Open formal research · applied with an evidence boundary

From machine-checked claims to verifiable decisions about nature.

Viridis publishes the mathematics, preserves the assumptions, and builds the compute-and-verification layer that connects ecological evidence to a model-bound economic or land decision.

Explore the public canon ↗Run the decision services
Evidence boundary

Lean proves conditional mathematics; it does not establish empirical inputs, ecological effect sizes, registry eligibility, or commercial outcomes. Those require field data, domain review, and independent validation.

Intelligence Bound → explicit decision contracts

The Intelligence Bound provides the foundational constraint framework. Other results retain their own assumptions, evidence and relationship to that root. Our current research feedstock is established literature and formal representations, not a proprietary field sensor stream.

Inspect the foundational theoremResearch and kernel contracts

Controlled research-to-product lifecycle

New research produces a reviewable change.

The existing nightly science engine works with established knowledge and formal artifacts. It proposes models and candidate results. Candidate discovery is not accepted science, and a passed proof is not empirical validation.

  1. 01

    Established knowledge

    Literature, explicit assumptions, existing formalizations and prior Viridis results.

  2. 02

    Candidate connection

    Proposed connections retain citations and novelty uncertainty.

  3. 03

    Explicit model

    Definitions, domains, assumptions and failure cases are stated.

  4. 04

    Formal statement

    A frozen statement identifies exactly what is being claimed.

  5. 05

    Proof checking

    Pinned toolchains, dependencies and exact verification receipts are inspected.

  6. 06

    Implementation & tests

    An executable contract is checked against its limited formal model.

  7. 07

    Scientific & engineering review

    Publication merit, runtime utility and empirical status are reviewed separately.

  8. 08

    Versioned kernel release

    Reviewed source and implementation hashes enter an explicit release lock.

  9. 09

    Adoption by Viridis OS

    A deliberate runtime release changes new decisions; saved records retain their pins.

Empirical validation runs as a separate evidence track. Nightly output does not automatically alter production, issue a customer decision, or rewrite a saved scenario.

Inspect package evidencePublic scientific record

Research library · by admission state

195 canon records, grouped by what they are today.

A Lean-checked record is conditional mathematics. Its tier below says whether it is a foundation constraint, an admitted runtime kernel, a held proposal with a named gate, a verified candidate without a product wrapper, or working corpus. None is empirical validation.

Foundation 5 — Constraint framework recorded on every decision; no numerical claim.
  • Biosphere Erasure Bound
    Viridis.BiosphereErasureBound · verified · gate-passed
    no DOI yet
  • Bounded Memory Dissipation
    Viridis.P0BoundedMemoryDissipation · working · not-canon-admitted
    no DOI yet
  • Bounded Memory Learning
    Viridis.BoundedMemoryLearning · verified · gate-passed
    no DOI yet
  • The Intelligence Bound
    P0_IntelligenceBound_COMPILED · verified · gate-passed
    doi:10.5281/zenodo.19317982
  • Thermodynamic Economics
    P4_ThermodynamicEconomics · verified · gate-passed
    no DOI yet
Admitted kernel 41 — Bound to a runtime module: DOI, kernel-checked Lean, schema, units, runner, receipt.
Held proposal 5 — Kernel-verified, not wired; each has a named next gate.
Kernel candidate 15 — Verified and gate-passed in the canon; no product wrapper yet.
  • D-Score
    DScore · verified · gate-passed
    no DOI yet
  • Conservation Feasibility
    P3_Impossibility · verified · gate-passed
    no DOI yet
  • Decoherent Selection
    Viridis.Decision.DecoherentSelection · verified · gate-passed
    no DOI yet
  • Effortless Equilibrium
    Viridis.Computation.EffortlessEquilibrium · verified · gate-passed
    no DOI yet
  • Energy
    NitrogenFixation · verified · gate-passed
    no DOI yet
  • Foundations
    NitrogenFixation · verified · gate-passed
    no DOI yet
  • Geodesic Saturation
    Viridis.GeodesicSaturation · verified · gate-passed
    no DOI yet
  • Integration
    NitrogenFixation · verified · gate-passed
    no DOI yet
  • Mutualistic Attestation
    Viridis.Ecoservices.MutualisticAttestation · verified · gate-passed
    no DOI yet
  • P2 HDFM POC
    P2_HDFM_POC · verified · gate-passed
    no DOI yet
  • Perennial Corridor
    Viridis.Corridors.PerennialCorridor · verified · gate-passed
    no DOI yet
  • PSIT Symplectic
    PSIT · verified · gate-passed
    no DOI yet
  • Stewardship Setpoint
    Viridis.Stewardship.StewardshipSetpoint · verified · gate-passed
    no DOI yet
  • Thermodynamic Attention
    Viridis.Attention.ThermodynamicAttention · verified · gate-passed
    no DOI yet
  • Universal Waterfilling
    Viridis.Meta.UniversalWaterfilling · verified · gate-passed
    no DOI yet
Working corpus 128 — Correct, comparator-certified, self-classified as not a runtime kernel.
Quarantined 1 — Removed from the verified spine pending repair.
  • AI Safety — Quarantined
    P9_AI_Safety · quarantined · quarantined
    no DOI yet

