### 16 Simple Statistics

- PDA_NSCOR

Calculates the approximate expected values of normal order statistics.
- PDA_V11

Calculates an approximation to the variance of the largest normal order statistic.
- PDA_COVMAT

Approximates the covariance matrix of normal order statistics.
- PDA_DCOV

Calculates the covariance matrix for a nonlinear data fitting problem

#### 16.1 Replacing calls to G01DBF

The routine PDA_NSCOR calculates the same values as G01DBF, except that only N2 values are
returned, rather than N. This is since the values are symmetric and can therefore be simply
derived.

#### 16.2 Replacing calls to G01DCF

The routine PDA_COVMAT calculates the same statistics as G01DCF. To use it requires that you also
supply the variance of the largest normal order statistic (see PDA_V11) and that you increase the
space required for the variance array. This is now a full array of values rather than a packed array.
The following code shows how to convert this to the same form as output by the NAG
routine:

K = 1

DO 3 J = 1, N

DO 4 L = 1, J

NAGVEC( K ) = V( L, J )

K = K + 1

4 CONTINUE

3 CONTINUE