About CMF Atlas

An open mathematical research database of Conservative Matrix Fields — 2D and higher-dimensional lattice recurrences with path-independent matrix products, encoding continued fractions, hypergeometric series, and named constants.

BETA

Disclaimer & Project Status

Beta Version & Research Goals

The CMF Atlas is currently in its beta stage. This platform is designed to facilitate mathematical research by making Conservative Matrix Fields (CMF) more accessible to the global community.

Ongoing Development & Verification

Please be aware that this is a live, ongoing project. Our current efforts are focused on two primary areas:

Feedback & Corrections

As this is a work in progress, errors in data may occur. We highly value community involvement — if you identify any discrepancies or bugs, please let us know via the Contact Page.

Independent Project & Credits

This is an independent project dedicated to highlighting the research conducted by the Ramanujan Machine Team. The structures presented here are built upon their discovery of the Conservative Matrix Field.

For further technical details, please refer to their official resources:

What is a Conservative Matrix Field?

A Conservative Matrix Field is a pair of matrix-valued functions \(K_1(k,m)\) and \(K_2(k,m)\) on a 2D integer lattice satisfying the flatness condition:

\[K_1(k,m)\cdot K_2(k+1,\,m) = K_2(k,m)\cdot K_1(k,\,m+1)\]

This means the matrix product along any lattice path from \((0,0)\) to \((K,M)\) is path-independent. The limiting value — as \(K,M\to\infty\) — converges to constants such as \(\zeta(3)\), \(\pi\), \(\ln 2\), and \(\zeta(5)\).

For the telescope family with polynomial form \(f(x,y)\):

\[K_1(k,m) = \begin{pmatrix} 0 & 1 \\ b(k{+}1) & a(k,m) \end{pmatrix}, \quad K_2(k,m) = \begin{pmatrix} \bar{g}(k,m) & 1 \\ b(k) & g(k,m) \end{pmatrix}\]

where \(g(k,m)=f(k,m)\), \(\bar{g}(k,m)=\bar{f}(k,m)\), \(b(k)=g(k,0)\cdot\bar{g}(k,0)\), \(a(k,m)=g(k,m)-\bar{g}(k{+}1,m)\).

Database Contents

CMF Atlas v2.3 contains entries from multiple discovery pipelines, including pFq CMFs from RamanujanTools.

Versionv2.3
Total entries (2D+)Loading…
Dimensions2D (~280 entries), 3D (2 entries), 5D+ (pFq, 1 entry)
SourcesRamanujanTools (A+), Gauge Transformed, CMF Hunter
Certification levelsA_plus (symbolically certified), A_certified (verified), B_verified_numeric (numeric), C_scouting (scouting)
Bulk downloadcmfs.json  (JSON, CC BY 4.0)
DOIPending registration — cite via BibTeX below in the interim
LicenseCC BY 4.0

Data Schema v2.3

All public CMF entries expose the following canonical fields via GET /cmfs/browse and GET /cmfs/{id}/full:

Core fields

idUnique integer identifier (stable across releases)
dimensionLattice dimension (integer ≥ 2)
f_polySymPy polynomial f(x,y) — present for telescope-polynomial CMFs
fbar_polyConjugate polynomial f̄(x,y)
degreeTotal polynomial degree of f
primary_constantCanonical label of the identified limit (e.g. “ζ(3)”, “π”)
identified_constantPSLQ-identified exact form, if found
flatness_verifiedBoolean — true if symbolic flatness proof exists
canonical_fingerprintHash-based unique key for the representation
entry_uriCanonical URL for the entry page
release_versionAtlas version in which this entry is included

Enum fields (canonical vocabulary)

certification_level A_plus A_certified B_verified_numeric C_scouting
proof_statussymbolically_certified | verified | numeric_only | unverified
identification_statuspslq_identified | matched | unidentified
construction_typetelescope_polynomial | hypergeometric_pfq | matrix_explicit
source_familyRamanujanTools | Gauge Transformed | CMF Hunter | …

Arithmetic status (constants registry)

irrationality_statusproven_transcendental | proven_irrational | open_question | algebraic
transcendence_statusproven_transcendental | open_question | algebraic

Full schema: /docs · Machine-readable release manifest: /release

How to Cite

If you use CMF Atlas in research, please cite it. Data licensed under CC BY 4.0. A DOI is pending registration — use the BibTeX below in the interim.

Project-level citation

@misc{vesterlund2026cmfatlas,
  author    = {Vesterlund, David},
  title     = {{CMF Atlas}: A Database of Conservative Matrix Fields},
  year      = {2026},
  url       = {https://davidvesterlund.com/cmf-atlas/},
  note      = {Version 2.3, open research dataset. DOI pending.},
  license   = {CC BY 4.0}
}

Individual entry citation

@misc{vesterlund2026cmf1,
  author    = {Vesterlund, David},
  title     = {{CMF Atlas} Entry \#1: $f(x,y) = (x+y)^3$},
  year      = {2026},
  url       = {https://davidvesterlund.com/cmf-atlas/entry.html?id=1},
  note      = {CMF Atlas v2.3, limit = $\zeta(3)$},
  license   = {CC BY 4.0}
}

Each entry page provides a pre-filled BibTeX block with the correct formula and metadata.

REST API

CMF Atlas is served by a FastAPI backend at cmfatlas-production.up.railway.app. Interactive Swagger docs: /docs.

All endpoints are available at both /endpoint and /api/v1/endpoint (stable versioned alias, redirects 307).

GET
/release
Release manifest — version, counts, schema
GET
/constants
Constants registry with entry counts
GET
/cmfs/browse
Search & filter CMFs
GET
/cmfs/{id}/full
Full entry with K-matrices
GET
/cmfs/{id}/relations
Linked constant, source family, related entries
GET
/cmfs/{id}/walk
Numerical walk / convergence
GET
/cmfs/{id}/conservative-test
2D flatness residual test
GET
/export/cmfs.json
Bulk dataset download (CC BY 4.0, JSON)
GET
/stats/detailed
Dashboard statistics & breakdown

Versioned alias example: GET /api/v1/cmfs/browse?dimension=2

RamanujanTools / pFq Integration

CMF Atlas includes CMFs imported directly from the RamanujanTools repository. Currently imported: e, pi, symmetric_pi, zeta3, hypergeometric_derived_2F1 (3D), hypergeometric_derived_3F2 (5D, at z=½), and five FFbar parametric families. All carry A+ certification.

# Browse pFq CMFs in the atlas
GET /cmfs/browse?source_type=pfq

# Individual entry with hypergeometric parameters
GET /cmfs/{id}/full

License

The CMF Atlas dataset and documentation are released under Creative Commons Attribution 4.0 (CC BY 4.0).

You are free to share, adapt, and use the data for any purpose including commercial and academic research, provided you give appropriate credit.