1 |
# $Id$ |
2 |
|
3 |
""" |
4 |
Provides some tools related to PDEs. |
5 |
|
6 |
Currently includes: |
7 |
- Projector - to project a discontinuous |
8 |
- Locator - to trace values in data objects at a certain location |
9 |
- TimeIntegrationManager - to handel extraplotion in time |
10 |
|
11 |
@var __author__: name of author |
12 |
@var __copyright__: copyrights |
13 |
@var __license__: licence agreement |
14 |
@var __url__: url entry point on documentation |
15 |
@var __version__: version |
16 |
@var __date__: date of the version |
17 |
""" |
18 |
|
19 |
__author__="Lutz Gross, l.gross@uq.edu.au" |
20 |
__copyright__=""" Copyright (c) 2006 by ACcESS MNRF |
21 |
http://www.access.edu.au |
22 |
Primary Business: Queensland, Australia""" |
23 |
__license__="""Licensed under the Open Software License version 3.0 |
24 |
http://www.opensource.org/licenses/osl-3.0.php""" |
25 |
__url__="http://www.iservo.edu.au/esys" |
26 |
__version__="$Revision$" |
27 |
__date__="$Date$" |
28 |
|
29 |
|
30 |
import escript |
31 |
import linearPDEs |
32 |
import numarray |
33 |
import util |
34 |
|
35 |
class TimeIntegrationManager: |
36 |
""" |
37 |
a simple mechanism to manage time dependend values. |
38 |
|
39 |
typical usage is:: |
40 |
|
41 |
dt=0.1 # time increment |
42 |
tm=TimeIntegrationManager(inital_value,p=1) |
43 |
while t<1. |
44 |
v_guess=tm.extrapolate(dt) # extrapolate to t+dt |
45 |
v=... |
46 |
tm.checkin(dt,v) |
47 |
t+=dt |
48 |
|
49 |
@note: currently only p=1 is supported. |
50 |
""" |
51 |
def __init__(self,*inital_values,**kwargs): |
52 |
""" |
53 |
sets up the value manager where inital_value is the initial value and p is order used for extrapolation |
54 |
""" |
55 |
if kwargs.has_key("p"): |
56 |
self.__p=kwargs["p"] |
57 |
else: |
58 |
self.__p=1 |
59 |
if kwargs.has_key("time"): |
60 |
self.__t=kwargs["time"] |
61 |
else: |
62 |
self.__t=0. |
63 |
self.__v_mem=[inital_values] |
64 |
self.__order=0 |
65 |
self.__dt_mem=[] |
66 |
self.__num_val=len(inital_values) |
67 |
|
68 |
def getTime(self): |
69 |
return self.__t |
70 |
def getValue(self): |
71 |
out=self.__v_mem[0] |
72 |
if len(out)==1: |
73 |
return out[0] |
74 |
else: |
75 |
return out |
76 |
|
77 |
def checkin(self,dt,*values): |
78 |
""" |
79 |
adds new values to the manager. the p+1 last value get lost |
80 |
""" |
81 |
o=min(self.__order+1,self.__p) |
82 |
self.__order=min(self.__order+1,self.__p) |
83 |
v_mem_new=[values] |
84 |
dt_mem_new=[dt] |
85 |
for i in range(o-1): |
86 |
v_mem_new.append(self.__v_mem[i]) |
87 |
dt_mem_new.append(self.__dt_mem[i]) |
88 |
v_mem_new.append(self.__v_mem[o-1]) |
89 |
self.__order=o |
90 |
self.__v_mem=v_mem_new |
91 |
self.__dt_mem=dt_mem_new |
92 |
self.__t+=dt |
93 |
|
94 |
def extrapolate(self,dt): |
95 |
""" |
96 |
extrapolates to dt forward in time. |
97 |
""" |
98 |
if self.__order==0: |
99 |
out=self.__v_mem[0] |
100 |
else: |
101 |
out=[] |
102 |
for i in range(self.__num_val): |
103 |
out.append((1.+dt/self.__dt_mem[0])*self.__v_mem[0][i]-dt/self.__dt_mem[0]*self.__v_mem[1][i]) |
104 |
|
105 |
if len(out)==0: |
106 |
return None |
107 |
elif len(out)==1: |
108 |
return out[0] |
109 |
else: |
110 |
return out |
111 |
|
112 |
|
113 |
class Projector: |
114 |
""" |
115 |
The Projector is a factory which projects a discontiuous function onto a |
116 |
continuous function on the a given domain. |
117 |
""" |
118 |
def __init__(self, domain, reduce = True, fast=True): |
119 |
""" |
120 |
Create a continuous function space projector for a domain. |
121 |
|
122 |
@param domain: Domain of the projection. |
123 |
@param reduce: Flag to reduce projection order (default is True) |
124 |
@param fast: Flag to use a fast method based on matrix lumping (default is true) |
125 |
""" |
126 |
self.__pde = linearPDEs.LinearPDE(domain) |
127 |
if fast: |
128 |
self.__pde.setSolverMethod(linearPDEs.LinearPDE.LUMPING) |
129 |
self.__pde.setSymmetryOn() |
130 |
self.__pde.setReducedOrderTo(reduce) |
131 |
