E01TGF (PDF version)
E01 Chapter Contents
E01 Chapter Introduction
NAG Library Manual

NAG Library Routine Document


Note:  before using this routine, please read the Users' Note for your implementation to check the interpretation of bold italicised terms and other implementation-dependent details.

+ Contents

    1  Purpose
    7  Accuracy

1  Purpose

E01TGF generates a three-dimensional interpolant to a set of scattered data points, using a modified Shepard method.

2  Specification

REAL (KIND=nag_wp)  X(M), Y(M), Z(M), F(M), RQ(LRQ)

3  Description

E01TGF constructs a smooth function Qx,y,z which interpolates a set of m scattered data points xr,yr,zr,fr, for r=1,2,,m, using a modification of Shepard's method. The surface is continuous and has continuous first partial derivatives.
The basic Shepard method, which is a generalization of the two-dimensional method described in Shepard (1968), interpolates the input data with the weighted mean
Qx,y,z=r=1mwrx,y,zqr r=1mwrx,y,z ,
qr=fr ​ and ​ wrx,y,z= 1dr2 ​ and ​ dr2= x-xr 2+ y-yr 2+ z-zr 2.
The basic method is global in that the interpolated value at any point depends on all the data, but this routine uses a modification (see Franke and Nielson (1980) and Renka (1988a)), whereby the method becomes local by adjusting each wrx,y,z to be zero outside a sphere with centre xr,yr,zr and some radius Rw. Also, to improve the performance of the basic method, each qr above is replaced by a function qrx,y,z, which is a quadratic fitted by weighted least squares to data local to xr,yr,zr and forced to interpolate xr,yr,zr,fr. In this context, a point x,y,z is defined to be local to another point if it lies within some distance Rq of it. Computation of these quadratics constitutes the main work done by this routine.
The efficiency of the routine is further enhanced by using a cell method for nearest neighbour searching due to Bentley and Friedman (1979).
The radii Rw and Rq are chosen to be just large enough to include Nw and Nq data points, respectively, for user-supplied constants Nw and Nq. Default values of these parameters are provided by the routine, and advice on alternatives is given in Section 8.2.
This routine is derived from the routine QSHEP3 described by Renka (1988b).
Values of the interpolant Qx,y,z generated by this routine, and its first partial derivatives, can subsequently be evaluated for points in the domain of the data by a call to E01THF.

4  References

Bentley J L and Friedman J H (1979) Data structures for range searching ACM Comput. Surv. 11 397–409
Franke R and Nielson G (1980) Smooth interpolation of large sets of scattered data Internat. J. Num. Methods Engrg. 15 1691–1704
Renka R J (1988a) Multivariate interpolation of large sets of scattered data ACM Trans. Math. Software 14 139–148
Renka R J (1988b) Algorithm 661: QSHEP3D: Quadratic Shepard method for trivariate interpolation of scattered data ACM Trans. Math. Software 14 151–152
Shepard D (1968) A two-dimensional interpolation function for irregularly spaced data Proc. 23rd Nat. Conf. ACM 517–523 Brandon/Systems Press Inc., Princeton

5  Parameters

1:     M – INTEGERInput
On entry: m, the number of data points.
Constraint: M10.
2:     X(M) – REAL (KIND=nag_wp) arrayInput
3:     Y(M) – REAL (KIND=nag_wp) arrayInput
4:     Z(M) – REAL (KIND=nag_wp) arrayInput
On entry: Xr, Yr, Zr must be set to the Cartesian coordinates of the data point xr,yr,zr, for r=1,2,,m.
Constraint: these coordinates must be distinct, and must not all be coplanar.
5:     F(M) – REAL (KIND=nag_wp) arrayInput
On entry: Fr must be set to the data value fr, for r=1,2,,m.
6:     NW – INTEGERInput
On entry: the number Nw of data points that determines each radius of influence Rw, appearing in the definition of each of the weights wr, for r=1,2,,m (see Section 3). Note that Rw is different for each weight. If NW0 the default value NW=min32,M-1 is used instead.
Constraint: NWmin40,M-1.
7:     NQ – INTEGERInput
On entry: the number Nq of data points to be used in the least squares fit for coefficients defining the nodal functions qrx,y,z (see Section 3). If NQ0 the default value NQ=min17,M-1 is used instead.
Constraint: NQ0 or 9NQmin40,M-1.
8:     IQ(LIQ) – INTEGER arrayOutput
On exit: integer data defining the interpolant Qx,y,z.
9:     LIQ – INTEGERInput
On entry: the dimension of the array IQ as declared in the (sub)program from which E01TGF is called.
Constraint: LIQ2×M+1.
10:   RQ(LRQ) – REAL (KIND=nag_wp) arrayOutput
On exit: real data defining the interpolant Qx,y,z.
11:   LRQ – INTEGERInput
On entry: the dimension of the array RQ as declared in the (sub)program from which E01TGF is called.
Constraint: LRQ10×M+7.
12:   IFAIL – INTEGERInput/Output
On entry: IFAIL must be set to 0, -1​ or ​1. If you are unfamiliar with this parameter you should refer to Section 3.3 in the Essential Introduction for details.
For environments where it might be inappropriate to halt program execution when an error is detected, the value -1​ or ​1 is recommended. If the output of error messages is undesirable, then the value 1 is recommended. Otherwise, if you are not familiar with this parameter, the recommended value is 0. When the value -1​ or ​1 is used it is essential to test the value of IFAIL on exit.
On exit: IFAIL=0 unless the routine detects an error or a warning has been flagged (see Section 6).

6  Error Indicators and Warnings

If on entry IFAIL=0 or -1, explanatory error messages are output on the current error message unit (as defined by X04AAF).
Errors or warnings detected by the routine:
On entry,M<10,
or NQ > min40,M-1 ,
or NW > min40,M-1 ,
On entry,Xi,Yi,Zi=Xj,Yj,Zj for some ij.
On entry,all the data points are coplanar. No unique solution exists.

7  Accuracy

On successful exit, the function generated interpolates the input data exactly and has quadratic accuracy.

8  Further Comments

8.1  Timing

The time taken for a call to E01TGF will depend in general on the distribution of the data points. If X, Y and Z are uniformly randomly distributed, then the time taken should be O(M). At worst OM2 time will be required.

8.2  Choice of Nw and Nq

Default values of the parameters Nw and Nq may be selected by calling E01TGF with NW0 and NQ0. These default values may well be satisfactory for many applications.
If nondefault values are required they must be supplied to E01TGF through positive values of NW and NQ. Increasing these parameters makes the method less local. This may increase the accuracy of the resulting interpolant at the expense of increased computational cost. The default values NW = min32,M-1  and NQ = min17,M-1  have been chosen on the basis of experimental results reported in Renka (1988a). In these experiments the error norm was found to vary smoothly with Nw and Nq, generally increasing monotonically and slowly with distance from the optimal pair. The method is not therefore thought to be particularly sensitive to the parameter values. For further advice on the choice of these parameters see Renka (1988a).

9  Example

This program reads in a set of 30 data points and calls E01TGF to construct an interpolating function Qx,y,z. It then calls E01THF to evaluate the interpolant at a set of points.
Note that this example is not typical of a realistic problem: the number of data points would normally be larger.

9.1  Program Text

Program Text (e01tgfe.f90)

9.2  Program Data

Program Data (e01tgfe.d)

9.3  Program Results

Program Results (e01tgfe.r)

E01TGF (PDF version)
E01 Chapter Contents
E01 Chapter Introduction
NAG Library Manual

© The Numerical Algorithms Group Ltd, Oxford, UK. 2012