1 |
|
2 |
############################################################################## |
3 |
# |
4 |
# Copyright (c) 2003-2020 by The University of Queensland |
5 |
# http://www.uq.edu.au |
6 |
# |
7 |
# Primary Business: Queensland, Australia |
8 |
# Licensed under the Apache License, version 2.0 |
9 |
# http://www.apache.org/licenses/LICENSE-2.0 |
10 |
# |
11 |
# Development until 2012 by Earth Systems Science Computational Center (ESSCC) |
12 |
# Development 2012-2013 by School of Earth Sciences |
13 |
# Development from 2014 by Centre for Geoscience Computing (GeoComp) |
14 |
# Development from 2019 by School of Earth and Environmental Sciences |
15 |
# |
16 |
############################################################################## |
17 |
|
18 |
from __future__ import division, print_function |
19 |
|
20 |
__copyright__="""Copyright (c) 2003-2020 by The University of Queensland |
21 |
http://www.uq.edu.au |
22 |
Primary Business: Queensland, Australia""" |
23 |
__license__="""Licensed under the Apache License, version 2.0 |
24 |
http://www.apache.org/licenses/LICENSE-2.0""" |
25 |
__url__="https://launchpad.net/escript-finley" |
26 |
|
27 |
from esys.escript import length, wherePositive, whereNegative, exp, inf, sup |
28 |
from esys.escript.modelframe import Model,ParameterSet |
29 |
from esys.escript.linearPDEs import LinearPDE |
30 |
from math import log |
31 |
import numpy |
32 |
|
33 |
class Sequencer(Model): |
34 |
""" |
35 |
Runs through time until t_end is reached. |
36 |
|
37 |
:ivar t_end: model is terminated when t_end is passed, default 1 (in). |
38 |
:type t_end: ``float`` |
39 |
:ivar dt_max: maximum time step size, default `Model.UNDEF_DT` (in) |
40 |
:type dt_max: ``float`` |
41 |
:ivar t: current time stamp (in/out). By default it is initialized with zero. |
42 |
:type t: ``float`` |
43 |
|
44 |
""" |
45 |
def __init__(self,**kwargs): |
46 |
""" |
47 |
""" |
48 |
super(Sequencer,self).__init__(**kwargs) |
49 |
self.declareParameter(t=0., |
50 |
t_end=1., |
51 |
dt_max=Model.UNDEF_DT) |
52 |
|
53 |
def doInitialization(self): |
54 |
""" |
55 |
initialize time integration |
56 |
""" |
57 |
self.__t_old = self.t |
58 |
|
59 |
def doStepPreprocessing(self, dt): |
60 |
self.t = self.__t_old+dt |
61 |
|
62 |
def doStepPostprocessing(self, dt): |
63 |
self.__t_old = self.t |
64 |
|
65 |
def finalize(self): |
66 |
""" |
67 |
returns true when `t` has reached `t_end` |
68 |
""" |
69 |
return self.t >= self.t_end |
70 |
|
71 |
def getSafeTimeStepSize(self, dt): |
72 |
""" |
73 |
returns `dt_max` |
74 |
""" |
75 |
return self.dt_max |
76 |
|
77 |
class GaussianProfile(ParameterSet): |
78 |
""" |
79 |
Generates a Gaussian profile at center x_c, width width and height A |
80 |
over a domain |
81 |
|
82 |
:note: Instance variable domain - domain |
83 |
:note: Instance variable x_c - center of the Gaussian profile (default [0.,0.,0.]) |
84 |
:note: Instance variable A - (in) height of the profile. A maybe a vector. (default 1.) |
85 |
:note: Instance variable width - (in) width of the profile (default 0.1) |
86 |
:note: Instance variable r - (in) radius of the circle (default = 0) |
87 |
|
88 |
In the case that the spatial dimension is two, The third component of |
89 |
x_c is dropped. |
90 |
""" |
91 |
def __init__(self,**kwargs): |
92 |
super(GaussianProfile, self).__init__(**kwargs) |
