Free cookie consent management tool by TermsFeed Policy Generator

source: branches/2789_MathNetNumerics-Exploration/HeuristicLab.Algorithms.DataAnalysis.Experimental/642_part2.f @ 16351

Last change on this file since 16351 was 15443, checked in by gkronber, 7 years ago

#2789 added interface to CUBGCV using DllImport of Fortran library

File size: 15.0 KB
Line 
1CUBGCV TEST DRIVER RESULTS:
2
3IER =   0
4VAR = 0.0279
5
6GENERALIZED CROSS VALIDATION = 0.0318
7MEAN SQUARE RESIDUAL         = 0.0246
8RESIDUAL DEGREES OF FREEDOM  =  43.97
9
10INPUT DATA                 OUTPUT RESULTS
11
12  I    X(I)    F(I)          Y(I)   SE(I)      C(I,1)      C(I,2)      C(I,3)
13  1  0.0046  0.2222        0.0342  0.1004  0.3630E+01  0.0000E+00  0.2542E+02
14  2  0.0360 -0.1098        0.1488  0.0750  0.3705E+01  0.2391E+01 -0.9537E+01
15  3  0.0435 -0.0658        0.1767  0.0707  0.3740E+01  0.2175E+01 -0.4233E+02
16  4  0.0735  0.3906        0.2900  0.0594  0.3756E+01 -0.1642E+01 -0.2872E+02
17  5  0.0955  0.6054        0.3714  0.0558  0.3642E+01 -0.3535E+01  0.2911E+01
18  6  0.1078  0.3034        0.4155  0.0549  0.3557E+01 -0.3428E+01 -0.1225E+02
19  7  0.1269  0.7386        0.4822  0.0544  0.3412E+01 -0.4131E+01  0.2242E+02
20  8  0.1565  0.4616        0.5800  0.0543  0.3227E+01 -0.2143E+01  0.6415E+01
21  9  0.1679  0.4315        0.6165  0.0543  0.3180E+01 -0.1923E+01 -0.1860E+02
22 10  0.1869  0.5716        0.6762  0.0544  0.3087E+01 -0.2985E+01 -0.3274E+02
23 11  0.2149  0.6736        0.7595  0.0542  0.2843E+01 -0.5733E+01 -0.4435E+02
24 12  0.2356  0.7388        0.8155  0.0539  0.2549E+01 -0.8486E+01 -0.5472E+02
25 13  0.2557  1.1953        0.8630  0.0537  0.2139E+01 -0.1180E+02 -0.9784E+01
26 14  0.2674  1.0299        0.8864  0.0536  0.1860E+01 -0.1214E+02  0.9619E+01
27 15  0.2902  0.7981        0.9225  0.0534  0.1322E+01 -0.1149E+02 -0.7202E+01
28 16  0.3155  0.8973        0.9485  0.0532  0.7269E+00 -0.1203E+02 -0.1412E+02
29 17  0.3364  1.2695        0.9583  0.0530  0.2040E+00 -0.1292E+02  0.2796E+02
30 18  0.3557  0.7253        0.9577  0.0527 -0.2638E+00 -0.1130E+02 -0.3453E+01
31 19  0.3756  1.2127        0.9479  0.0526 -0.7176E+00 -0.1151E+02  0.3235E+02
32 20  0.3881  0.7304        0.9373  0.0525 -0.9889E+00 -0.1030E+02  0.4381E+01
33 21  0.4126  0.9810        0.9069  0.0525 -0.1486E+01 -0.9977E+01  0.1440E+02
34 22  0.4266  0.7117        0.8842  0.0525 -0.1756E+01 -0.9373E+01 -0.8925E+01
35 23  0.4566  0.7203        0.8227  0.0524 -0.2344E+01 -0.1018E+02 -0.2278E+02
36 24  0.4704  0.9242        0.7884  0.0524 -0.2637E+01 -0.1112E+02 -0.4419E+01
37 25  0.4914  0.7345        0.7281  0.0523 -0.3110E+01 -0.1140E+02 -0.3562E+01
