Follow each problem back to the paper, conference, or research community that posed it. This directory shows the collections represented in the index and how their evidence is used.
Coverage varies by source. Counts below are records currently in this index; nested questions are kept inside their parent record. Full snapshot coverage does not mean every record has been independently reviewed.
New: 223 automatically checkable problems from 3 sources ↓
126 questions from 15 research sources (September) ↓
The complete AMS / zbMATH classification connects subject areas to selected journals, reviewed articles and dated question summaries with original problem numbers and page references.
The saved International Congress of Mathematicians catalogue, with original statements, remaining questions, and proceedings links.
2898 / 2,898 source records in source snapshot
Complete saved catalogue: 2,898 records with 4,466 nested remaining questions. Only 10 records are marked reviewed by the source. Different records may address related or equivalent conjectures.
Snapshot: 2026-09-08
30 papers from the supplied COLT archive, with original text, figures, and dated status notes.
30 / 30 papers in source snapshot
All 30 papers in the supplied local manifest are included, containing 88 question labels. Subquestions are kept within each paper.
Snapshot: 2026-09-05
Mathematical problems with explicit statements, research context, and a record of progress.
3909 / 3,909 public source records in source snapshot
Full public directory: 2,678 source-curated records and 1,231 backlog candidates. Candidates are marked status unassessed. Full statement text is included; research histories and linked project artifacts remain at the source. Source Lean claims are not local verification.
Snapshot: 2026-09-08
Ten categories spanning quantum information, many-body theory, quantum fields, spectral theory, and foundations.
114 / 114 problems in source snapshot
All 10 category files fetched: 113 active records, plus the earlier E2 entry retained as archived. Other merged/background records are retained in the archive metadata, not counted as separate active problems.
Snapshot: 2026-09-08
Research questions developed in American Institute of Mathematics workshops.
3,470 source records after reconciling 165 AIMPL root lists with the earlier 140-link AIM directory. This round adds 1,695 records; five demonstration lists excluded. 57 PDF lists remain available as documents, without individual question extraction. Historical status is unassessed unless an explicit source resolution is recorded.
Snapshot: 2026-09-10
The foundational mathematical challenges selected by the Clay Mathematics Institute.
Seven problems posed in 2000, each carrying a US$1,000,000 prize. One (Poincaré) has been resolved.
New: seven more specialist collections and the AMR directory ↓
100 prepared additions from 15 Wikipedia and specialist collections. This is a partial selection; current status is unassessed. Existing formulations receive additional references. Counts below reflect the imported records.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification. The source carries factual-accuracy and sourcing warnings; these selections require specialist review.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Selected questions inspected on 9 September 2026; partial coverage. Editorial summaries, not full article transcriptions. Current status unassessed; inclusion is not independent verification.
Six selected questions from the open-problem sections, inspected 9 September 2026. Solved section excluded. Historical formulations; current status unassessed.
All 10 numbered questions collected on 6 June 2022, summarized with approximation factors and computational models retained. Inspected 9 September 2026; subsequent status unassessed.
All 78 numbered pages inspected; 37 selected formulations across two rounds, including one linked to an existing Dürer record. Source-marked resolutions excluded, along with stale open labels for problems 52 and 67. Current status unassessed; partial catalogue coverage.
Discovery index for the subject lists above. Undecidability and NP-/PSPACE-completeness lists are not automatically open problems.
Located through the parent index; no medical questions imported in this selection. Individual formulations and current evidence still need review.
223 new problem records from 3 actively maintained collections whose answers are constructions, certificates or numerical bounds that a program can check. Each record keeps the source’s own status label and dated bounds; reported solutions and bounds are not independently certified here.
All 49 listed research problems with automatically verifiable answers, from Epoch’s published data archive and problem pages (data updated 7 October 2026). Expanded after June 2026 workshops and featured by Scientific American (2 October 2026). Status follows Epoch’s labels: 39 unsolved, 10 marked solved (5 by AI, 5 by human + AI).
