1 | """ |
---|
2 | This module implements invariant and its related computations. |
---|
3 | @author: Gervaise B. Alina/UTK |
---|
4 | """ |
---|
5 | |
---|
6 | import math |
---|
7 | import numpy |
---|
8 | |
---|
9 | from DataLoader.data_info import Data1D as LoaderData1D |
---|
10 | #from DataLoader.data_info import Data1D as LoaderData1D |
---|
11 | from DataLoader.qsmearing import smear_selection |
---|
12 | |
---|
13 | |
---|
14 | # The minimum q-value to be used when extrapolating |
---|
15 | Q_MINIMUM = 1e-5 |
---|
16 | |
---|
17 | # The maximum q-value to be used when extrapolating |
---|
18 | Q_MAXIMUM = 10 |
---|
19 | |
---|
20 | # Number of steps in the extrapolation |
---|
21 | INTEGRATION_NSTEPS = 1000 |
---|
22 | |
---|
23 | def guinier(x, scale=1, radius=0.1): |
---|
24 | """ |
---|
25 | Compute a F(x) = scale* e-((radius*x)**2/3). |
---|
26 | @param x: a vector of q values |
---|
27 | @param scale: the scale value |
---|
28 | @param radius: the guinier radius value |
---|
29 | @return F(x) |
---|
30 | """ |
---|
31 | value = numpy.array([math.exp(-((radius * i)**2/3)) for i in x ]) |
---|
32 | return scale * value |
---|
33 | |
---|
34 | def power_law(x, scale=1, power=4): |
---|
35 | """ |
---|
36 | F(x) = scale* (x)^(-power) |
---|
37 | when power= 4. the model is porod |
---|
38 | else power_law |
---|
39 | The model has three parameters: |
---|
40 | @param x: a vector of q values |
---|
41 | @param power: power of the function |
---|
42 | @param scale : scale factor value |
---|
43 | @param F(x) |
---|
44 | """ |
---|
45 | value = numpy.array([ math.pow(i, -power) for i in x ]) |
---|
46 | return scale * value |
---|
47 | |
---|
48 | class FitFunctor: |
---|
49 | """ |
---|
50 | compute f(x) |
---|
51 | """ |
---|
52 | def __init__(self,data , function): |
---|
53 | """ |
---|
54 | @param function :the function used for computing residuals |
---|
55 | @param Data: data used for computing residuals |
---|
56 | """ |
---|
57 | self.function = function |
---|
58 | self.data = data |
---|
59 | x_len = len(self.data.x) -1 |
---|
60 | #fitting range |
---|
61 | self.qmin = self.data.x[0] |
---|
62 | if self.qmin == 0: |
---|
63 | self.qmin = Q_MINIMUM |
---|
64 | |
---|
65 | self.qmax = self.data.x[x_len] |
---|
66 | #Unsmeared q range |
---|
67 | self._qmin_unsmeared = 0 |
---|
68 | self._qmax_unsmeared = self.data.x[x_len] |
---|
69 | |
---|
70 | #bin for smear data |
---|
71 | self._first_unsmeared_bin = 0 |
---|
72 | self._last_unsmeared_bin = x_len |
---|
73 | |
---|
74 | # Identify the bin range for the unsmeared and smeared spaces |
---|
75 | self.idx = (self.data.x >= self.qmin) & (self.data.x <= self.qmax) |
---|
76 | self.idx_unsmeared = (self.data.x >= self._qmin_unsmeared) \ |
---|
77 | & (self.data.x <= self._qmax_unsmeared) |
---|
78 | |
---|
79 | #get the smear object of data |
---|
80 | self.smearer = smear_selection( self.data ) |
---|
81 | |
---|
82 | |
---|
83 | def set_fit_range(self ,qmin=None, qmax=None): |
---|
84 | """ to set the fit range""" |
---|
85 | if qmin is not None: |
---|
86 | self.qmin = qmin |
---|
87 | if qmax is not None: |
---|
88 | self.qmax = qmax |
---|
89 | |
---|