38 26  0.5084  0.7378        0.6720  0.0524 -0.3500E+01 -0.1158E+02  0.5336E+01
39 27  0.5277  0.7441        0.6002  0.0525 -0.3941E+01 -0.1127E+02  0.2479E+02
40 28  0.5450  0.5612        0.5286  0.0527 -0.4310E+01 -0.9980E+01  0.2920E+02
41 29  0.5641  0.5049        0.4429  0.0529 -0.4659E+01 -0.8309E+01  0.3758E+02
42 30  0.5857  0.4725        0.3390  0.0531 -0.4964E+01 -0.5878E+01  0.5563E+02
43 31  0.6159  0.1380        0.1850  0.0531 -0.5167E+01 -0.8307E+00  0.4928E+02
44 32  0.6317  0.1412        0.1036  0.0531 -0.5157E+01  0.1499E+01  0.5437E+02
45 33  0.6446 -0.1110        0.0371  0.0531 -0.5091E+01  0.3614E+01  0.3434E+02
46 34  0.6707 -0.2605       -0.0927  0.0532 -0.4832E+01  0.6302E+01  0.1164E+02
47 35  0.6853 -0.1284       -0.1619  0.0533 -0.4640E+01  0.6812E+01  0.1617E+02
48 36  0.7064 -0.3452       -0.2564  0.0536 -0.4332E+01  0.7834E+01  0.4164E+01
49 37  0.7310 -0.5527       -0.3582  0.0538 -0.3939E+01  0.8141E+01 -0.2214E+02
50 38  0.7531 -0.3459       -0.4415  0.0540 -0.3611E+01  0.6674E+01 -0.9205E+01
51 39  0.7686 -0.5902       -0.4961  0.0541 -0.3410E+01  0.6245E+01 -0.2193E+02
52 40  0.7952 -0.7644       -0.5828  0.0541 -0.3125E+01  0.4494E+01 -0.4649E+02
53 41  0.8087 -0.5392       -0.6242  0.0541 -0.3029E+01  0.2614E+01 -0.3499E+02
54 42  0.8352 -0.4247       -0.7031  0.0539 -0.2964E+01 -0.1603E+00  0.2646E+01
55 43  0.8501 -0.6327       -0.7476  0.0538 -0.2967E+01 -0.4132E-01  0.1817E+02
56 44  0.8726 -0.9983       -0.8139  0.0538 -0.2942E+01  0.1180E+01 -0.6774E+01
57 45  0.8874 -0.9082       -0.8574  0.0542 -0.2911E+01  0.8778E+00 -0.1364E+02
58 46  0.9139 -0.8930       -0.9340  0.0566 -0.2893E+01 -0.2044E+00 -0.8094E+01
59 47  0.9271 -1.0233       -0.9723  0.0593 -0.2903E+01 -0.5258E+00 -0.1498E+02
60 48  0.9473 -0.8839       -1.0313  0.0665 -0.2942E+01 -0.1433E+01  0.4945E+01
61 49  0.9652 -1.0172       -1.0843  0.0766 -0.2989E+01 -0.1168E+01  0.1401E+02
62 50  0.9930 -1.2715       -1.1679  0.0998
63Documentation:
64C   COMPUTER            - VAX/DOUBLE
65C
66C   AUTHOR              - M.F.HUTCHINSON
67C                         CSIRO DIVISION OF MATHEMATICS AND STATISTICS
68C                         P.O. BOX 1965
69C                         CANBERRA, ACT 2601
70C                         AUSTRALIA
71C
72C   LATEST REVISION     - 15 AUGUST 1985
73C
74C   PURPOSE             - CUBIC SPLINE DATA SMOOTHER
75C
76C   USAGE               - CALL CUBGCV (X,F,DF,N,Y,C,IC,VAR,JOB,SE,WK,IER)
77C
78C   ARGUMENTS    X      - VECTOR OF LENGTH N CONTAINING THE
79C                           ABSCISSAE OF THE N DATA POINTS
80C                           (X(I),F(I)) I=1..N. (INPUT) X
81C                           MUST BE ORDERED SO THAT
82C                           X(I) .LT. X(I+1).