Scientific review record

Open findings and held sources

Later review can identify limits in an already published paper. A certificate for named formal statements does not validate every paper claim. These findings remain visible and block new kernel qualification.

BCAN · Run-142 · qualification held

The necessary route-length implication following Proposition 2 does not follow from the stated lower bound on directional progress.

Coincident unit directions and one unit segment give projected displacement 1. A declared alignment floor of 1/2 satisfies the assumptions, but the claimed necessary length 1 / (1/2) = 2 exceeds the actual length 1.

The positive-length aggregation inequality itself is not refuted by this example. The retained Lean certificate covers narrower scalar statements; it does not prove this paper implication.

A corrected statement, corresponding numerical control and later independent scientific review are required before new kernel qualification. The published record remains unchanged.

Inspect the published source

Independent review SHA-256: 80cb55ebb079b79b795fae1faa4d4e297228bca097a94dcde93e57555793d89a

RAAC · Run-134 · qualification held

The current claim map labels general optimization and alignment claims as formally verified, while the cited Lean targets establish narrower algebraic identities.

A one-dimensional completing-square identity and two-dimensional polynomial factorizations do not by themselves establish the claimed general multidimensional optimization results.

This finding concerns proof coverage in the claim map. It does not refute the narrower algebraic statements or imply that every analytical paper result is false.

A corrected claim-to-proof map and later independent review are required before new kernel qualification. Original papers and proof artifacts remain unchanged.

Inspect the published source

Independent review SHA-256: a0038bfa09df4fdf50a45ee69f5466d0f1a755e7a85c3d3a7edd7d314dc94401

The research-to-field chain

One claim, one evidence trail, one accountable decision.

01Open research
Viridis Research

Claims that can be inspected

Lean proof objects, explicit assumptions, permanent DOI records, and exact public source. A passed proof means the conclusion follows from the stated hypotheses.

02V1 · public foundation
Viridis OS

Evidence that can be reproduced

The V1 public foundation carries source, method version, uncertainty, limitations, formal backing, and receipt status alongside each output. External adoption remains unproven.

03Controlled beta
Viridis Conservation

Verified Decision Chains

Viridis Conservation connects appropriate research, ecological evidence, economic computation, and human approvals into one accountable chain for a real parcel or project.

Science Foundry → products

Research contributes a method—not automatic product truth.

Each mapping names the product contribution, its current maturity, and the next external gate. Publication or formal proof never activates a customer claim by itself.

Science Foundry

Carbon Continuity after wildfire

Product
Carbon Continuity
Remote Remaining-Asset Screen
Contribution

Supplies the Lean-checked threshold, pool-accounting boundary, and reproducibility requirements used in the first packaged Decision Review.

activeRung 4 · controlled beta

Next gate: Authorized project evidence, decision-owner review, and field or institutional validation appropriate to the claim

Boundary: Formal verification does not establish field coefficients, registry treatment, credit validity, insurance treatment, or commercial value.

Science Foundry

D-Score and biodiversity condition

Product
MEASURE
Ecological condition and priority evidence
Contribution

Supplies bounded ecological-condition, connectivity, and conservation-priority calculations for a Decision Record.

candidateRung 4 · calibration required

Next gate: Partner review, field calibration, and held-out ecological validation for the intended landscape

Boundary: A model score does not establish title, permanence, field condition, legal protection, or registry acceptance.