self.__pde.setValue(D = 1.) |
132 |
return |
133 |
|
134 |
def __del__(self): |
135 |
return |
136 |
|
137 |
def __call__(self, input_data): |
138 |
""" |
139 |
Projects input_data onto a continuous function |
140 |
|
141 |
@param input_data: The input_data to be projected. |
142 |
""" |
143 |
out=escript.Data(0.,input_data.getShape(),self.__pde.getFunctionSpaceForSolution()) |
144 |
if input_data.getRank()==0: |
145 |
self.__pde.setValue(Y = input_data) |
146 |
out=self.__pde.getSolution() |
147 |
elif input_data.getRank()==1: |
148 |
for i0 in range(input_data.getShape()[0]): |
149 |
self.__pde.setValue(Y = input_data[i0]) |
150 |
out[i0]=self.__pde.getSolution() |
151 |
elif input_data.getRank()==2: |
152 |
for i0 in range(input_data.getShape()[0]): |
153 |
for i1 in range(input_data.getShape()[1]): |
154 |
self.__pde.setValue(Y = input_data[i0,i1]) |
155 |
out[i0,i1]=self.__pde.getSolution() |
156 |
elif input_data.getRank()==3: |
157 |
for i0 in range(input_data.getShape()[0]): |
158 |
for i1 in range(input_data.getShape()[1]): |
159 |
for i2 in range(input_data.getShape()[2]): |
160 |
self.__pde.setValue(Y = input_data[i0,i1,i2]) |
161 |
out[i0,i1,i2]=self.__pde.getSolution() |
162 |
else: |
163 |
for i0 in range(input_data.getShape()[0]): |
164 |
for i1 in range(input_data.getShape()[1]): |
165 |
for i2 in range(input_data.getShape()[2]): |
166 |
for i3 in range(input_data.getShape()[3]): |
167 |
self.__pde.setValue(Y = input_data[i0,i1,i2,i3]) |
168 |
out[i0,i1,i2,i3]=self.__pde.getSolution() |
169 |
return out |
170 |
|
171 |
class NoPDE: |
172 |
""" |
173 |
solves the following problem for u: |
174 |
|
175 |
M{kronecker[i,j]*D[j]*u[j]=Y[i]} |
176 |
|
177 |
with constraint |
178 |
|
179 |
M{u[j]=r[j]} where M{q[j]>0} |
180 |
|
181 |
where D, Y, r and q are given functions of rank 1. |
182 |
|
183 |
In the case of scalars this takes the form |
184 |
|
185 |
M{D*u=Y} |
186 |
|
187 |
with constraint |
188 |
|
189 |
M{u=r} where M{q>0} |
190 |
|
191 |
where D, Y, r and q are given scalar functions. |
192 |
|
193 |
The constraint is overwriting any other condition. |
194 |
|
195 |
@note: This class is similar to the L{linearPDEs.LinearPDE} class with A=B=C=X=0 but has the intention |
196 |
that all input parameter are given in L{Solution} or L{ReducedSolution}. The whole |
197 |
thing is a bit strange and I blame Robert.Woodcock@csiro.au for this. |
198 |
""" |
199 |
def __init__(self,domain,D=None,Y=None,q=None,r=None): |
200 |
""" |
201 |
initialize the problem |
202 |
|
203 |
@param domain: domain of the PDE. |
204 |
@type domain: L{Domain} |
205 |
@param D: coefficient of the solution. |
206 |
@type D: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
207 |
@param Y: right hand side |
208 |
@type Y: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
209 |
@param q: location of constraints |
210 |
@type q: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
211 |
@param r: value of solution at locations of constraints |
212 |
@type r: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
213 |
""" |
214 |
self.__domain=domain |
215 |
self.__D=D |
216 |
self.__Y=Y |
217 |
self.__q=q |
218 |
self.__r=r |
219 |
self.__u=None |
220 |
self.__function_space=escript.Solution(self.__domain) |
221 |
def setReducedOn(self): |
222 |
""" |
223 |
sets the L{FunctionSpace} of the solution to L{ReducedSolution} |
224 |
""" |
225 |
self.__function_space=escript.ReducedSolution(self.__domain) |
226 |
self.__u=None |
227 |
|
228 |
def setReducedOff(self): |
229 |
""" |
230 |
sets the L{FunctionSpace} of the solution to L{Solution} |
231 |
""" |
232 |
self.__function_space=escript.Solution(self.__domain) |
233 |
self.__u=None |
234 |
|
235 |
def setValue(self,D=None,Y=None,q=None,r=None): |
236 |
""" |
237 |
assigns values to the parameters. |
238 |
|
239 |
@param D: coefficient of the solution. |
240 |
@type D: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
241 |
@param Y: right hand side |
242 |
