smoothing parameter 0 to be > 0 renders the

specified by the user; the value of this parameter

reflects the trade-off between goodness of fit to the

computational problem more complex.

data, measured by the first term, and smoothness,

as measured by *Q*. If 0 is chosen to be 0, the

spline is an exact interpolator and passes through

optimization problem (Equation 7-7) may also be

all the data points. If 0 > 0, the spline is not an

obtained by a kriging algorithm if the smoothing

parameter 0 is taken to be equal to the variance of

exact interpolator. (Splines that are not exact

interpolators are referred to as smoothing splines.)

measurement error and if a special form is chosen

There are a number of numerical procedures that

for the covariance function. Therefore, in this

may be used for fitting splines, but allowing the

situation, spline approximation is a special type of

7-4

Integrated Publishing, Inc. |