Sensitivity Analysis
Sensitivity analysis measures how much each input matters to an output, focusing effort on the variables that actually drive results.
Which knobs matter
A model may have dozens of inputs, but usually only a few strongly influence any given output. Sensitivity analysis identifies which inputs move the answer most, so that measurement, refinement, and control effort go where they count.
Local versus global
- Local sensitivity: how the output responds to a small change in one input near a chosen point.
- Global sensitivity: how much each input contributes to output variation across its whole range, including interactions.
A simple local measure
The partial derivative of an output with respect to an input, evaluated at an operating point, is the most basic sensitivity. Scaling it by the input’s typical size gives a dimensionless measure that lets you rank very different variables on the same footing.
Why it precedes optimization
Before searching a design space, it helps to know its shape. Sensitivity analysis reveals which parameters are worth optimizing and which can be fixed, shrinking the search and clarifying the trade-offs. It also exposes dangerous cliffs where a small input change causes a large output swing.
In practice
For a machine like the breeder Hyperion, understanding whether performance is more sensitive to magnetic field, plasma current, or profile shape guides both design priorities and the diagnostics needed in operation. Sensitivity analysis and uncertainty quantification are natural partners: one ranks the drivers, the other bounds their effect.