83C                F      - VECTOR OF LENGTH N CONTAINING THE
84C                           ORDINATES (OR FUNCTION VALUES)
85C                           OF THE N DATA POINTS (INPUT).
86C                DF     - VECTOR OF LENGTH N. (INPUT/OUTPUT)
87C                           DF(I) IS THE RELATIVE STANDARD DEVIATION
88C                           OF THE ERROR ASSOCIATED WITH DATA POINT I.
89C                           EACH DF(I) MUST BE POSITIVE.  THE VALUES IN
90C                           DF ARE SCALED BY THE SUBROUTINE SO THAT
91C                           THEIR MEAN SQUARE VALUE IS 1, AND UNSCALED
92C                           AGAIN ON NORMAL EXIT.
93C                           THE MEAN SQUARE VALUE OF THE DF(I) IS RETURNED
94C                           IN WK(7) ON NORMAL EXIT.
95C                           IF THE ABSOLUTE STANDARD DEVIATIONS ARE KNOWN,
96C                           THESE SHOULD BE PROVIDED IN DF AND THE ERROR
97C                           VARIANCE PARAMETER VAR (SEE BELOW) SHOULD THEN
98C                           BE SET TO 1.
99C                           IF THE RELATIVE STANDARD DEVIATIONS ARE UNKNOWN,
100C                           SET EACH DF(I)=1.
101C                N      - NUMBER OF DATA POINTS (INPUT).
102C                           N MUST BE .GE. 3.
103C                Y,C    - SPLINE COEFFICIENTS. (OUTPUT) Y
104C                           IS A VECTOR OF LENGTH N. C IS
105C                           AN N-1 BY 3 MATRIX. THE VALUE
106C                           OF THE SPLINE APPROXIMATION AT T IS
107C                           S(T)=((C(I,3)*D+C(I,2))*D+C(I,1))*D+Y(I)
108C                           WHERE X(I).LE.T.LT.X(I+1) AND
109C                           D = T-X(I).
110C                IC     - ROW DIMENSION OF MATRIX C EXACTLY
111C                           AS SPECIFIED IN THE DIMENSION
112C                           STATEMENT IN THE CALLING PROGRAM. (INPUT)
113C                VAR    - ERROR VARIANCE. (INPUT/OUTPUT)
114C                           IF VAR IS NEGATIVE (I.E. UNKNOWN) THEN
115C                           THE SMOOTHING PARAMETER IS DETERMINED
116C                           BY MINIMIZING THE GENERALIZED CROSS VALIDATION
117C                           AND AN ESTIMATE OF THE ERROR VARIANCE IS
118C                           RETURNED IN VAR.
119C                           IF VAR IS NON-NEGATIVE (I.E. KNOWN) THEN THE
120C                           SMOOTHING PARAMETER IS DETERMINED TO MINIMIZE
121C                           AN ESTIMATE, WHICH DEPENDS ON VAR, OF THE TRUE
122C                           MEAN SQUARE ERROR, AND VAR IS UNCHANGED.
123C                           IN PARTICULAR, IF VAR IS ZERO, THEN AN
124C                           INTERPOLATING NATURAL CUBIC SPLINE IS CALCULATED.
125C                           VAR SHOULD BE SET TO 1 IF ABSOLUTE STANDARD
126C                           DEVIATIONS HAVE BEEN PROVIDED IN DF (SEE ABOVE).
127C                JOB    - JOB SELECTION PARAMETER. (INPUT)
128C                         JOB = 0 SHOULD BE SELECTED IF POINT STANDARD ERROR
129C                           ESTIMATES ARE NOT REQUIRED IN SE.
130C                         JOB = 1 SHOULD BE SELECTED IF POINT STANDARD ERROR
131C                           ESTIMATES ARE REQUIRED IN SE.
132C                SE     - VECTOR OF LENGTH N CONTAINING BAYESIAN STANDARD
133C                           ERROR ESTIMATES OF THE FITTED SPLINE VALUES IN Y.