49 catalogue entries · 49 new problem records
All 115 tabulated constants with their descriptions and the repository’s best known lower and upper bounds at the pinned commit. The maintainers review submissions for plausibility, not certified correctness; asterisked bounds have minimal verification.
115 catalogue entries · 115 new problem records
59 of the 67 problem pages, with AlphaEvolve result categories and recorded bounds; 8 pages are rediscovery tests of classical results, competition problems or puzzles and remain catalogue-only. The repository records AlphaEvolve performance, not the current status of each question.
67 catalogue entries · 59 new problem records
492 Lean-formalized OEIS conjectures with language-model resolution results (160 settled). Largely overlaps the 164 OEIS-derived Formal Conjectures records already indexed; better used as a status cross-check than as new problem records.
Weekly LLM-extracted conjectures from new arXiv math papers (890 at launch). Openness and interest labels are model-generated and no license is stated; needs question-level review before import.
68 Erdős problems posed for Lean-verified resolution; already covered by the Erdős Problems source, retained as a cross-reference for status updates.
126 new question records from 15 sources, with MSC2020 subject mappings and original section or page links. Licensed statements and editorial summaries are identified on each record. Numbered source pointers and resolved items remain separate from the main count.
All 22 numbered problems from the August 2026 version, with contributor context and definitions. Nested subquestions remain grouped.
22 catalogue entries · 22 new problem records
Explicit research-question bullets from the March 2025 author version (published in Stochastic Systems). Definitions and existing-result bullets are context, not new records. Systems-design directions are labeled programmes.
42 catalogue entries · 42 new problem records
All 18 numbered topics in version 2 indexed. Three postscripts report solutions (9, 11, 13); these are excluded from the new problem records. Optical-fiber subquestions remain grouped.
18 catalogue entries · 15 new problem records
All 94 source-labeled problems indexed; two selected questions summarized. Terminology is defined in section 2 of the linked paper.
94 catalogue entries · 2 new problem records
All 47 numbered problems in version 3 indexed. Two benchmark-design directions are selected; numbered pointers alone do not certify current openness. Read the source updates for later solutions.
47 catalogue entries · 2 new problem records
Eight selected questions from sections 4.1–4.9; section 4.2 remains a source pointer pending a formulation and progress check.
9 catalogue entries · 8 new problem records
Four selected questions from the five-section open-problem session. Datalog convergence is retained only as a pointer because later bounds require reconciling the exact model.
5 catalogue entries · 4 new problem records
Selected questions from pp. 2756–2760 of Oberwolfach Report 51/2025. The source-marked solution sketch and formulations needing clarification are excluded.
12 catalogue entries · 11 new problem records
Selected source questions; current status unassessed.
1 catalogue entries · 1 new problem records
Selected source questions; current status unassessed.
4 catalogue entries · 4 new problem records
Three selected questions from the Tokyo FDIG 2025 contributors; this is not a complete transcription of all eight contributor sections.
3 catalogue entries · 3 new problem records
Two questions selected from section 3.1 of SIAG/OPT Views and News 33(1). This is an authored perspective, not an official exhaustive SIAM problem list.
2 catalogue entries · 2 new problem records
Selected source questions; current status unassessed.
3 catalogue entries · 3 new problem records
Four selected research programmes from sections 3.2–3.5, retaining the source discussion. Broad modelling directions are not presented as sharp conjectures.
4 catalogue entries · 4 new problem records
Three grouped research programmes from the Outlook, with the complete corresponding subsection as context. Speculative connections are labeled as programmes.
3 catalogue entries · 3 new problem records
Curated combinatorics and optimization wiki; individual formulations and solved-category membership need review before import.
Historical list; major subsequent solutions make blanket open-status import inappropriate.
Historical expert problem collection; dated status and individual formulations require review.
Published problem session; retained as discovery pending full question-level review.
Seminar report; broad talk abstracts are not treated as self-contained problem records.
Scientific white paper retained for a later question-level extraction.
Seven collections and the AMR discovery directory. 522 new problem records; numbered book indexes and directory links are counted separately. Browse each catalogue to find original pages, linked formulations and source-marked solutions.
