e01zmc generates a multidimensional interpolant to a set of scattered data points, using a modified Shepard method. When the number of dimensions is no more than five, there are corresponding functions in Chapter E01 which are specific to the given dimensionality. e01sgc generates the two-dimensional interpolant, while e01tgc, e01tkc and e01tmc generate the three-, four- and five-dimensional interpolants respectively.

2 Specification

#include <nag.h>

void	e01zmc (Integer d, Integer m, const double x[], const double f[], Integer nw, Integer nq, Integer iq[], double rq[], NagError *fail)

The function may be called by the names: e01zmc, nag_interp_dimn_scat_shep or nag_nd_shep_interp.

3 Description

e01zmc constructs a smooth function

Q (x)

x \in ℝ^{d}

which interpolates a set of

m

scattered data points

(x_{r}, f_{r})

, for

r = 1, 2, \dots, 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

Q (x) = \frac{\sum_{r = 1}^{m} w_{r} (x) q_{r}}{\sum_{r = 1}^{m} w_{r} (x)},

where

q_{r} = f_{r}

w_{r} (x) = \frac{1}{{‖x - x_{r}‖}_{2}^{2}}

The basic method is global in that the interpolated value at any point depends on all the data, but e01zmc uses a modification (see Franke and Nielson (1980) and Renka (1988a)), whereby the method becomes local by adjusting each

w_{r} (x)

to be zero outside a hypersphere with centre

x_{r}

and some radius

R_{w}

. Also, to improve the performance of the basic method, each

q_{r}

above is replaced by a function

q_{r} (x)

, which is a quadratic fitted by weighted least squares to data local to

x_{r}

and forced to interpolate

(x_{r}, f_{r})

. In this context, a point

x

is defined to be local to another point if it lies within some distance

R_{q}

of it.

The efficiency of e01zmc is enhanced by using a cell method for nearest neighbour searching due to Bentley and Friedman (1979) with a cell density of

3

The radii

R_{w}

and

R_{q}

are chosen to be just large enough to include

N_{w}

and

N_{q}

data points, respectively, for user-supplied constants

N_{w}

and

N_{q}

. Default values of these parameters are provided, and advice on alternatives is given in Section 9.2.

e01zmc is derived from the new implementation of QSHEP3 described by Renka (1988b). It uses the modification for high-dimensional interpolation described by Berry and Minser (1999).

Values of the interpolant

Q (x)

generated by e01zmc, and its first partial derivatives, can subsequently be evaluated for points in the domain of the data by a call to e01znc.

4 References

Bentley J L and Friedman J H (1979) Data structures for range searching ACM Comput. Surv. 11 397–409

Berry M W, Minser K S (1999) Algorithm 798: high-dimensional interpolation using the modified Shepard method ACM Trans. Math. Software 25 353–366

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 Arguments

1: $d$ – Integer Input: On entry: $d$ , the number of dimensions.

Constraint: $d \geq 2$ .
2: $m$ – Integer Input: On entry: $m$ , the number of data points.
Note: on the basis of experimental results reported in Berry and Minser (1999), when $d \geq 5$ it is recommended to use $m \geq 4000$ .

Constraint: $m \geq (d + 1) \times (d + 2) / 2 + 2$ .
3: $x [d \times m]$ – const double Input: Note: the $(i, j)$ th element of the matrix $X$ is stored in $x [(j - 1) \times d + i - 1]$ .

On entry: the d components of the first data point must be stored in elements $0, 1, \dots, d - 1$ of x. The second data point must be stored in elements $d, d + 1, \dots, 2 \times d - 1$ of x, and so on. In general, the m data points must be stored in $x [i \times d + j]$ , for $i = 0, 1, \dots, m - 1$ and $j = 0, 1, \dots, d - 1$ .

Constraint: these coordinates must be distinct, and must not all lie on the same $(d - 1)$ -dimensional hypersurface.
4: $f [m]$ – const double Input: On entry: $f [r - 1]$ must be set to the data value $f_{r}$ , for $r = 1, 2, \dots, m$ .
5: $nw$ – Integer Input: On entry: the number $N_{w}$ of data points that determines each radius of influence $R_{w}$ , appearing in the definition of each of the weights $w_{r}$ , for $r = 1, 2, \dots, m$ (see Section 3). Note that $R_{w}$ is different for each weight. If $nw \leq 0$ the default value $nw = \min (2 \times (d + 1) \times (d + 2), m - 1)$ is used instead.

Suggested value: $nw = - 1$ .

Constraint: $nw \leq m - 1$ .
6: $nq$ – Integer Input: On entry: the number $N_{q}$ of data points to be used in the least squares fit for coefficients defining the quadratic functions $q_{r} (x)$ (see Section 3). If $nq \leq 0$ the default value $nq = \min ((d + 1) \times (d + 2) \times 6 / 5, m - 1)$ is used instead.

Suggested value: $nq = - 1$ .