93 |
self.declareParameter(domain=None, |
94 |
x_c=numpy.zeros([3]), |
95 |
A=1., |
96 |
width=0.1, |
97 |
r=0) |
98 |
|
99 |
def out(self): |
100 |
""" |
101 |
Generate the Gaussian profile |
102 |
|
103 |
Link against this method to get the output of this model. |
104 |
""" |
105 |
x = self.domain.getX() |
106 |
dim = self.domain.getDim() |
107 |
l = length(x-self.x_c[:dim]) |
108 |
m = whereNegative(l-self.r) |
109 |
|
110 |
return (m+(1.-m)*exp(-log(2.)*(l/self.width)**2))*self.A |
111 |
|
112 |
class InterpolateOverBox(ParameterSet): |
113 |
""" |
114 |
Returns values at each time. The values are defined through given values |
115 |
at time node. For two dimensional domains back values are ignored. |
116 |
|
117 |
:note: Instance variable domain - domain |
118 |
:note: Instance variable value_left_bottom_front - (in) value at left,bottom,front corner |
119 |
:note: Instance variable value_right_bottom_front - (in) value at right, bottom, front corner |
120 |
:note: Instance variable value_left_top_front - (in) value at left,top,front corner |
121 |
:note: Instance variable value_right_top_front - (in) value at right,top,front corner |
122 |
:note: Instance variable value_left_bottom_back - (in) value at left,bottom,back corner |
123 |
:note: Instance variable value_right_bottom_back - (in) value at right,bottom,back corner |
124 |
:note: Instance variable value_left_top_back - (in) value at left,top,back corner |
125 |
:note: Instance variable value_right_top_back - (in) value at right,top,back corner |
126 |
""" |
127 |
|
128 |
def __init__(self, **kwargs): |
129 |
super(InterpolateOverBox, self).__init__(self) |
130 |
self.declareParameter(domain=None, |
131 |
value_left_bottom_front=0., |
132 |
value_right_bottom_front=0., |
133 |
value_left_top_front=0., |
134 |
value_right_top_front=0., |
135 |
value_left_bottom_back=0., |
136 |
value_right_bottom_back=0., |
137 |
value_left_top_back=0., |
138 |
value_right_top_back=0.) |
139 |
|
140 |
|
141 |
def out(self): |
142 |
""" |
143 |
values at domain locations by bilinear interpolation of the given values. |
144 |
|
145 |
Link against this method to get the output of this model. |
146 |
""" |
147 |
x = self.domain.getX() |
148 |
if self.domain.getDim() == 2: |
149 |
x0,x1=x[0],x[1] |
150 |
left_bottom_front0,right_top_back0=inf(x0),sup(x0) |
151 |
left_bottom_front1,right_top_back1=inf(x1),sup(x1) |
152 |
f_right = (x0 - left_bottom_front0)/(right_top_back0 -left_bottom_front0) |
153 |
f_left = 1. - f_right |
154 |
f_top = (x1 - left_bottom_front1)/(right_top_back1 - left_bottom_front1) |
155 |
f_bottom = 1. - f_top |
156 |
out = f_left * f_bottom * self.value_left_bottom_front \ |
157 |
+ f_right * f_bottom * self.value_right_bottom_front \ |
158 |
+ f_left * f_top * self.value_left_top_front \ |
159 |
+ f_right * f_top * self.value_right_top_front |
160 |
else: |
161 |
x0,x1,x2=x[0],x[1],x[2] |
162 |
left_bottom_front0,right_top_back0=inf(x0),sup(x0) |
163 |
left_bottom_front1,right_top_back1=inf(x1),sup(x1) |
164 |
left_bottom_front2,right_top_back2=inf(x2),sup(x2) |
165 |
f_right = (x0 - left_bottom_front0)/(right_top_back0 - left_bottom_front0) |
166 |
f_left = 1. - f_right |
167 |
f_top = (x1 - left_bottom_front1)/(right_top_back1 - left_bottom_front1) |
168 |
f_bottom = 1. - f_top |
169 |