90 | # Determine the range needed in unsmeared-Q to cover |
---|
91 | # the smeared Q range |
---|
92 | self._qmin_unsmeared = self.qmin |
---|
93 | self._qmax_unsmeared = self.qmax |
---|
94 | |
---|
95 | self._first_unsmeared_bin = 0 |
---|
96 | self._last_unsmeared_bin = len(self.data.x)-1 |
---|
97 | |
---|
98 | if self.smearer!=None: |
---|
99 | self._first_unsmeared_bin, self._last_unsmeared_bin = self.smearer.get_bin_range(self.qmin, self.qmax) |
---|
100 | self._qmin_unsmeared = self.data.x[self._first_unsmeared_bin] |
---|
101 | self._qmax_unsmeared = self.data.x[self._last_unsmeared_bin] |
---|
102 | |
---|
103 | # Identify the bin range for the unsmeared and smeared spaces |
---|
104 | self.idx = (self.data.x >= self.qmin) & (self.data.x <= self.qmax) |
---|
105 | self.idx_unsmeared = (self.data.x >= self._qmin_unsmeared) \ |
---|
106 | & (self.data.x <= self._qmax_unsmeared) |
---|
107 | |
---|
108 | def fit(self): |
---|
109 | """ |
---|
110 | Fit data for y = ax + b return a and b |
---|
111 | |
---|
112 | """ |
---|
113 | fx = self.data.y |
---|
114 | ## Smear theory data |
---|
115 | if self.smearer is not None: |
---|
116 | fx = self.smearer(fx, self._first_unsmeared_bin, |
---|
117 | self._last_unsmeared_bin) |
---|
118 | A = numpy.vstack([ self.data.x, numpy.ones(len(self.data.x))]).T |
---|
119 | |
---|
120 | a, b = numpy.linalg.lstsq(A, fx)[0] |
---|
121 | return a, b |
---|
122 | |
---|
123 | class InvariantCalculator(object): |
---|
124 | """ |
---|
125 | Compute invariant if data is given. |
---|
126 | Can provide volume fraction and surface area if the user provides |
---|
127 | Porod constant and contrast values. |
---|
128 | @precondition: the user must send a data of type DataLoader.Data1D |
---|
129 | the user provide background and scale values. |
---|
130 | |
---|
131 | @note: Some computations depends on each others. |
---|
132 | """ |
---|
133 | |
---|
134 | |
---|
135 | def __init__(self, data, background=0, scale=1 ): |
---|
136 | """ |
---|
137 | Initialize variables |
---|
138 | @param data: data must be of type DataLoader.Data1D |
---|
139 | @param contrast: contrast value of type float |
---|
140 | @param pConst: Porod Constant of type float |
---|
141 | """ |
---|
142 | # Background and scale should be private data member if the only way to |
---|
143 | # change them are by instantiating a new object. |
---|
144 | self._background = background |
---|
145 | self._scale = scale |
---|
146 | |
---|
147 | # The data should be private |
---|
148 | self._data = self._get_data(data) |
---|
149 | |
---|
150 | # Since there are multiple variants of Q*, you should force the |
---|
151 | # user to use the get method and keep Q* a private data member |
---|
152 | self._qstar = None |
---|
153 | |
---|
154 | # You should keep the error on Q* so you can reuse it without |
---|
155 | # recomputing the whole thing. |
---|
156 | self._qstar_err = 0 |
---|
157 | |
---|
158 | # Extrapolation parameters |
---|
159 | self._low_extrapolation_npts = 4 |
---|
160 | self._low_extrapolation_function = guinier |
---|
161 | self._low_extrapolation_power = 4 |
---|
162 | |
---|
163 | self._high_extrapolation_npts = 4 |
---|