134C                           SE IS NOT REFERENCED IF JOB=0. (OUTPUT)
135C                WK     - WORK VECTOR OF LENGTH 7*(N + 2). ON NORMAL EXIT THE
136C                           FIRST 7 VALUES OF WK ARE ASSIGNED AS FOLLOWS:-
137C
138C                           WK(1) = SMOOTHING PARAMETER (= RHO/(RHO + 1))
139C                           WK(2) = ESTIMATE OF THE NUMBER OF DEGREES OF
140C                                   FREEDOM OF THE RESIDUAL SUM OF SQUARES
141C                           WK(3) = GENERALIZED CROSS VALIDATION
142C                           WK(4) = MEAN SQUARE RESIDUAL
143C                           WK(5) = ESTIMATE OF THE TRUE MEAN SQUARE ERROR
144C                                   AT THE DATA POINTS
145C                           WK(6) = ESTIMATE OF THE ERROR VARIANCE
146C                           WK(7) = MEAN SQUARE VALUE OF THE DF(I)
147C
148C                           IF WK(1)=0 (RHO=0) AN INTERPOLATING NATURAL CUBIC
149C                           SPLINE HAS BEEN CALCULATED.
150C                           IF WK(1)=1 (RHO=INFINITE) A LEAST SQUARES
151C                           REGRESSION LINE HAS BEEN CALCULATED.
152C                           WK(2) IS AN ESTIMATE OF THE NUMBER OF DEGREES OF
153C                           FREEDOM OF THE RESIDUAL WHICH REDUCES TO THE
154C                           USUAL VALUE OF N-2 WHEN A LEAST SQUARES REGRESSION
155C                           LINE IS CALCULATED.
156C                           WK(3),WK(4),WK(5) ARE CALCULATED WITH THE DF(I)
157C                           SCALED TO HAVE MEAN SQUARE VALUE 1.  THE
158C                           UNSCALED VALUES OF WK(3),WK(4),WK(5) MAY BE
159C                           CALCULATED BY DIVIDING BY WK(7).
160C                           WK(6) COINCIDES WITH THE OUTPUT VALUE OF VAR IF
161C                           VAR IS NEGATIVE ON INPUT.  IT IS CALCULATED WITH
162C                           THE UNSCALED VALUES OF THE DF(I) TO FACILITATE
163C                           COMPARISONS WITH A PRIORI VARIANCE ESTIMATES.
164C
165C                IER    - ERROR PARAMETER. (OUTPUT)
166C                         TERMINAL ERROR
167C                           IER = 129, IC IS LESS THAN N-1.
168C                           IER = 130, N IS LESS THAN 3.
169C                           IER = 131, INPUT ABSCISSAE ARE NOT
170C                             ORDERED SO THAT X(I).LT.X(I+1).
171C                           IER = 132, DF(I) IS NOT POSITIVE FOR SOME I.
172C                           IER = 133, JOB IS NOT 0 OR 1.
173C
174C   PRECISION/HARDWARE  - DOUBLE
175C
176C   REQUIRED ROUTINES   - SPINT1,SPFIT1,SPCOF1,SPERR1
177C
178C   REMARKS      THE NUMBER OF ARITHMETIC OPERATIONS REQUIRED BY THE
179C                SUBROUTINE IS PROPORTIONAL TO N.  THE SUBROUTINE
180C                USES AN ALGORITHM DEVELOPED BY M.F. HUTCHINSON AND
181C                F.R. DE HOOG, 'SMOOTHING NOISY DATA WITH SPLINE
182C                FUNCTIONS', NUMER. MATH. (IN PRESS)
183C
184C-----------------------------------------------------------------------
185C
186
187
188ALGORITHM
189
190CUBGCV calculates a natural cubic spline curve which smoothes a given set
191of data points, using statistical considerations to determine the amount
192of smoothing required, as described in reference 2.  If the error variance
193is known, it should be supplied to the routine in VAR.  The degree of
194smoothing is then determined by minimizing an unbiased estimate of the true
195mean square error.  On the other hand, if the error variance is not known, VAR
196should be set to -1.0.  The routine then determines the degree of smoothing
197by minimizing the generalized cross validation.  This is asymptotically the
198same as minimizing the true mean square error (see reference 1).  In this
199case, an estimate of the error variance is returned in VAR which may be
200compared with any a priori approximate estimates.  In either case, an
201estimate of the true mean square error is returned in WK(5).  This estimate,
202however, depends on the error variance estimate, and should only be accepted
203if the error variance estimate is reckoned to be correct.
