Uncertainty Quantification
Uncertainty quantification tracks how imperfect inputs and models translate into uncertainty in predictions, turning point estimates into honest ranges.
Every number has a spread
No input to a simulation is known exactly, and no model is perfect. Uncertainty quantification (UQ) is the discipline of propagating those imperfections through a computation so the output carries a defensible error range instead of false precision.
Two kinds of uncertainty
- Aleatoric: inherent randomness that cannot be reduced by more knowledge.
- Epistemic: uncertainty from incomplete knowledge, which better data or models can reduce.
How it is done
A common approach is to sample the uncertain inputs from their distributions, run the model on each sample, and study the spread of outputs. This forward propagation reveals which outputs are robust and which are fragile. More efficient methods exist, but the logic is the same: map input uncertainty to output uncertainty.
Why it matters in design
A design that meets a target only for perfectly known inputs is not a design; it is a coincidence. Stating that a breeding ratio, a Q value, or a peak field holds within a quantified range is what makes a predicted number engineering-grade rather than aspirational.
Honesty as output
UQ forces a program to state what it does not know. Reporting a prediction with its uncertainty — and the assumptions behind that uncertainty — is more credible than a bare figure, and it is the only responsible way to present simulation results that will inform real construction decisions.