-
Notifications
You must be signed in to change notification settings - Fork 1
/
demo_pspline_2d.py
206 lines (162 loc) · 5.38 KB
/
demo_pspline_2d.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
#!/usr/bin/env python
"""
2-D spline in float64 precision
"""
import sys, time
import numpy as np
from pypspline3.pspline_2d import pspline, griddata
eps = 1.e-6
n1, n2 = 11, 21
bcs1 = (0,0)
bcs2 = (0,0) # not-a-knot
x1min, x1max = 0., 1.
x2min, x2max = 0., 1.
x1 = np.arange(x1min, x1max+eps, (x1max-x1min)/float(n1-1))
x2 = np.arange(x2min, x2max+eps, (x2max-x2min)/float(n2-1))
tic = time.time()
xx1, xx2 = griddata(x1, x2)
toc = time.time()
print("griddata : time->%10.1f secs" % (toc-tic))
tic = time.time()
#########################
f = xx1**3 + 2*xx1*xx2**3
#########################
toc = time.time()
print("function evaluations: time->%10.1f secs" % (toc-tic))
tic = time.time()
spl = pspline(x1, x2)
# may set BCs if not-a-knot
spl.setup(f)
toc = time.time()
print("init/setup: %d original grid nodes time->%10.1f secs" %
(n1*n2, toc-tic))
# skip save/load for now
# new mesh
n1, n2 = 2*n2-1, 2*n2 # 3*n2-1, 3*n2
x1 = np.arange(x1min, x1max+eps, (x1max-x1min)/float(n1-1))
x2 = np.arange(x2min, x2max+eps, (x2max-x2min)/float(n2-1))
xx1, xx2 = griddata(x1, x2)
fexact = xx1**3 + 2*xx1*xx2**3
# point interpolation
nint = n1*n2
error = 0
tic = time.time()
for i2 in range(n2):
for i1 in range(n1):
fi, ier, iwarn = spl.interp_point(x1[i1], x2[i2])
error += (fi - fexact[i2,i1])**2
toc = time.time()
error /= nint
error = np.sqrt(error)
print("interp_point: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# cloud interpolation
n = n1*n2
xc1 = np.arange(x1min, x1max+eps, (x1max-x1min)/float(n-1))
xc2 = np.arange(x2min, x2max+eps, (x2max-x2min)/float(n-1))
fcexact = xc1**3 + 2*xc1*xc2**3
tic = time.time()
fi, ier, iwarn = spl.interp_cloud(xc1, xc2)
toc = time.time()
error = np.sum((fi-fcexact)**2)/nint
print("interp_cloud: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# array interpolation
tic = time.time()
fi, ier, iwarn = spl.interp_array(x1, x2)
toc = time.time()
error = np.sum(np.sum(np.sum((fi-fexact)**2)))/nint
print("interp_array: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
## df/dx
fexact = 3*xx1**2 + 2*xx2**3
# point df/dx
tic = time.time()
error = 0
for i2 in range(n2):
for i1 in range(n1):
fi, ier, iwarn = spl.derivative_point(1, 0, x1[i1], x2[i2])
error += (fi - fexact[i2,i1])**2
toc = time.time()
error /= nint
error = np.sqrt(error)
print("derivative_point df/dx: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# array df/dx
tic = time.time()
fi, ier, iwarn = spl.derivative_array(1, 0, x1, x2)
toc = time.time()
error = np.sum(np.sum(np.sum((fi-fexact)**2)))/nint
print("derivative_array df/dx: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# cloud df/dx
fcexact = 3*xc1**2 + 2*xc2**3
tic = time.time()
fi, ier, iwarn = spl.derivative_cloud(1, 0, xc1, xc2)
toc = time.time()
error = np.sum((fi-fcexact)**2)/nint
print("derivative_cloud df/dx: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
## d^2f/dy^2
fexact = 12*xx1*xx2
# point d^2f/dy^2
error = 0
tic = time.time()
for i2 in range(n2):
for i1 in range(n1):
fi, ier, iwarn = spl.derivative_point(0,2, x1[i1], x2[i2])
error += (fi - fexact[i2,i1])**2
toc = time.time()
error /= nint
error = np.sqrt(error)
print("derivative_point d^2f/dy^2: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# array d^2f/dy^2
tic = time.time()
fi, ier, iwarn = spl.derivative_array(0,2, x1, x2)
toc = time.time()
error = np.sum(np.sum(np.sum((fi-fexact)**2)))/nint
print("derivative_array d^2f/dy^2: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# cloud d^2f/dy^2
fcexact = 12*xc1*xc2
tic = time.time()
fi, ier, iwarn = spl.derivative_cloud(0,2, xc1, xc2)
toc = time.time()
error = np.sum((fi-fcexact)**2)/nint
print("derivative_cloud d^2f/dy^2: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
## gradients
f1exact = 3*xx1**2 + 2* xx2**3
f2exact = 6*xx1*xx2**2
# point
error = 0
tic = time.time()
for i2 in range(n2):
for i1 in range(n1):
f1,f2, ier, iwarn = spl.gradient_point(x1[i1], x2[i2])
error += ( \
(f1 - f1exact[i2,i1])**2 +
(f2 - f2exact[i2,i1])**2
)/2.
toc = time.time()
error /= nint
error = np.sqrt(error)
print("gradient_point: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# array
tic = time.time()
f1,f2, ier, iwarn = spl.gradient_array(x1, x2)
toc = time.time()
error = np.sum(np.sum(np.sum(((f1-f1exact)**2 + (f2-f2exact)**2)/2.)))/nint
print("gradient_array: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))
# cloud
f1cexact = 3*xc1**2 + 2* xc2**3
f2cexact = 6*xc1*xc2**2
tic = time.time()
f1,f2, ier, iwarn = spl.gradient_cloud(xc1, xc2)
toc = time.time()
error = np.sum(((f1-f1cexact)**2 + (f2-f2cexact)**2/2.))/nint
print("gradient_cloud: %d evaluations (error=%g) ier=%d iwarn=%d time->%10.1f secs" %
(nint, error, ier, iwarn, toc-tic))