A research model that keeps its evidence in view.
Research does not have one canonical workflow. Grid's role is to keep sources, transformations, methods, results, narrative, and review state connected—while specialist tools, institutional controls, and scholarly judgment remain explicit. The result is a computational medium for the work between raw evidence and a reviewable claim.
Make the work behind the finding inspectable.
A practical case for explicit assumptions, inspectable analysis, and repeatable computation. Explore the methodological research, Grid’s academic feature set, and worked examples of uncertainty, experiment comparisons, and preserved results.
Covers the conditions for repeatability, Grid Pro research workflows, and clearly identified proposals for extending them.
Exact work, as source.
symbol x poly = A1*x^2 + 1 # exact, not floating-point-lucky derivative_result = SYMBOLIC.D(poly, x) slope = derivative_result.expression roots_result = SYMBOLIC.SOLVE(poly == 0, x) decision_result = SYMBOLIC.DECIDE(x^2 >= 0) # numerics beside the algebra traj = ODE_SOLVE(LAMBDA(t, y, -0.5*y), 1, 0, 10, 101)
Exact mathematics is one native research path, not the whole research story. Here, A1 is a reactive input and assumptions gate rewrites: x/x only carries a nonzero condition until x ≠ 0 is supported.
Symbolic mathematics
Canonicalize, simplify, factor, differentiate, solve exactly with certified root intervals, and approximate at explicit precision within declared limits — with a replayable certificate per step.
Symbolic mathematicsScientific computing
ODE and PDE solvers with automatic stiffness handling, spectral linear algebra, FFTs, curve fitting, splines — reactive, in cells.
The function catalogEvidence you can question
Structured predicates and WHY paths make review state and support chains queryable, while leaving the model's facts and decisions under explicit human ownership.
Predicates and evidenceDifferent evidence. One reviewable medium.
Each path starts from the workflow researchers actually have, defines a bounded role for Grid, preserves the right authorities, and names what Grid does not yet claim.
Experimental and quantitative research
Keep observations, cleaning decisions, models, figures, and claims in one inspectable dependency chain without displacing the systems that acquired or govern the data.
- Biology
- Chemistry
- Psychology
- Economics
- Kinesiology
- Communication
- Typical current stack
- Instrument exports or survey platforms → ELN/LIMS or secure storage → R, Python, Stata, or a notebook → figures and manuscript.
- Where Grid can fit
- Typed formulas, relational transformations, reactive dependencies, and evidence relations can connect approved inputs, analysis outputs, tables, figures, and the claims that depend on them.
Artifact chain
- Protocol
- Frozen observations
- Cleaning + QA
- Analysis model
- Tables + figures
- Claims
- Review package
- Authority retained
- Acquisition instruments, ELN/LIMS, approved source repositories, validated statistical packages, and the publication or archival repository remain systems of record.
Review boundary
Researchers and reviewers still own study design, exclusions, stopping rules, model choice, uncertainty, and interpretation. Grid can expose those decisions; it cannot certify that they are scientifically warranted.
Capability evidence
Documented primitivesCurrent public docs cover typed formulas, relational transformations, statistical functions, reactive dependencies, and queryable explanation paths.
Not claimed · product gap
Not yet established publicly: validated instrument and LIMS adapters, preregistered-analysis locking, discipline-specific method packs, and publication-grade figure/report workflows.
Simulation, engineering, and optimization
Make assumptions, units, calibrated parameters, scenarios, solves, and decision rationale travel together while specialist engines keep doing the physics they are trusted to do.
- Aerospace engineering
- Civil engineering
- Operations research
- Geosciences
- Physics
- Planning
- Typical current stack
- MATLAB or Python → CAD/CAE and domain simulators → solver scripts → CSV handoffs → sensitivity workbook → design or policy report.
- Where Grid can fit
- Grid can act as the scenario and decision layer: unit-checked inputs, bounded resident simulation, optimization, calibration, sensitivity, and recorded run identities in the same model.
Artifact chain
- Requirements
- Equations + units
- Calibration
- Scenario matrix
- Simulation + solve
- Sensitivity
- Decision package
- Authority retained
- Validated CAE and domain simulators, solver binaries, calibration datasets, configuration control, and safety-case or design-record systems remain authoritative.
Review boundary
Domain experts must validate equations, boundary conditions, meshes, convergence, solver settings, feasibility, and safety margins. A replayable run is evidence of what ran, not proof that the model is valid.
Capability evidence
Documented primitivesThe public simulation surface records stable run IDs and result digests and documents parameter matrices, calibration, seeded Monte Carlo, ODE/PDE functions, LP primitives, and MIP in solver-enabled builds.
Not claimed · product gap
Not yet established publicly: production co-simulation adapters, complete solver-version and setting capture, discipline benchmark suites, and certification-ready templates.
Mixed-methods and human-subject research
Join quantitative results and coded qualitative evidence at an approved level of detail while keeping consent, access, and release authority outside the analytical model.
- Education
- Clinical psychology
- Public health
- Sociology
- Social work
- Linguistics
- Typical current stack
- IRB and consent records → REDCap or Qualtrics → controlled storage → R, Stata, or SPSS plus NVivo or Dedoose → manuscript.
- Where Grid can fit
- Over already de-identified, approved inputs, relational analysis and explicit predicates can make declared transformations, review state, quantitative outputs, coded themes, and triangulated claims inspectable.
