What is a Daitum Model?

A Daitum model is a complete, programmatic description of an optimisation application.

It defines what data to use, how the problem is solved, and how results are presented — all written in Python and serialised to JSON for execution on the Daitum platform.

The three parts

A Daitum model is composed of three artefacts, each built with a dedicated library:

Model definition (daitum_model)

Defines the data structures and core logic of the problem.

• Tables — named collections of records (e.g. Jobs, Machines)

• Fields — typed columns within a table (STRING, DECIMAL, DATETIME, …)

• Calculations — derived values at the model level (e.g. total cost)

• Parameters — user-editable scalar inputs (e.g. a budget limit)

UI definition (daitum_ui)

Defines the user interface presented in the platform.

• Views — tables, Gantt charts, forms, cards, maps, and more

• Navigation — navigation structure linking views

• Context variables — shared runtime state (filters, toggles)

Configuration (daitum_configuration)

Defines how the optimisation is executed and data is transformed.

• Algorithm — the solving approach (e.g. Genetic Algorithm, VNS)

• Optimisation model — decision variables, objectives, and constraints

• Data sources — input data mappings, data transforms, report properties

How the parts fit together

The three artefacts share a common object model. Tables and fields defined in daitum_model are referenced directly in the UI and configuration layers.

This allows a single definition (e.g. a field) to be reused across: - UI components (columns, charts) - optimisation logic (decision variables, constraints) - calculations and outputs

daitum_model  ──────────────┐
     │  Tables, Fields,     ├──► Packaged and deployed to Daitum
     │  Calculations        │
     ▼                      │
daitum_ui                   │
     │  Views, Navigation   │
     ▼                      │
daitum_configuration        │
        Algorithm, Data ────┘

Deployment

Each artefact is serialised to a Python dict and written to JSON. These files are bundled into a ZIP archive and uploaded to the Daitum platform.

See the Quickstart for a complete end-to-end example.