164 | self._high_extrapolation_function = power_law |
---|
165 | self._high_extrapolation_power = 4 |
---|
166 | |
---|
167 | |
---|
168 | def set_extrapolation(self, range, npts=4, function=None, power=4): |
---|
169 | """ |
---|
170 | Set the extrapolation parameters for the high or low Q-range. |
---|
171 | Note that this does not turn extrapolation on or off. |
---|
172 | @param range: a keyword set the type of extrapolation . type string |
---|
173 | @param npts: the numbers of q points of data to consider for extrapolation |
---|
174 | @param function: a keyword to select the function to use for extrapolation. |
---|
175 | of type string. |
---|
176 | @param power: an power to apply power_low function |
---|
177 | |
---|
178 | """ |
---|
179 | range = range.lower() |
---|
180 | if range not in ['high', 'low']: |
---|
181 | raise ValueError, "Extrapolation range should be 'high' or 'low'" |
---|
182 | function = function.lower() |
---|
183 | if function not in ['power_law', 'guinier']: |
---|
184 | raise ValueError, "Extrapolation function should be 'guinier' or 'power_law'" |
---|
185 | |
---|
186 | if range == 'high': |
---|
187 | if function != 'power_law': |
---|
188 | raise ValueError, "Extrapolation only allows a power law at high Q" |
---|
189 | self._high_extrapolation_npts = npts |
---|
190 | self._high_extrapolation_power = power |
---|
191 | else: |
---|
192 | if function == 'power_law': |
---|
193 | self._low_extrapolation_function = power_law |
---|
194 | else: |
---|
195 | self._low_extrapolation_function = guinier |
---|
196 | self._low_extrapolation_npts = npts |
---|
197 | self._low_extrapolation_power = power |
---|
198 | |
---|
199 | def _get_data(self, data): |
---|
200 | """ |
---|
201 | @note this function must be call before computing any type |
---|
202 | of invariant |
---|
203 | @return data= self._scale *data - self._background |
---|
204 | """ |
---|
205 | if not issubclass(data.__class__, LoaderData1D): |
---|
206 | #Process only data that inherited from DataLoader.Data_info.Data1D |
---|
207 | raise ValueError,"Data must be of type DataLoader.Data1D" |
---|
208 | return self._scale * data - self._background |
---|
209 | |
---|
210 | def _fit(self, function, qmin=Q_MINIMUM, qmax=Q_MAXIMUM): |
---|
211 | """ |
---|
212 | fit data with function using |
---|
213 | data= self._get_data() |
---|
214 | fx= Functor(data , function) |
---|
215 | y = data.y |
---|
216 | out, cov_x = linalg.lstsq(y,fx) |
---|
217 | @param qmin: data first q value to consider during the fit |
---|
218 | @param qmax: data last q value to consider during the fit |
---|
219 | @param function: the function to use during the fit |
---|
220 | @return a: the scale of the function |
---|
221 | @return b: the other parameter of the function for guinier will be radius |
---|
222 | for power_law will be the power value |
---|
223 | """ |
---|
224 | if function.__name__ == "guinier": |
---|
225 | fit_x = self._data.x * self._data.x |
---|
226 | qmin = qmin**2 |
---|
227 | qmax = qmax**2 |
---|
228 | fit_y = [math.log(y) for y in self._data.y] |
---|
229 | elif function.__name__ == "power_law": |
---|
230 | fit_x = [math.log(x) for x in self._data.x] |