Artifact chain
- Approved protocol
- Consented collection
- External de-identification
- Quant + qual analysis
- Triangulation
- Disclosure review
- Release
- Authority retained
- The IRB protocol, consent and withdrawal system, identity provider, secure source repository, and qualitative source system remain authoritative for permission and access.
Review boundary
Grid must not infer consent, secondary-use permission, re-identification risk, or disclosure safety. Those decisions belong to participants, investigators, and institutional governance processes.
Capability evidence
Composable primitivesTyped predicates can represent explicit review facts and evidence paths, while relational and statistical primitives can support approved analysis. Grid can only evaluate facts explicitly supplied; it does not establish permission or consent.
Not claimed · product gap
Not yet established publicly: consent and withdrawal propagation, field- and row-level research policy enforcement, privacy-preserving release checks, and governed REDCap/qualitative-tool connectors.
Qualitative coding and participatory research
Turn the path from source material to code, memo, theme, counterexample, participant review, and claim into a visible source-to-claim chain—not an opaque AI summary.
- Anthropology
- Ethnography
- Communication studies
- Education
- Sociology
- Community research
- Typical current stack
- Recordings and field notes → transcripts → NVivo, ATLAS.ti, MAXQDA, or Dedoose → memo documents → codebook spreadsheets → collaborative review.
- Where Grid can fit
- Structured predicates and graph relations can support an analytic matrix linking coded excerpts, codebook versions, memos, themes, negative cases, review state, and supported claims.
Artifact chain
- Source record
- Transcript + context
- Versioned codebook
- Coded excerpts
- Memos + negative cases
- Participant review
- Claim appendix
- Authority retained
- Master recordings and transcripts, consent or community agreements, controlled qualitative repositories, and participant or researcher interpretations remain authoritative.
Review boundary
AI-proposed codes or summaries must remain suggestions. People decide what a code means, whether a theme is warranted, how positionality affects interpretation, and whether member review or community authority changes a conclusion.
Capability evidence
Composable primitivesStructured many-valued predicates, reactive relation graphs, and WHY paths provide useful primitives for code-to-claim traceability; they are not yet a complete qualitative analysis environment.
Not claimed · product gap
Not yet established publicly: span-level audio/video annotation, codebook branching and reconciliation, redaction tooling, participant-controlled review, and standard qualitative exports.
Humanities, archives, editions, and media
Keep catalog records, surrogates, transcriptions, encodings, variants, transformations, annotations, and interpretive claims connected without flattening them into a dataset.
- History
- Literature
- Classics
- Art history
- Religious studies
- Media studies
- Typical current stack
- Archive or catalog → IIIF images and OCR/HTR → TEI/XML or annotations → scripts, GIS, and reference manager → edition, article, or exhibit.
- Where Grid can fit
- Custom grammars, structured predicates, and graphs can underpin a research ledger joining identifiers, source passages, editorial states, variants, transformations, networks, chronologies, and claims.
Artifact chain
- Catalogued object
- Image or surrogate
- Transcription/OCR
- Encoded text
- Variants + annotations
- Analysis + claim
- Edition or exhibit
- Authority retained
- Archive catalogs and persistent identifiers, source surrogates, rights records, authoritative transcriptions, editorial policy, and preservation or publishing repositories remain authoritative.
Review boundary
Scholars, archives, and source communities retain authority over provenance, authenticity, attribution, culturally sensitive access, editorial emendation, and interpretation.
Capability evidence
Composable primitivesPublic docs establish custom grammars and parsers, typed predicates, graph traversal, and dependency explanations—the substrate for an edition, not a finished scholarly-edition product.
Not claimed · product gap
Not yet established publicly: IIIF, TEI, METS/MODS, and citation-manager adapters; facsimile/text synchronization; critical-apparatus UI; rights policy; and preservation-grade export.
Field, spatial, ecological, and observational research
Carry sampling design, raw observations, calibration and QA, coordinates, time, specimens, spatial derivations, models, maps, and claims as one reviewable chain.
- Ecology
- Geography
- Geology
- Archaeology
- Urban planning
- Environmental engineering
- Typical current stack
- Field forms, sensors, GPS, or specimens → collections database and file store → QGIS or ArcGIS → R/Python and remote-sensing tools → maps and paper.
- Where Grid can fit
- The field-to-claim model: sampling units, typed observations, calibration and QA flags, coordinate/time normalization, spatial joins, trajectories, scenarios, outputs, and rationale.
Artifact chain
- Sampling design
- Raw field record
- Calibration + QA
- Spatial derivation
- Model or scenario
- Map + table + claim
- Archive handoff
- Authority retained
- Raw device files, specimen and collections records, trusted CRS/projection libraries, permits, chain-of-custody records, and archival repositories remain authoritative.
Review boundary
Experts must assess sampling validity, taxonomy, calibration, coordinate reference systems, detection bias, missingness, ground truth, permit constraints, and protection of sensitive locations.
Capability evidence
Documented primitivesThe public function catalog includes CRS-preserving geometry, reprojection, spatial joins, H3 aggregation, routing, and trajectory operations; simulation also documents spatial-agent playback.
Not claimed · product gap
Not yet established publicly: offline field capture and conflict-safe sync, raster/vector workflow breadth, sensor and specimen connectors, calibration provenance, and sensitive-location masking.