f_back = (x2 - left_bottom_front1)/(right_top_back2 - left_bottom_front2) |
170 |
f_front = 1. - f_back |
171 |
out = f_left * f_bottom * f_front * self.value_left_bottom_front\ |
172 |
+ f_right * f_bottom * f_front * self.value_right_bottom_front\ |
173 |
+ f_left * f_top * f_front * self.value_left_top_front\ |
174 |
+ f_right * f_top * f_front * self.value_right_top_front\ |
175 |
+ f_left * f_bottom * f_back * self.value_left_bottom_back\ |
176 |
+ f_right * f_bottom * f_back * self.value_right_bottom_back\ |
177 |
+ f_left * f_top * f_back * self.value_left_top_back\ |
178 |
+ f_right * f_top * f_back * self.value_right_top_back |
179 |
return out |
180 |
|
181 |
|
182 |
class InterpolatedTimeProfile(ParameterSet): |
183 |
""" |
184 |
|
185 |
Returns values at each time. The values are defined through given |
186 |
values at time node. |
187 |
|
188 |
value[i] defines the value at time nodes[i]. Between nodes linear |
189 |
interpolation is used. |
190 |
|
191 |
For time t<nodes[0], value[0] is used and for t>nodes[l], values[l] |
192 |
is used where l=len(nodes)-1. |
193 |
|
194 |
:note: Instance variable t - (in) current time |
195 |
:note: Instance variable node - (in) list of time nodes |
196 |
:note: Instance variable values - (in) list of values at time nodes |
197 |
""" |
198 |
|
199 |
def __init__(self,**kwargs): |
200 |
super( InterpolatedTimeProfile, self).__init__(**kwargs) |
201 |
self.declareParameter(t=0., \ |
202 |
nodes=[0.,1.],\ |
203 |
values=[1.,1.]) |
204 |
def out(self): |
205 |
""" |
206 |
current value |
207 |
|
208 |
Link against this method to get the output of this model. |
209 |
""" |
210 |
l = len(self.nodes) - 1 |
211 |
t = self.t |
212 |
if t <= self.nodes[0]: |
213 |
return self.values[0] |
214 |
else: |
215 |
for i in range(1,l): |
216 |
if t < self.nodes[i]: |
217 |
m = (self.values[i-1] - self.values[i])/\ |
218 |
(self.nodes[i-1] - self.nodes[i]) |
219 |
return m*(t-self.nodes[i-1]) + self.values[i-1] |
220 |
return self.values[l] |
221 |
|
222 |
class ScalarDistributionFromTags(ParameterSet): |
223 |
""" |
224 |
creates a scalar distribution on a domain from tags, If tag_map is given |
225 |
the tags can be given a names and tag_map is used to map it into domain tags. |
226 |
|
227 |
:ivar domain: domain |
228 |
:type domain: `esys.escript.Domain` |
229 |
:ivar default: default value |
230 |
:ivar tag0: tag 0 |
231 |
:type tag0: ``int`` |
232 |
:ivar value0: value for tag 0 |
233 |
:type value0: ``float`` |
234 |
:ivar tag1: tag 1 |
235 |
:type tag1: ``int`` |
236 |
:ivar value1: value for tag 1 |
237 |
:type value1: ``float`` |
238 |
:ivar tag2: tag 2 |
239 |
:type tag2: ``int`` |
240 |
:ivar value2: value for tag 2 |
241 |
:type value2: ``float`` |
242 |
:ivar tag3: tag 3 |
243 |
:type tag3: ``int`` |
244 |
:ivar value3: value for tag 3 |
245 |
:type value3: ``float`` |
246 |
:ivar tag4: tag 4 |
247 |
:type tag4: ``int`` |
248 |
:ivar value4: value for tag 4 |
249 |
:type value4: ``float`` |
250 |
:ivar tag5: tag 5 |
251 |
:type tag5: ``int`` |
252 |
:ivar value5: value for tag 5 |
253 |
:type value5: ``float`` |
254 |
:ivar tag6: tag 6 |
255 |
:type tag6: ``int`` |
256 |
:ivar value6: value for tag 6 |
257 |
:type value6: ``float`` |
258 |
:ivar tag7: tag 7 |
259 |
:type tag7: ``int`` |
260 |
:ivar value7: value for tag 7 |
261 |
:type value7: ``float`` |
262 |
:ivar tag8: tag 8 |