---|
231 | qmin = math.log(qmin) |
---|
232 | qmax = math.log(qmax) |
---|
233 | fit_y = [math.log(y) for y in self._data.y] |
---|
234 | else: |
---|
235 | raise ValueError("Unknown function used to fit %s"%function.__name__) |
---|
236 | |
---|
237 | fit_data = LoaderData1D(x=fit_x, y=fit_y) |
---|
238 | fit_data.dxl = self._data.dxl |
---|
239 | fit_data.dxw = self._data.dxw |
---|
240 | |
---|
241 | functor = FitFunctor(data=fit_data, function= function) |
---|
242 | functor.set_fit_range(qmin=qmin, qmax=qmax) |
---|
243 | a, b = functor.fit() |
---|
244 | |
---|
245 | if function.__name__ == "guinier": |
---|
246 | # b is the radius value of the guinier function |
---|
247 | print "b",b |
---|
248 | b = math.sqrt(-3 * b) |
---|
249 | |
---|
250 | # a is the scale of the guinier function |
---|
251 | a = math.exp(a) |
---|
252 | return a, math.fabs(b) |
---|
253 | |
---|
254 | def _get_qstar(self, data): |
---|
255 | """ |
---|
256 | Compute invariant for data |
---|
257 | @param data: data to use to compute invariant. |
---|
258 | |
---|
259 | """ |
---|
260 | if data is None: |
---|
261 | return 0 |
---|
262 | if data.is_slit_smeared(): |
---|
263 | return self._get_qstar_smear(data) |
---|
264 | else: |
---|
265 | return self._get_qstar_unsmear(data) |
---|
266 | |
---|
267 | |
---|
268 | def get_qstar(self, extrapolation=None): |
---|
269 | """ |
---|
270 | Compute the invariant of the local copy of data. |
---|
271 | Implementation: |
---|
272 | if slit smear: |
---|
273 | qstar_0 = self._get_qstar_smear() |
---|
274 | else: |
---|
275 | qstar_0 = self._get_qstar_unsmear() |
---|
276 | if extrapolation is None: |
---|
277 | return qstar_0 |
---|
278 | if extrapolation==low: |
---|
279 | return qstar_0 + self._get_qstar_low() |
---|
280 | elif extrapolation==high: |
---|
281 | return qstar_0 + self._get_qstar_high() |
---|
282 | elif extrapolation==both: |
---|
283 | return qstar_0 + self._get_qstar_low() + self._get_qstar_high() |
---|
284 | |
---|
285 | @param extrapolation: string to apply optional extrapolation |
---|
286 | @return q_star: invariant of the data within data's q range |
---|
287 | |
---|
288 | @warning: When using setting data to Data1D , the user is responsible of |
---|
289 | checking that the scale and the background are properly apply to the data |
---|
290 | |
---|
291 | @warning: if error occur self._get_qstar_low() or self._get_qstar_low() |
---|
292 | their returned value will be ignored |
---|
293 | """ |
---|
294 | qstar_0 = self._get_qstar(data=self._data) |
---|
295 | |
---|
296 | if extrapolation is None: |
---|
297 | self._qstar = qstar_0 |
---|
298 | return self._qstar |
---|
299 | # Compute invariant plus invaraint of extrapolated data |
---|
300 | extrapolation = extrapolation.lower() |
---|
301 | if extrapolation == "low": |
---|
302 | self._qstar = qstar_0 + self._get_qstar_low() |
---|
303 | return self._qstar |
---|
304 | elif extrapolation == "high": |
---|
305 | self._qstar = qstar_0 + self._get_qstar_high() |
---|
306 | return self._qstar |
---|
307 | elif extrapolation == "both": |
---|
308 | self._qstar = qstar_0 + self._get_qstar_low() + self._get_qstar_high() |
---|