204
205If job is set to 1, bayesian estimates of the standard error of each
206smoothed data value are returned in the array SE.  These also depend on
207the error variance estimate and should only be accepted if the error
208variance estimate is reckoned to be correct.  See reference 4.
209
210The number of arithmetic operations and the amount of storage required by
211the routine are both proportional to N, so that very large data sets may be
212analysed.  The data points do not have to be equally spaced or uniformly
213weighted.  The residual and the spline coefficients are calculated in the
214manner described in reference 3, while the trace and various statistics,
215including the generalized cross validation, are calculated in the manner
216described in reference 2.
217
218When VAR is known, any value of N greater than 2 is acceptable.  It is
219advisable, however, for N to be greater than about 20 when VAR is unknown.
220If the degree of smoothing done by CUBGCV when VAR is unknown is not
221satisfactory, the user should try specifying the degree of smoothing by
222setting VAR to a reasonable value.
223
224References:
225
2261.  Craven, Peter and Wahba, Grace, "Smoothing noisy data with spline
227    functions", Numer. Math. 31, 377-403 (1979).
2282.  Hutchinson, M.F. and de Hoog, F.R., "Smoothing noisy data with spline
229    functions", Numer. Math. (in press).
2303.  Reinsch, C.H., "Smoothing by spline functions", Numer. Math. 10,
231    177-183 (1967).
2324.  Wahba, Grace, "Bayesian 'confidence intervals' for the cross-validated
233    smoothing spline", J.R.Statist. Soc. B 45, 133-150 (1983).
234
235
236Example
237
238A sequence of 50 data points are generated by adding a random variable with
239uniform density in the interval [-0.3,0.3] to the curve y=sin(3*pi*x/2).
240The abscissae are unequally spaced in [0,1].  Point standard error estimates
241are returned in SE by setting JOB to 1.  The error variance estimate is
242returned in VAR.  It compares favourably with the true value of 0.03.
243The IMSL function GGUBFS is used to generate sample values of a uniform
244variable on [0,1].
245
246
247INPUT:
248
249      INTEGER          N,IC,JOB,IER
250      DOUBLE PRECISION X(50),F(50),Y(50),DF(50),C(49,3),WK(400),
251     *                 VAR,SE(50)
252      DOUBLE PRECISION GGUBFS,DSEED,DN
253      DATA DSEED/1.2345D4/
254C
255      N=50
256      IC=49
257      JOB=1
258      VAR=-1.0D0
259      DN=N
260C
261      DO 10 I=1,N
262      X(I)=(I - 0.5)/DN + (2.0*GGUBFS(DSEED) - 1.0)/(3.0*DN)
263      F(I)=DSIN(4.71238*X(I)) + (2.0*GGUBFS(DSEED) - 1.0)*0.3
264      DF(I)=1.0D0
265  10  CONTINUE
266      CALL CUBGCV(X,F,DF,N,Y,C,IC,VAR,JOB,SE,WK,IER)
267       .
268       .
269       .
270      END
271
272OUTPUT:
273
274IER = 0
275VAR = 0.0279
276GENERALIZED CROSS VALIDATION = 0.0318
277MEAN SQUARE RESIDUAL = 0.0246
278RESIDUAL DEGREES OF FREEDOM = 43.97
279FOR CHECKING PURPOSES THE FOLLOWING OUTPUT IS GIVEN:
280
281X(1)  = 0.0046    F(1)  =  0.2222     Y(1)  =  0.0342     SE(1)  = 0.1004
282X(21) = 0.4126    F(21) =  0.9810     Y(21) =  0.9069     SE(21) = 0.0525
283X(41) = 0.8087    F(41) = -0.5392     Y(41) = -0.6242     SE(41) = 0.0541
284
Note: See TracBrowser for help on using the repository browser.