All 100 original numbers indexed by PDF page. Source-marked solved entries remain visibly labeled in the catalogue. Available formalizations are linked separately; the author PDF is not republished.
100 catalogue entries · 0 new problem records
September 2026 edition. Active sections and the solved archive are indexed separately; active-section membership does not certify current openness. Includes all 150 problems in Issue 21.
1,935 catalogue entries · 0 new problem records
2026 author version. Problem numbers, subject sections and page links are indexed. The book explicitly prohibits reposting; its text and PDF are not copied into the public site.
366 catalogue entries · 0 new problem records
All nine submitted topic sections indexed with direct links. A section may contain multiple questions. The index is not counted as nine fully transcribed problems.
9 catalogue entries · 0 new problem records
All 16 numbered problems, 16–31, from the March 2026 paper. Definitions and surrounding section context are retained. Overlapping named problems link to existing records.
16 catalogue entries · 15 new problem records
All 50 problem numbers from version 3, April 2026. Problems 10 and 30 belong to the solved section and are catalogued there, not imported as open candidates.
50 catalogue entries · 48 new problem records
Fixed repository commit. Only files with actual research-open declarations are eligible for problem records; tests, textbook exercises and solved-only files are excluded. Source declarations are not locally verified proofs.
1,250 catalogue entries · 459 new problem records
Bibliographic links from the directory are searchable. Directory links are not counted as individual open problems and are not recursively imported.
109 catalogue entries · 0 new problem records
678 additional source records. Historical questions are marked status unassessed; they may have since been resolved. Related formulations across collections are not necessarily distinct problems. COLT remains limited to the 30 papers in the supplied verified archive.
634 unresolved source records represented: 631 additions and 3 existing formulations. All 1,220 numbered pages examined; 586 resolved or independence-labelled records excluded.
All 12 numbered questions in the 2021 article, with section context. Current status unassessed.
All 20 questions in the 2014 revision, with control-model definitions. Current status unassessed.
10 substantive slides from the 2024 workshop. The closing broad discussion prompt is not counted. Current status unassessed.
2 additional formulations from 2025. Two other questions are linked to the corresponding FOCS entries. Current status unassessed.
3 research directions from one seminar report, published May 2026. This does not cover the full Dagstuhl archive. Current status unassessed.
13 historical topic pages examined. Some are explicitly solved; others need status review and careful transcription of image-based formulas.
Historical 2017 survey. Overlap with foundational lists and subsequent progress need review before import.
Institutional listing confirmed; the linked problem repository could not be retrieved. Separate from Open Quantum Problems already in the index.
Existing curated entries retained. Unrelated spam appears in the current recent-post feed, so additional pages need individual review.
Collection located. Individual article questions have not yet been extracted and checked.
Useful report archives; only explicitly formulated research problems should become index entries. Directory links and talks are not problem counts.
arXiv, OpenReview, OpenAlex, Semantic Scholar, Crossref and similar services help locate papers and later progress. Their search results are not counted as open problems. ICM, AIM, Clay and the supplied COLT archive retain the coverage described above.
Read the complete workshop documents, including their original formulas and figures. These are document counts; questions inside the PDFs have not yet been individually transcribed into the index.
Links from the saved AIM directory. A workshop list can contain many questions; these links are not counted as imported problems.
A source date is not a proof of current status. Imported entries display a dated source assessment. A historical workshop list may contain questions that have since been resolved. We select individual questions and retain the limits of the available evidence.
Original statement and subsequent progress are separate. COLT entries preserve the original published text and figures alongside the archive’s remaining questions and status notes. Transcription checks establish readability and fidelity, not the validity of a proof.
A claimed result remains a claim. Preprints, numerical searches, and independently verified results carry different evidential weight. Source-side “improved” labels are shown as partial progress, with the original qualifications.
Attribution travels with the problem. Every imported entry links to the source record or primary paper. Imported material keeps its original terms; this platform does not relicense third-party papers.