309 | return self._qstar |
---|
310 | |
---|
311 | def _get_qstar_unsmear(self, data): |
---|
312 | """ |
---|
313 | Compute invariant for pinhole data. |
---|
314 | This invariant is given by: |
---|
315 | |
---|
316 | q_star = x0**2 *y0 *dx0 +x1**2 *y1 *dx1 |
---|
317 | + ..+ xn**2 *yn *dxn |
---|
318 | |
---|
319 | where n >= len(data.x)-1 |
---|
320 | dxi = 1/2*(xi+1 - xi) + (xi - xi-1) |
---|
321 | dx0 = (x1 - x0)/2 |
---|
322 | dxn = xn - xn-1 |
---|
323 | @param data: the data to use to compute invariant. |
---|
324 | @return q_star: invariant value for pinhole data. q_star > 0 |
---|
325 | """ |
---|
326 | if len(data.x) <= 1 or len(data.y) <= 1 or len(data.x)!= len(data.y): |
---|
327 | msg = "Length x and y must be equal" |
---|
328 | msg += " and greater than 1; got x=%s, y=%s"%(len(data.x), len(data.y)) |
---|
329 | raise ValueError, msg |
---|
330 | else: |
---|
331 | n = len(data.x)- 1 |
---|
332 | #compute the first delta q |
---|
333 | dx0 = (data.x[1] - data.x[0])/2 |
---|
334 | #compute the last delta q |
---|
335 | dxn = data.x[n] - data.x[n-1] |
---|
336 | sum = 0 |
---|
337 | sum += data.x[0] * data.x[0] * data.y[0] * dx0 |
---|
338 | sum += data.x[n] * data.x[n] * data.y[n] * dxn |
---|
339 | |
---|
340 | if len(data.x) == 2: |
---|
341 | return sum |
---|
342 | else: |
---|
343 | #iterate between for element different from the first and the last |
---|
344 | for i in xrange(1, n-1): |
---|
345 | dxi = (data.x[i+1] - data.x[i-1])/2 |
---|
346 | sum += data.x[i] * data.x[i] * data.y[i] * dxi |
---|
347 | return sum |
---|
348 | |
---|
349 | def _get_qstar_smear(self, data): |
---|
350 | """ |
---|
351 | Compute invariant for slit-smeared data. |
---|
352 | This invariant is given by: |
---|
353 | q_star = x0*dxl *y0*dx0 + x1*dxl *y1 *dx1 |
---|
354 | + ..+ xn*dxl *yn *dxn |
---|
355 | where n >= len(data.x)-1 |
---|
356 | dxi = 1/2*(xi+1 - xi) + (xi - xi-1) |
---|
357 | dx0 = x0+ (x1 - x0)/2 |
---|
358 | dxn = xn - xn-1 |
---|
359 | dxl: slit smear value |
---|
360 | |
---|
361 | @return q_star: invariant value for slit smeared data. |
---|
362 | """ |
---|
363 | if not data.is_slit_smeared(): |
---|
364 | msg = "_get_qstar_smear need slit smear data " |
---|
365 | msg += "Hint :dxl= %s , dxw= %s"%(str(data.dxl), str(data.dxw)) |
---|
366 | raise ValueError, msg |
---|
367 | |
---|
368 | if len(data.x) <= 1 or len(data.y) <= 1 or len(data.x) != len(data.y)\ |
---|
369 | or len(data.x)!= len(data.dxl): |
---|
370 | msg = "x, dxl, and y must be have the same length and greater than 1" |
---|
371 | raise ValueError, msg |
---|
372 | else: |
---|
373 | n = len(data.x)-1 |
---|
374 | #compute the first delta |
---|
375 | dx0 = (data.x[1] + data.x[0])/2 |
---|
376 | #compute the last delta |
---|
377 | dxn = data.x[n] - data.x[n-1] |
---|
378 | sum = 0 |
---|
379 | sum += data.x[0] * data.dxl[0] * data.y[0] * dx0 |
---|
380 | sum += data.x[n] * data.dxl[n] * data.y[n] * dxn |
---|
381 | |
---|
382 | if len(data.x)==2: |
---|
383 | return sum |
---|
384 | else: |
---|
385 | #iterate between for element different from the first and the last |
---|
386 | for i in xrange(1, n-1): |
---|
387 | dxi = (data.x[i+1] - data.x[i-1])/2 |
