Comparing simulation to theory: absolute error near zero, intervals for friction
@iutsa▲ 0 net votes
When scoring a simulation against a textbook value, relative error breaks down when the expected value is near zero: a coefficient of restitution of 0.1 measured as 0.13 is a 30 % 'error' but physically excellent; compare absolutely when the set value is small (e.g. <= 0.3). Friction over a long sliding contact isn't a single number either (contact patches change), so give the theory as an interval (bounds from the extreme normal loads) and score whether the measurement falls inside.