Constraint: $nq \leq 0$ or $(d + 1) \times (d + 2) / 2 - 1 \leq nq \leq m - 1$ .
7: $iq [2 \times m + 1]$ – Integer Output: On exit: integer data defining the interpolant $Q (x)$ .
8: $rq [\dim]$ – double Output: Note: the dimension, dim, of the array rq must be at least $((d + 1) \times (d + 2) / 2) \times m + 2 \times d + 1$ .

On exit: real data defining the interpolant $Q (x)$ .
9: $fail$ – NagError * Input/Output: The NAG error argument (see Section 7 in the Introduction to the NAG Library CL Interface).

6 Error Indicators and Warnings

NE_ALLOC_FAIL: Dynamic memory allocation failed.
See Section 3.1.2 in the Introduction to the NAG Library CL Interface for further information.
NE_BAD_PARAM: On entry, argument $〈value〉$ had an illegal value.
NE_DATA_HYPERSURFACE: On entry, all the data points lie on the same hypersurface. No unique solution exists.
NE_DUPLICATE_NODE: There are duplicate nodes in the dataset. $x [(i - 1) \times d + k - 1] = x [(j - 1) \times d + k - 1]$ , for $i = 〈value〉$ , $j = 〈value〉$ and $k = 1, 2, \dots, d$ . The interpolant cannot be derived.
NE_INT: On entry, $d = 〈value〉$ .
Constraint: $d \geq 2$ .
NE_INT_2: On entry, $((d + 1) \times (d + 2) / 2) \times m + 2 \times d + 1$ exceeds the largest machine integer.
$d = 〈value〉$ and $m = 〈value〉$ .

On entry, $m = 〈value〉$ and $d = 〈value〉$ .
Constraint: $m \geq (d + 1) \times (d + 2) / 2 + 2$ .

On entry, $nq = 〈value〉$ and $d = 〈value〉$ .
Constraint: $nq \leq 0$ or $nq \geq (d + 1) \times (d + 2) / 2 - 1$ .

On entry, $nq = 〈value〉$ and $m = 〈value〉$ .
Constraint: $nq \leq m - 1$ .

On entry, $nw = 〈value〉$ and $m = 〈value〉$ .
Constraint: $nw \leq m - 1$ .
NE_INTERNAL_ERROR: An internal error has occurred in this function. Check the function call and any array sizes. If the call is correct then please contact NAG for assistance.
See Section 7.5 in the Introduction to the NAG Library CL Interface for further information.
NE_NO_LICENCE: Your licence key may have expired or may not have been installed correctly.
See Section 8 in the Introduction to the NAG Library CL Interface for further information.

7 Accuracy

In experiments undertaken by Berry and Minser (1999), the accuracies obtained for a conditional function resulting in sharp functional transitions were of the order of

10^{- 1}

at best. In other cases in these experiments, the function generated interpolates the input data with maximum absolute error of the order of

10^{- 2}

8 Parallelism and Performance

e01zmc is threaded by NAG for parallel execution in multithreaded implementations of the NAG Library.

e01zmc makes calls to BLAS and/or LAPACK routines, which may be threaded within the vendor library used by this implementation. Consult the documentation for the vendor library for further information.

Please consult the X06 Chapter Introduction for information on how to control and interrogate the OpenMP environment used within this function. Please also consult the Users' Note for your implementation for any additional implementation-specific information.

9 Further Comments

9.1 Timing

The time taken for a call to e01zmc will depend in general on the distribution of the data points and on the choice of

N_{w}

and

N_{q}

parameters. If the data points are uniformly randomly distributed, then the time taken should be

O (m)

. At worst

O (m^{2})

time will be required.

9.2 Choice of $N_{w}$ and $N_{q}$

Default values of the parameters

N_{w}

and

N_{q}

may be selected by calling e01zmc with

nw \leq 0

and

nq \leq 0

. These default values may well be satisfactory for many applications.

If non-default values are required they must be supplied to e01zmc through positive values of nw and nq. Increasing these argument values 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 = \min (2 \times (d + 1) \times (d + 2), m - 1)

and

nq = \min ((d + 1) \times (d + 2) \times 6 / 5, m - 1)

have been chosen on the basis of experimental results reported in Renka (1988a) and Berry and Minser (1999). For further advice on the choice of these arguments see Renka (1988a) and Berry and Minser (1999).

10 Example

This program reads in a set of

30

data points and calls e01zmc to construct an interpolating function

Q (x)

. It then calls e01znc 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 very much larger.

e01zm: FL CL CPP AD

NAG CL Interfacee01zmc (dimn_​scat_​shep)

▸▿ Contents

1 Purpose

2 Specification

3 Description

4 References

5 Arguments

6 Error Indicators and Warnings

7 Accuracy

8 Parallelism and Performance

9 Further Comments

9.1 Timing

9.2 Choice of Nw and Nq

10 Example

10.1 Program Text

10.2 Program Data

10.3 Program Results

NAG CL Interface
e01zmc (dimn_scat_shep)

9.2 Choice of $N_{w}$ and $N_{q}$