---|
388 | sum += data.x[i] * data.dxl[i] * data.y[i] * dxi |
---|
389 | return sum |
---|
390 | |
---|
391 | def _get_qstar_uncertainty(self, data=None): |
---|
392 | """ |
---|
393 | Compute uncertainty of invariant value |
---|
394 | Implementation: |
---|
395 | if data is None: |
---|
396 | data = self.data |
---|
397 | |
---|
398 | if data.slit smear: |
---|
399 | return self._get_qstar_smear_uncertainty(data) |
---|
400 | else: |
---|
401 | return self._get_qstar_unsmear_uncertainty(data) |
---|
402 | |
---|
403 | @param: data use to compute the invariant which allow uncertainty |
---|
404 | computation. |
---|
405 | @return: uncertainty |
---|
406 | """ |
---|
407 | if data is None: |
---|
408 | data = self.data |
---|
409 | |
---|
410 | if data.is_slit_smeared(): |
---|
411 | return self._get_qstar_smear_uncertainty(data) |
---|
412 | else: |
---|
413 | return self._get_qstar_unsmear_uncertainty(data) |
---|
414 | |
---|
415 | def _get_qstar_unsmear_uncertainty(self, data=None): |
---|
416 | """ |
---|
417 | Compute invariant uncertainty with with pinhole data. |
---|
418 | This uncertainty is given as follow: |
---|
419 | dq_star = math.sqrt[(x0**2*(dy0)*dx0)**2 + |
---|
420 | (x1**2 *(dy1)*dx1)**2 + ..+ (xn**2 *(dyn)*dxn)**2 ] |
---|
421 | where n >= len(data.x)-1 |
---|
422 | dxi = 1/2*(xi+1 - xi) + (xi - xi-1) |
---|
423 | dx0 = x0+ (x1 - x0)/2 |
---|
424 | dxn = xn - xn-1 |
---|
425 | dyn: error on dy |
---|
426 | |
---|
427 | @param data: |
---|
428 | note: if data doesn't contain dy assume dy= math.sqrt(data.y) |
---|
429 | """ |
---|
430 | if data is None: |
---|
431 | data = self.data |
---|
432 | |
---|
433 | if len(data.x) <= 1 or len(data.y) <= 1 or \ |
---|
434 | len(self.data.x) != len(self.data.y): |
---|
435 | msg = "Length of data.x and data.y must be equal" |
---|
436 | msg += " and greater than 1; got x=%s, y=%s"%(len(data.x), |
---|
437 | len(data.y)) |
---|
438 | raise ValueError, msg |
---|
439 | else: |
---|
440 | #Create error for data without dy error |
---|
441 | if (data.dy is None) or (not data.dy): |
---|
442 | dy = math.sqrt(y) |
---|
443 | else: |
---|
444 | dy = data.dy |
---|
445 | |
---|
446 | n = len(data.x) - 1 |
---|
447 | #compute the first delta |
---|
448 | dx0 = (data.x[1] - data.x[0])/2 |
---|
449 | #compute the last delta |
---|
450 | dxn= data.x[n] - data.x[n-1] |
---|
451 | sum = 0 |
---|
452 | sum += (data.x[0] * data.x[0] * dy[0] * dx0)**2 |
---|
453 | sum += (data.x[n] * data.x[n] * dy[n] * dxn)**2 |
---|
454 | if len(data.x) == 2: |
---|
455 | return math.sqrt(sum) |
---|
456 | else: |
---|
457 | #iterate between for element different from the first and the last |
---|
458 | for i in xrange(1, n-1): |
---|
459 | dxi = (data.x[i+1] - data.x[i-1])/2 |
---|
460 | sum += (data.x[i] * data.x[i] * dy[i] * dxi)**2 |
---|
461 | return math.sqrt(sum) |
---|
462 | |
---|
463 | def _get_qstar_smear_uncertainty(self): |
---|
464 | """ |
---|
465 | Compute invariant uncertainty with slit smeared data. |
---|
466 | This uncertainty is given as follow: |
---|
467 | dq_star = x0*dxl *dy0 *dx0 + x1*dxl *dy1 *dx1 |
---|
468 | + ..+ xn*dxl *dyn *dxn |
---|
469 | where n >= len(data.x)-1 |