263 |
:type tag8: ``int`` |
264 |
:ivar value8: value for tag 8 |
265 |
:type value8: ``float`` |
266 |
:ivar tag9: tag 9 |
267 |
:type tag9: ``int`` |
268 |
:ivar value9: value for tag 9 |
269 |
:type value9: ``float`` |
270 |
""" |
271 |
def __init__(self,**kwargs): |
272 |
super(ScalarDistributionFromTags, self).__init__(**kwargs) |
273 |
self.declareParameter(domain=None, |
274 |
default=0., |
275 |
tag0=None, |
276 |
value0=0., |
277 |
tag1=None, |
278 |
value1=0., |
279 |
tag2=None, |
280 |
value2=0., |
281 |
tag3=None, |
282 |
value3=0., |
283 |
tag4=None, |
284 |
value4=0., |
285 |
tag5=None, |
286 |
value5=0., |
287 |
tag6=None, |
288 |
value6=0., |
289 |
tag7=None, |
290 |
value7=0., |
291 |
tag8=None, |
292 |
value8=0., |
293 |
tag9=None, |
294 |
value9=0.) |
295 |
|
296 |
|
297 |
def out(self): |
298 |
""" |
299 |
returns a `esys.escript.Data` object |
300 |
Link against this method to get the output of this model. |
301 |
""" |
302 |
d=Scalar(self.default,Function(self.domain)) |
303 |
if not self.tag0 is None: d.setTaggedValue(self.tag0,self.value0) |
304 |
if not self.tag1 is None: d.setTaggedValue(self.tag1,self.value1) |
305 |
if not self.tag2 is None: d.setTaggedValue(self.tag2,self.value2) |
306 |
if not self.tag3 is None: d.setTaggedValue(self.tag3,self.value3) |
307 |
if not self.tag4 is None: d.setTaggedValue(self.tag4,self.value4) |
308 |
if not self.tag5 is None: d.setTaggedValue(self.tag5,self.value5) |
309 |
if not self.tag6 is None: d.setTaggedValue(self.tag6,self.value6) |
310 |
if not self.tag7 is None: d.setTaggedValue(self.tag7,self.value7) |
311 |
if not self.tag8 is None: d.setTaggedValue(self.tag8,self.value8) |
312 |
if not self.tag9 is None: d.setTaggedValue(self.tag9,self.value9) |
313 |
return d |
314 |
|
315 |
class SmoothScalarDistributionFromTags(ParameterSet): |
316 |
""" |
317 |
creates a smooth scalar distribution on a domain from region tags |
318 |
|
319 |
:ivar domain: domain |
320 |
:type domain: `esys.escript.Domain` |
321 |
:ivar default: default value |
322 |
:ivar tag0: tag 0 |
323 |
:type tag0: ``int`` |
324 |
:ivar value0: value for tag 0 |
325 |
:type value0: ``float`` |
326 |
:ivar tag1: tag 1 |
327 |
:type tag1: ``int`` |
328 |
:ivar value1: value for tag 1 |
329 |
:type value1: ``float`` |
330 |
:ivar tag2: tag 2 |
331 |
:type tag2: ``int`` |
332 |
:ivar value2: value for tag 2 |
333 |
:type value2: ``float`` |
334 |
:ivar tag3: tag 3 |
335 |
:type tag3: ``int`` |
336 |
:ivar value3: value for tag 3 |
337 |
:type value3: ``float`` |
338 |
:ivar tag4: tag 4 |
339 |
:type tag4: ``int`` |
340 |
:ivar value4: value for tag 4 |
341 |
:type value4: ``float`` |
342 |
:ivar tag5: tag 5 |
343 |
:type tag5: ``int`` |
344 |
:ivar value5: value for tag 5 |
345 |
:type value5: ``float`` |
346 |
:ivar tag6: tag 6 |
347 |
:type tag6: ``int`` |
348 |
:ivar value6: value for tag 6 |
349 |
:type value6: ``float`` |
350 |
:ivar tag7: tag 7 |
351 |
:type tag7: ``int`` |
352 |
:ivar value7: value for tag 7 |
353 |
:type value7: ``float`` |
354 |
:ivar tag8: tag 8 |
355 |
:type tag8: ``int`` |
356 |
:ivar value8: value for tag 8 |
357 |
:type value8: ``float`` |
358 |
:ivar tag9: tag 9 |
359 |
:type tag9: ``int`` |
360 |
:ivar value9: value for tag 9 |
361 |
:type value9: ``float`` |
362 |
""" |
363 |
def __init__(self,**kwargs): |
364 |