Science Foundry

HDFM and dendritic corridor mathematics

Product
VERIFY
Working-forest corridor and monitoring evidence
Contribution

Supplies graph structure, corridor calculations, remote-observation seams, and attestation inputs for working-forest review.

candidateRung 4 · field validation required

Next gate: One field-reviewed parcel, live provider evidence, and independent ecological review

Boundary: The idealized corridor result and software checks do not prove ecological performance on a real property.

Science Foundry

Afforestation and forest nucleation

Product
Stewardship
Restoration design candidate
Contribution

Provides candidate design laws for concentrating limited establishment effort and comparing planting geometry.

candidateRung 4 · research candidate

Next gate: Site-specific ecological calibration, field trial, and qualified practitioner review

Boundary: The formal model is not a universal planting prescription and has not been activated as a customer method.

Science Foundry

Native grassland stewardship

Product
Stewardship
Grassland Decision Review module
Contribution

Supplies grassland-specific condition, disturbance, carbon-floor, and woody-encroachment rules.

candidateRung 3 · legal and field review required

Next gate: Local ecology, grazing or fire expertise, counsel review, field evidence, and partner acceptance

Boundary: The covenant design is not a registry-approved method, executed legal instrument, or universal field rule.

Science Foundry

Wu Wei stewardship

Product
Stewardship
Bounded-intervention planning candidate
Contribution

Provides a candidate constraint for keeping intervention within a living system's assimilation and recovery capacity.

backlogRung 3 · research candidate

Next gate: Model-specific calibration, domain review, and monitored field application

Boundary: The principle is not an independently validated universal management law or active customer method.

Science Foundry

Public Canon and reproducible research records

Product
DATA
Public research assistant and evidence exports
Contribution

Supplies digest-checked public research records, DOI and source links, validation state, and bounded evidence exports used by ViridisOS.

activeRung 4 · controlled beta

Next gate: Outside reuse, independent replication, accepted data license, and decision-specific evidence review

Boundary: Publication, downloads, repository traffic, or assistant retrieval do not establish adoption, empirical validation, or decision value.

Science Foundry

Provenance, attestation, and evidence-state design

Product
ViridisOS
Every Verified Decision Chain
Contribution

Binds source, method version, computation, uncertainty, limitations, review state, and reproducibility receipt to the output.

activeRung 4 · public foundation

Next gate: Independent verification, outside integration, repeat organizational use, and accepted delivery evidence

Boundary: A valid receipt establishes integrity and traceability, not empirical truth, institutional acceptance, adoption, or conservation impact.

Science Foundry

EcoChain and conservation-finance architecture

Product
FINANCE
Future legal permanence, aggregation, settlement, and payment path
Contribution

Defines guarded state transitions that could connect verified stewardship evidence to externally authorized finance.

backlogRung 3 · future held

Next gate: Executed legal instrument, qualified holder, live parcel evidence, registry authority, buyer acceptance, settlement, KYC, and payout receipt

Boundary: No public credit issuance, sale, owner payout, recorded conserved acreage, or live conservation-finance network is active.

Public research record

Inspect before you trust.

Every public surface has a distinct job: DOI records preserve releases, GitHub exposes executable source, the canon explorer makes scope and limitations legible, and ORCID anchors the human researcher.

Browse the canon↗Inspect Lean source↗Open Zenodo records↗Verify the researcher↗
From research to one decision

Bring one consequential land decision. We will make its reasoning reproducible.

For a land trust, restoration developer, or forestry adviser with one real parcel or project. Research enters the workflow only when its evidence state and limitations fit the decision.

Request a reviewed analysis →
What you receive
  • ✓ A scoped Verified Decision Chain for one parcel or restoration project
  • ✓ Inputs, assumptions, uncertainty, and provenance kept visible
  • ✓ A reproducible method version and partner review session
  • ✓ A clear list of evidence still needed before operational use

Not included: registry accreditation, credit issuance, legal certification, or a claim that machine verification substitutes for ecological or professional review.

Justin D. Hart, founder and research lead. Research questions, replication results, and institutional partnerships: justin@viridisconservation.com
Start a decision →