UTE aims to support scientific and operational theories. That requires a numeric layer: quantities with units, equations, and
models such as kinetics, dose-response, and differential/difference dynamics.
What exists today
The current system can express some numeric constraints via DSL operators and can solve CSP-style problems, but it does not have a
dedicated numeric modeling layer (units, dimensional analysis, solver integration, estimation workflows).
UTE requirement: numeric modeling is not “just arithmetic”
Units and dimensions
Quantities need units (mg, mL, mol/L) and dimensional checks to prevent silent errors.
Model execution
Some queries require evaluating models (equations, kinetics) as part of inference, with evidence/provenance.
Parameter estimation
Fit parameters to measurements (or constrain them) and keep a trace of assumptions and data used.
Proposed extensions (research direction)
- Quantity type (value + unit) and a small unit algebra.
- Model operators that call numeric evaluators but still yield explainable outputs.
- Solver adapters (CSP/optimization/ODE) behind a deterministic interface with trace objects.
Spec
This page is summarized and formalized in DS37.