality of ζ\zeta(3) Companion paper: A Formal Proof of the Irrationality of ζ⁡(3)\zeta(3) in Lean 4 [141]. Mathematical source: A Note on the Irrationality of ζ\zeta(2) and ζ\zeta(3) [30].
106 Jordan–Schönflies theorem No companion paper is listed in the retained upstream documentation; the repository documents the theorem and its proof structure.
107 Kahn–Kalai expectation-threshold theorem Mathematical source: A Short Proof of Kahn-Kalai Conjecture [209].
108 Known Bounds for the Hadwiger-Nelson Problem Mathematical source: The chromatic number of the plane is at least 5 – a human-verifiable proof [179].
109 Kolokolnikov’s ACMAX conjecture Companion paper: Maximizing Algebraic Connectivity with 2(n-2) Edges: The Large Vertex Number Case [231]. Source conjecture: Maximizing algebraic connectivity for certain families of graphs [120].
110 Kurosh and Schreier subgroup theorems via covering graphs Mathematical source: Die Untergruppen der freien Produkte von beliebigen Gruppen [127].
111 Lean-QuantumAlg Companion paper: Building Shor’s Algorithm in Lean: An Agentic Formalization of Quantum Attacks on RSA-2048 and P-256 [227]. Companion paper: An Agentic Formalization for Certified Quantum Neural Network Design [112].
112 Lehmer’s Polynomial and the E10 Coxeter Element No publication identified; see upstream documentation.
113 Lentil: Temporal Logic of Actions in Lean 4 No publication identified; see upstream documentation.
114 Leo Moser’s finite distinct-subset-sums inequality Mathematical source: Sets of Integers Whose Subsets Have Distinct Sums [89].
115 Maxima of Coxeter frieze patterns are Fibonacci numbers Companion paper: On Upper Bounds of Frieze Patterns [42].
116 Mean-field derivation and well-posedness of the Vlasov equation Companion paper: A Formalization of the Mean-Field Derivation of the Vlasov Equation [156].
117 Minimum modulus for the unique multiset-sum problem Mathematical source (preprint): The Exact Minimum Order for Unique Multiset Sums in Finite Abelian Groups [110]. Companion paper: Minimum modulus for the unique multiset-sum problem [69].
118 Misere combinatorial games No publication identified; see upstream documentation.
119 Modular forms and the generalized residue theorem Mathematical source: Non-Integer Valued Winding Numbers and a Generalized Residue Theorem [105].
120 Monlib4 Operator-Algebra and Quantum-Set Core Companion thesis: On quantum graph theory: non-commutative graph theory [169]. Mathematical source: A Constructive Elementary Proof of the Skolem-Noether Theorem for Matrix Algebras [203].
121 Monotonicity Formula for Stationary Harmonic Maps Mathematical source (book): Theorems on Regularity and Singularity of Energy Minimizing Maps [200].
122 Monsky’s Theorem Mathematical source: On Dividing A Square Into Triangles [158].
123 Moreira’s version of Sard’s theorem Mathematical source: Hausdorff measures and the Morse-Sard theorem [56].
124 MRiscX Mathematical source: Hoare-Style Logic for Unstructured Programs [147].
125 Nash-Williams fronts and 2-better-quasi-orders Mathematical source: Towards better: A motivated introduction to better-quasi-orders [180].
126 Near-perfect triangle packings from sum-zero triples Companion preprint: Linear-Error Clique Partitions of Split Graphs via Structured Triangle Packing [78].
127 Neukirch’s Algebraic Number Theory: Hilbert ramification theory Mathematical source (book): Algebraic Number Theory [165].
128 Nivat’s conjecture Companion AI-generated manuscript: Pattern complexity and Nivat’s conjecture [85].
129 On the paucity of lattice triangles Companion paper: On the paucity of lattice triangles [13].
130 Optimal Pebbling Number of the Hypercube Companion paper: Optimal pebbling of the hypercube [176].
131 Oracle Computability and Turing Degrees Prior formalization paper: Formalizing computability theory via partial recursive functions [38].
132 Osterwalder-Schrader Axioms for the Gaussian Free Field No publication identified; see upstream documentation.
133 Partial Combinatory Algebras Mathematical source: Partial Combinatory Algebras [178].
134 PCF Theory Mathematical source: Cardinal Arithmetic [2].
135 Period Lengths of Rational Cut-and-Project Gap Sequences Companion manuscript: Period Length Formulas for Rational Cut-and-Project Strip Projections: Multiset and Set Cases [124].
136 Planar Point Sets Whose Non-Diameter Distances Form a Geometric 3-Chain No publication identified; see upstream documentation.
137 Poincaré’s Nonintegrability Theorem for the Restricted Three-Body Problem Mathematical source: A new proof of Poincaré’s result on the restricted three-body problem [219].