@type Y: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
243 |
@param q: location of constraints |
244 |
@type q: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
245 |
@param r: value of solution at locations of constraints |
246 |
@type r: C{float}, C{int}, L{numarray.NumArray}, L{Data} |
247 |
""" |
248 |
if not D==None: |
249 |
self.__D=D |
250 |
self.__u=None |
251 |
if not Y==None: |
252 |
self.__Y=Y |
253 |
self.__u=None |
254 |
if not q==None: |
255 |
self.__q=q |
256 |
self.__u=None |
257 |
if not r==None: |
258 |
self.__r=r |
259 |
self.__u=None |
260 |
|
261 |
def getSolution(self): |
262 |
""" |
263 |
returns the solution |
264 |
|
265 |
@return: the solution of the problem |
266 |
@rtype: L{Data} object in the L{FunctionSpace} L{Solution} or L{ReducedSolution}. |
267 |
""" |
268 |
if self.__u==None: |
269 |
if self.__D==None: |
270 |
raise ValueError,"coefficient D is undefined" |
271 |
D=escript.Data(self.__D,self.__function_space) |
272 |
if D.getRank()>1: |
273 |
raise ValueError,"coefficient D must have rank 0 or 1" |
274 |
if self.__Y==None: |
275 |
self.__u=escript.Data(0.,D.getShape(),self.__function_space) |
276 |
else: |
277 |
self.__u=util.quotient(self.__Y,D) |
278 |
if not self.__q==None: |
279 |
q=util.wherePositive(escript.Data(self.__q,self.__function_space)) |
280 |
self.__u*=(1.-q) |
281 |
if not self.__r==None: self.__u+=q*self.__r |
282 |
return self.__u |
283 |
|
284 |
class Locator: |
285 |
""" |
286 |
Locator provides access to the values of data objects at a given |
287 |
spatial coordinate x. |
288 |
|
289 |
In fact, a Locator object finds the sample in the set of samples of a |
290 |
given function space or domain where which is closest to the given |
291 |
point x. |
292 |
""" |
293 |
|
294 |
def __init__(self,where,x=numarray.zeros((3,))): |
295 |
""" |
296 |
Initializes a Locator to access values in Data objects on the Doamin |
297 |
or FunctionSpace where for the sample point which |
298 |
closest to the given point x. |
299 |
""" |
300 |
if isinstance(where,escript.FunctionSpace): |
301 |
self.__function_space=where |
302 |
else: |
303 |
self.__function_space=escript.ContinuousFunction(where) |
304 |
self.__id=util.length(self.__function_space.getX()-x[:self.__function_space.getDim()]).mindp() |
305 |
|
306 |
def __str__(self): |
307 |
""" |
308 |
Returns the coordinates of the Locator as a string. |
309 |
""" |
310 |
return "<Locator %s>"%str(self.getX()) |
311 |
|
312 |
def getFunctionSpace(self): |
313 |
""" |
314 |
Returns the function space of the Locator. |
315 |
""" |
316 |
return self.__function_space |
317 |
|
318 |
def getId(self): |
319 |
""" |
320 |
Returns the identifier of the location. |
321 |
""" |
322 |
return self.__id |
323 |
|
324 |
def getX(self): |
325 |
""" |
326 |
Returns the exact coordinates of the Locator. |
327 |
""" |
328 |
return self(self.getFunctionSpace().getX()) |
329 |
|
330 |
def __call__(self,data): |
331 |
""" |
332 |
Returns the value of data at the Locator of a Data object otherwise |
333 |
the object is returned. |
334 |
""" |
335 |
return self.getValue(data) |
336 |
|
337 |
def getValue(self,data): |
338 |
""" |
339 |
Returns the value of data at the Locator if data is a Data object |
340 |
otherwise the object is returned. |
341 |
""" |
342 |
if isinstance(data,escript.Data): |
343 |
if data.getFunctionSpace()==self.getFunctionSpace(): |
344 |
#out=data.convertToNumArrayFromDPNo(self.getId()[0],self.getId()[1]) |
345 |
out=data.convertToNumArrayFromDPNo(self.getId()[0],self.getId()[1],self.getId()[2]) |
346 |
else: |
347 |
#out=data.interpolate(self.getFunctionSpace()).convertToNumArrayFromDPNo(self.getId()[0],self.getId()[1]) |
348 |
out=data.interpolate(self.getFunctionSpace()).convertToNumArrayFromDPNo(self.getId()[0],self.getId()[1],self.getId()[2]) |
349 |
if data.getRank()==0: |
350 |
return out[0] |
351 |
else: |
352 |
return out |
353 |
else: |
354 |
return data |
355 |
|
356 |
# vim: expandtab shiftwidth=4: |