The record formats¶
Every record Mokili stores, as JSON Schema (draft 2020-12). They are published under the Apache License 2.0: any tool may read and write these records.
A record says which format it follows in its schema field. What a version
number promises is in Record versions.
| Schema | Records | |
|---|---|---|
| calibration.schema.json | mokili.calibration/1 |
Parameter values fitted to a measurement, with their uncertainty. |
| campaign-run.schema.json | mokili.campaign-run/1 |
What a campaign found: one cell per variant and fault. |
| campaign.schema.json | mokili.campaign/1 |
Requirements checked across controller variants and faults: every variant meets every fault, from one base scenario. |
| choice.schema.json | mokili.choice/1 |
A person's choice after a decision's runs: the option taken, and why. |
| comparison-evidence.schema.json | mokili.comparison-evidence/1 |
A comparison and the evidence bundle of every run it compares. |
| comparison.schema.json | mokili.comparison/1 |
Runs compared to a baseline run, with the method and its assumptions stated. |
| credibility.schema.json | mokili.credibility/1 |
The case that a model can be trusted for one use: the question, how the model answers it, what is at stake, and the evidence gathered. |
| decision-run.schema.json | mokili.decision-run/1 |
What a decision's runs found. |
| decision.schema.json | mokili.decision/1 |
A choice between options, made on simulations: each option runs on the same draws of what is uncertain, so they are compared draw by draw. |
| evidence.schema.json | mokili.evidence/1 |
Everything a reviewer needs to read or rerun one run, in one file. |
| measurement.schema.json | mokili.measurement/1 |
Observations of a real system, from a CSV file stored by digest. |
| model.schema.json | mokili.model/1 |
|
| proposal.schema.json | mokili.proposal/1 |
A change someone suggests to a model, a scenario or a sweep. |
| requirement.schema.json | mokili.requirement/1 |
A requirement a run can be checked against: the sentence people agreed on, and the property that tests it. |
| result.schema.json | mokili.result/1 |
What the engine produced: simulated values from a simulation engine, or inferred conclusions from a qualitative analysis. |
| review.schema.json | mokili.review/1 |
A person's decision on a proposal. |
| run.schema.json | mokili.run/1 |
How a result came to exist, or why it does not. |
| scenario.schema.json | mokili.scenario/1 |
One experiment input, fixed: which model, what changes, for how long, with which seed, and which outputs to keep. |
| scene.schema.json | mokili.scene/1 |
A schematic picture of a system, driven by a run's series. |
| study-run.schema.json | mokili.study-run/1 |
What a study found. |
| study.schema.json | mokili.study/1 |
What an uncertain or finer run of a scenario changes, for the outputs and requirements that matter. |
| sweep.schema.json | mokili.sweep/1 |
A base scenario and its variations, run and compared to the base. |