Uncertainty

How Sounio represents and (eventually) propagates measurement uncertainty.

Uncertainty

Uncertainty is central to the Sounio story, but good docs need to separate the mathematical model from the exact surface the current artifact enforces. This page does both: it documents what the model is trying to express and what you can prove with the current fixtures.

What is verified now

  • Confidence-sensitive refusal is not hypothetical; it is exercised by current compile-fail fixtures.
  • The checked contract already demonstrates that weak evidence and low-confidence states can block a program before runtime.
  • The richer propagation model remains visible in research and reference docs even when a particular artifact exposes only part of it.

Model-level propagation rules worth knowing

  • For independent addition and subtraction, standard uncertainty is typically combined in quadrature.
  • For multiplication and division, relative uncertainty propagation is the more useful mental model.
  • Confidence and provenance are part of the decision boundary, not just the numeric payload.

Where to prove behavior in this repo

  • tests/run-pass/vancomycin_propagation.sio is the easiest current propagation example to validate with souc check.
  • tests/compile-fail/vancomycin_low_conf.sio proves the refusal path remains live.
  • docs/research/vancomycin-uncertainty.md explains the clinical case study behind these fixtures.

Validation commands

export SOUC_BIN="$(pwd)/artifacts/omega/souc-bin/souc-linux-x86_64-jit"
export SOUNIO_STDLIB_PATH="$(pwd)/stdlib"

"$SOUC_BIN" check tests/run-pass/vancomycin_propagation.sio
"$SOUC_BIN" check tests/compile-fail/vancomycin_low_conf.sio