super(SmoothScalarDistributionFromTags, self).__init__(**kwargs) |
365 |
self.declareParameter(domain=None, |
366 |
default=0., |
367 |
tag0=None, |
368 |
value0=0., |
369 |
tag1=None, |
370 |
value1=0., |
371 |
tag2=None, |
372 |
value2=0., |
373 |
tag3=None, |
374 |
value3=0., |
375 |
tag4=None, |
376 |
value4=0., |
377 |
tag5=None, |
378 |
value5=0., |
379 |
tag6=None, |
380 |
value6=0., |
381 |
tag7=None, |
382 |
value7=0., |
383 |
tag8=None, |
384 |
value8=0., |
385 |
tag9=None, |
386 |
value9=0.) |
387 |
|
388 |
|
389 |
def __update(self,tag,tag_value,value): |
390 |
if self.__pde==None: |
391 |
self.__pde=LinearPDE(self.domain,numSolutions=1) |
392 |
mask=Scalar(0.,Function(self.domain)) |
393 |
mask.setTaggedValue(tag,1.) |
394 |
self.__pde.setValue(Y=mask) |
395 |
mask=wherePositive(abs(self.__pde.getRightHandSide())) |
396 |
value*=(1.-mask) |
397 |
value+=tag_value*mask |
398 |
return value |
399 |
|
400 |
def out(self): |
401 |
""" |
402 |
returns a `esys.escript.Data` object |
403 |
Link against this method to get the output of this model. |
404 |
""" |
405 |
d=Scalar(self.default,Solution(self.domain)) |
406 |
self.__pde=None |
407 |
if not self.tag0 is None: d=self.__update(self.tag0,self.value0,d) |
408 |
if not self.tag1 is None: d=self.__update(self.tag1,self.value1,d) |
409 |
if not self.tag2 is None: d=self.__update(self.tag2,self.value2,d) |
410 |
if not self.tag3 is None: d=self.__update(self.tag3,self.value3,d) |
411 |
if not self.tag4 is None: d=self.__update(self.tag4,self.value4,d) |
412 |
if not self.tag5 is None: d=self.__update(self.tag5,self.value5,d) |
413 |
if not self.tag6 is None: d=self.__update(self.tag6,self.value6,d) |
414 |
if not self.tag7 is None: d=self.__update(self.tag7,self.value7,d) |
415 |
if not self.tag8 is None: d=self.__update(self.tag8,self.value8,d) |
416 |
if not self.tag9 is None: d=self.__update(self.tag9,self.value9,d) |
417 |
return d |
418 |
|
419 |
class LinearCombination(ParameterSet): |
420 |
""" |
421 |
Returns a linear combination of the f0*v0+f1*v1+f2*v2+f3*v3+f4*v4 |
422 |
|
423 |
:ivar f0: numerical object or None, default=None (in) |
424 |
:ivar v0: numerical object or None, default=None (in) |
425 |
:ivar f1: numerical object or None, default=None (in) |
426 |
:ivar v1: numerical object or None, default=None (in) |
427 |
:ivar f2: numerical object or None, default=None (in) |
428 |
:ivar v2: numerical object or None, default=None (in) |
429 |
:ivar f3: numerical object or None, default=None (in) |
430 |
:ivar v3: numerical object or None, default=None (in) |
431 |
:ivar f4: numerical object or None, default=None (in) |
432 |
:ivar v4: numerical object or None, default=None (in) |
433 |
""" |
434 |
def __init__(self,**kwargs): |
435 |
super(LinearCombination, self).__init__(**kwargs) |
436 |
self.declareParameter(f0=None, \ |
437 |
v0=None, \ |
438 |
f1=None, \ |
439 |
v1=None, \ |
440 |
f2=None, \ |
441 |
v2=None, \ |
442 |
f3=None, \ |
443 |
v3=None, \ |
444 |
f4=None, \ |
445 |
v4=None) |
446 |
|
447 |
def out(self): |
448 |
""" |
449 |
returns f0*v0+f1*v1+f2*v2+f3*v3+f4*v4. |
450 |
Link against this method to get the output of this model. |
451 |
""" |
452 |
if not self.f0 is None and not self.v0 is None: |
453 |
fv0 = self.f0*self.v0 |
454 |
else: |
455 |
fv0 = None |
456 |
|
457 |
if not self.f1 is None and not self.v1 is None: |
