ExtremumLocation Property |
Namespace: Meta.Numerics.Analysis
Note that numerical methods for finding typical a maximum or minimum cannot usually determine its location to full floating-point precision. Near a quadratic optimum, a change in x of ~ε will change f(x) by ~ε2. Thus the smallest detectable change in f(x) will typically correspond to a change in x of order of the square root of full precision. Full Double precision being ~16 digits, you should expect the location to be accurate only to ~8 digits.