138 Pointwise Birkhoff Ergodic Theorem Mathematical source (book): Ergodic Theory [181].
139 Poitou’s Explicit Odlyzko Bound for Root Discriminants Mathematical source (document): Sur les petits discriminants [184].
140 Polylean Unit Conjecture Counterexample Mathematical source: A counterexample to the unit conjecture for group rings [73].
141 Polynomial ABC (Mason–Stothers) and its corollaries Companion paper: Formalizing Mason-Stothers Theorem and its Corollaries in Lean 4 [21].
142 Polynomial parametrizations of Pythagorean triples Mathematical source: Parametrization of Pythagorean triples by a single triple of polynomials [71].
143 Polynomial-factor hardness of the closest vector problem Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
144 Prekopa-Leindler, Brunn-Minkowski, and the isoperimetric inequality No publication identified; see upstream documentation.
145 Primitive Sets Above x (Erdos Problem 1196) Companion paper: Primitive sets and von Mangoldt chains: Erdős Problem #1196 and beyond [8].
146 Pumping Lemma for Context-Free Grammars No publication identified; see upstream documentation.
147 Pólya’s enumeration theorem No publication identified; see upstream documentation.
148 Quadratic-order prime classification Mathematical source (thesis): Singular Moduli and the Ideal Class Group [74].
149 Quantum parallel repetition Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
150 Quartic-over-logarithmic lower bound for rational permanent formulas Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
151 Quasi-Borel Spaces Mathematical source: A Domain Theory for Statistical Probabilistic Programming [211]. Mathematical source: A convenient category for higher-order probability theory [96].
152 Radó’s theorem for Riemann surfaces No publication identified; see upstream documentation.
153 Rellich–Kondrachov compact embedding theorem No publication identified; see upstream documentation.
154 Riemann Mapping Theorem No publication identified; see upstream documentation.
155 Riemann–Roch for algebraic function fields No publication identified; see upstream documentation.
156 RL Theory in Lean Companion paper: Towards Formalizing Reinforcement Learning Theory: A Robbins-Siegmund Approach [228].
157 Sabidussi’s compatibility conjecture for Eulerian multigraphs Companion manuscript: Graph Puzzles III.1: A proof of Sabidussi’s compatibility conjecture [210].
158 Selberg Sieve Mathematical source: Lectures on sieves [95].
159 Sensitivity Conjecture: sqrt(deg) <= sensitivity Mathematical source: Induced subgraphs of hypercubes and a proof of the Sensitivity Conjecture [102].
160 Serre’s mass formula for totally ramified local-field extensions Mathematical source: Une formule de masse pour les extensions totalement ramifiées de degré donné d’un corps local [195].
161 Shannon Entropy Characterization Mathematical source: A Mathematical Theory of Communication [198].
162 Sharp asymptotic upper bounds for sphere packing Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
163 Sharp Five-Distance and Sup-Norm Gap Theorems Software release: Sharp five-distance and sup-norm gap theorems for Kronecker sequences (Lean 4 / Mathlib) [33]. Mathematical source: Higher dimensional gap theorems for the maximum metric [94].
164 Simple zeros of the Riemann zeta function Companion paper: A new proof that more than 2/32/3 of the zeros of the Riemann zeta function are simple and on the critical line [129].
165 Spectral positivity No publication identified; see upstream documentation.
166 Spectral theorem for compact self-adjoint operators Mathematical source: Chapter II. Compact Self-Adjoint Operators [41].
167 Square-difference-free polynomials beyond Naslund’s conjectured bound Companion proof note: Square-difference-free sets in F3​[T]F_{3}[T] past the conjectured bound [114]. Source conjecture: Paley Graphs and Sárközy’s Theorem in Function Fields [164].
168 Stable phase retrieval for Hermite-Fock expansions No publication identified; see upstream documentation.
169 Strengthened V0 Bounded Arithmetic Interfaces Mathematical source (book): Logical Foundations of Proof Complexity [52].
170 Sums of Distinct Factorials That Are Powers of Two (Erdos Problem 403) No publication identified; see upstream documentation.
171 Sums of Three Squares Mathematical source: Sums of three squares [14].
172 Sundog certificates No publication identified; see upstream documentation.
173 Superexponential lower bounds for multicolor triangle Ramsey numbers Companion paper: Ten Advances in Mathematics and Theoretical Computer Science [175].
174 Synthetic Euclidean geometry: Euclid’s Elements Book I Mathematical source: A FORMAL SYSTEM FOR EUCLID’SELEMENTS [18].
175 Thakur’s hypotheses on power sums of F_q[t] Companion paper: Thakur’s hypotheses on power sums of 𝔽q​[t]\mathbb{F}_{q}[t] [43].
176 The 5/8 theorem Mathematical source: What is the Probability that Two Group Elements Commute? [88]. The registry DOI resolves to an unrelated article; this row cites the Gustafson paper named by the registry.