458 |
fv1 = self.f1*self.v1 |
459 |
else: |
460 |
fv1 = None |
461 |
|
462 |
if not self.f2 is None and not self.v2 is None: |
463 |
fv2 = f2*v2 |
464 |
else: |
465 |
fv2 = None |
466 |
|
467 |
if not self.f3 is None and not self.v3 is None: |
468 |
fv3 = self.f3*self.v3 |
469 |
else: |
470 |
fv3 = None |
471 |
|
472 |
if not self.f4 is None and not self.v4 is None: |
473 |
fv4 = self.f4*self.v4 |
474 |
else: |
475 |
fv4 = None |
476 |
|
477 |
if fv0 is None: |
478 |
out = 0. |
479 |
else: |
480 |
out = fv0 |
481 |
if not fv1 is None: |
482 |
out += fv1 |
483 |
if not fv2 is None: |
484 |
out += fv2 |
485 |
if not fv3 is None: |
486 |
out += fv3 |
487 |
return out |
488 |
|
489 |
class MergeConstraints(ParameterSet): |
490 |
""" |
491 |
Returns a linear combination of the f0*v0+f1*v1+f2*v2+f3*v3+f4*v4 |
492 |
""" |
493 |
def __init__(self,**kwargs): |
494 |
super(MergeConstraints, self).__init__(**kwargs) |
495 |
self.declareParameter(location_of_constraint0=None, \ |
496 |
value_of_constraint0=None, \ |
497 |
location_of_constraint1=None, \ |
498 |
value_of_constraint1=None, \ |
499 |
location_of_constraint2=None, \ |
500 |
value_of_constraint2=None, \ |
501 |
location_of_constraint3=None, \ |
502 |
value_of_constraint3=None, \ |
503 |
location_of_constraint4=None, \ |
504 |
value_of_constraint4=None) |
505 |
def location_of_constraint(self): |
506 |
""" |
507 |
return the values used to constrain a solution |
508 |
|
509 |
:return: the mask marking the locations of the constraints |
510 |
:rtype: `escript.Scalar` |
511 |
""" |
512 |
out_loc=0 |
513 |
if not self.location_of_constraint0 is None: |
514 |
out_loc=wherePositive(out_loc+wherePositive(self.location_of_constraint0)) |
515 |
if not self.location_of_constraint1 is None: |
516 |
out_loc=wherePositive(out_loc+wherePositive(self.location_of_constraint1)) |
517 |
if not self.location_of_constraint2 is None: |
518 |
out_loc=wherePositive(out_loc+wherePositive(self.location_of_constraint2)) |
519 |
if not self.location_of_constraint3 is None: |
520 |
out_loc=wherePositive(out_loc+wherePositive(self.location_of_constraint3)) |
521 |
return out_loc |
522 |
|
523 |
def value_of_constraint(self): |
524 |
""" |
525 |
return the values used to constrain a solution |
526 |
|
527 |
:return: values to be used at the locations of the constraints. If |
528 |
``value`` is not given ``None`` is rerturned. |
529 |
:rtype: `escript.Scalar` |
530 |
""" |
531 |
out_loc=0 |
532 |
out=0 |
533 |
if not self.location_of_constraint0 is None: |
534 |
tmp=wherePositive(self.location_of_constraint0) |
535 |
out=out*(1.-tmp)+self.value_of_constraint0*tmp |
536 |
out_loc=wherePositive(out_loc+tmp) |
537 |
if not self.location_of_constraint1 is None: |
538 |
tmp=wherePositive(self.location_of_constraint1) |
539 |
out=out*(1.-tmp)+self.value_of_constraint1*tmp |
540 |
out_loc=wherePositive(out_loc+tmp) |
541 |
if not self.location_of_constraint2 is None: |
542 |
tmp=wherePositive(self.location_of_constraint2) |
543 |
out=out*(1.-tmp)+self.value_of_constraint2*tmp |
544 |
out_loc=wherePositive(out_loc+tmp) |
545 |
if not self.location_of_constraint3 is None: |
546 |
tmp=wherePositive(self.location_of_constraint3) |
547 |
out=out*(1.-tmp)+self.value_of_constraint3*tmp |
548 |
out_loc=wherePositive(out_loc+tmp) |
549 |
return out |
550 |
# vim: expandtab shiftwidth=4: |