177 The Anderson Conjecture: A Weakly Quasi-Complete Ring Need Not Be Quasi-Complete Mathematical source: Quasi-complete Semilocal Rings and Modules [11].
178 The Bannai-Bannai-Stanton Bound on Distance Sets Companion report: Formalisation of the Bannai-Bannai-Stanton Theorem in Lean 4 [59].
179 The Bollobás–Nikiforov conjecture Companion manuscript: The Bollobás–Nikiforov inequality for nonnegative edge weights [53].
180 The Convex-Octagon Case of Erdős Problem 97 Mathematical source: On Sets of Distances of n Points [63].
181 The Cramer-Wold theorem Mathematical source: Some Theorems on Distribution Functions [55].
182 The Erdős–Graham–Ruzsa–Straus two-prime theorem Companion draft: A Machine-Checked Proof of the Erdős–Graham–Ruzsa–Straus Two-Prime Theorem [62]. Mathematical source: On the prime factors of (2​nn)\binom{2n}{n} [64].
183 The Erdős–Sós conjecture (Erdős Problem 548) The retained upstream documentation states that a paper is in preparation; it links the Erdős Problems statement and its Lean formalization.
184 The Erdős–Tuza–Valtr conjecture Companion paper: On the Erdős–Tuza–Valtr conjecture [20].
185 The Fundamental Inequality of Valued Fields No publication identified; see upstream documentation.
186 The Hanson-Wright inequality for sub-Gaussian quadratic forms Mathematical source (book): High-Dimensional Probability [215].
187 The Jacobian of a Compact Riemann Surface No publication identified; see upstream documentation.
188 The Johnson–Lindenstrauss Lemma and Quantized JL Mathematical source (QJL): QJL: 1-Bit Quantized JL Transform for KV Cache Quantization with Zero Overhead [225]. Mathematical source: Extensions of Lipschitz mappings into a Hilbert space [113].
189 The Komlós and Beck–Fiala bounds with constant 36 Companion paper: An elementary proof of the Komlós conjecture [116].
190 The Kunen inconsistency theorem Mathematical source: Elementary embeddings and infinitary combinatorics [123].
191 The main theorem of polytopes No publication identified; see upstream documentation.
192 The optimal exponent relating sumsets and difference sets Companion paper: Settling the Optimal Exponent Relating Sumsets and Difference Sets [138].
193 The polynomial Freiman–Ruzsa conjecture Mathematical source: Marton’s Conjecture in abelian groups with bounded torsion [83]. Mathematical source: Improved Exponent for Marton’s Conjecture in 𝔽2n\mathbb{F}_{2}^{n} [137]. Mathematical source: On a conjecture of Marton [82].
194 The polynomial method and restricted sums of congruence classes Mathematical source: The Polynomial Method and Restricted Sums of Congruence Classes [9].
195 The Ramanujan-Nagell theorem Companion paper: A formal proof of the Ramanujan–Nagell theorem in Lean 4 [23].
196 The Rupert Problem for convex polyhedra No publication identified; see upstream documentation.
197 The Three-Gap (Steinhaus) Theorem Companion manuscript: A Lean 4 Formalization of the Three-Gap (Steinhaus) Theorem, Uniform in the Rotation Number [125]. Mathematical source: The Three Gap Theorem (Steinhaus Conjecture) [213].
198 The transcendence of π\pi Mathematical source: The Transcendence of π\pi [166].
199 The Wallace problem in ZFC Companion paper: The Wallace problem and countably compact torsion-free Abelian groups in ZFC [70].
200 The Zhang-Yeung non-Shannon information inequality Mathematical source: On characterization of entropy function via information inequalities [229].
201 Turán’s theorem (the "Book" weighting proof) Mathematical source (book): Proofs from THE BOOK [6].
202 Ulm’s theorem for countable reduced abelian p-groups No publication identified; see upstream documentation.
203 Unconditional Schauder Bases Mathematical source (book): Classical Banach Spaces I and II [140].
204 Uniqueness of Shannon Capacity-Achieving Priors Mathematical source (book): Elements of Information Theory [54].
205 Verified canonical labelling of finite graphs No publication identified; see upstream documentation.
206 Verified interval-Cauchy real arithmetic Prior Coq formalization paper: Certified Exact Transcendental Real Number Computation in Coq [167].
207 Virasoro Project Mathematical source (book): Infinite-Dimensional Lie Algebras [115].
208 Weierstrass models and singular points for Tate’s algorithm Mathematical source (document): Elliptic Curve Handbook [50].
209 Whitehead’s theorem for CW-complexes No publication identified; see upstream documentation.
210 Wigner Semicircle Distribution Mathematical source (lecture notes): Math 247A: Introduction to Random Matrix Theory [117].
211 ZFLean Mathematical source (book): Set Theory [111].
Concept demonstration

Open image ↗