---|
470 | dxi = 1/2*(xi+1 - xi) + (xi - xi-1) |
---|
471 | dx0 = x0+ (x1 - x0)/2 |
---|
472 | dxn = xn - xn-1 |
---|
473 | dxl: slit smearing value |
---|
474 | dyn : error on dy |
---|
475 | @param data: data of type Data1D where the scale is applied |
---|
476 | and the background is subtracted. |
---|
477 | |
---|
478 | note: if data doesn't contain dy assume dy= math.sqrt(data.y) |
---|
479 | """ |
---|
480 | if data is None: |
---|
481 | data = self._data |
---|
482 | |
---|
483 | if not data.is_slit_smeared(): |
---|
484 | msg = "_get_qstar_smear_uncertainty need slit smear data " |
---|
485 | msg += "Hint :dxl= %s , dxw= %s"%(str(data.dxl), str(data.dxw)) |
---|
486 | raise ValueError, msg |
---|
487 | |
---|
488 | if len(data.x) <= 1 or len(data.y) <= 1 or len(data.x) != len(data.y)\ |
---|
489 | or len(data.x) != len(data.dxl): |
---|
490 | msg = "x, dxl, and y must be have the same length and greater than 1" |
---|
491 | raise ValueError, msg |
---|
492 | else: |
---|
493 | #Create error for data without dy error |
---|
494 | if (data.dy is None) or (not data.dy): |
---|
495 | dy = math.sqrt(y) |
---|
496 | else: |
---|
497 | dy = data.dy |
---|
498 | |
---|
499 | n = len(data.x) - 1 |
---|
500 | #compute the first delta |
---|
501 | dx0 = (data.x[1] - data.x[0])/2 |
---|
502 | #compute the last delta |
---|
503 | dxn = data.x[n] - data.x[n-1] |
---|
504 | sum = 0 |
---|
505 | sum += (data.x[0] * data.dxl[0] * dy[0] * dx0)**2 |
---|
506 | sum += (data.x[n] * data.dxl[n] * dy[n] * dxn)**2 |
---|
507 | |
---|
508 | if len(data.x) == 2: |
---|
509 | return math.sqrt(sum) |
---|
510 | else: |
---|
511 | #iterate between for element different from the first and the last |
---|
512 | for i in xrange(1, n-1): |
---|
513 | dxi = (data.x[i+1] - data.x[i-1])/2 |
---|
514 | sum += (data.x[i] * data.dxl[i] * dy[i] * dxi)**2 |
---|
515 | return math.sqrt(sum) |
---|
516 | |
---|
517 | def get_surface(self,contrast, porod_const): |
---|
518 | """ |
---|
519 | Compute the surface of the data. |
---|
520 | |
---|
521 | Implementation: |
---|
522 | V= self.get_volume_fraction(contrast) |
---|
523 | |
---|
524 | Compute the surface given by: |
---|
525 | surface = (2*pi *V(1- V)*porod_const)/ q_star |
---|
526 | |
---|
527 | @param contrast: contrast value to compute the volume |
---|
528 | @param porod_const: Porod constant to compute the surface |
---|
529 | @return: specific surface |
---|
530 | """ |
---|
531 | #Check whether we have Q star |
---|
532 | if self._qstar is None: |
---|
533 | self._qstar = self.get_star() |
---|
534 | if self._qstar == 0: |
---|
535 | raise RuntimeError("Cannot compute surface, invariant value is zero") |
---|
536 | # Compute the volume |
---|
537 | volume = self.get_volume_fraction(contrast) |
---|
538 | return 2 * math.pi * volume *(1 - volume) * float(porod_const)/self._qstar |
---|
539 | |
---|
540 | |
---|
541 | def get_volume_fraction(self, contrast): |
---|
542 | """ |
---|
543 | Compute volume fraction is deduced as follow: |
---|
544 | |
---|
545 | q_star = 2*(pi*contrast)**2* volume( 1- volume) |
---|
546 | for k = 10^(-8)*q_star/(2*(pi*|contrast|)**2) |
---|
547 | we get 2 values of volume: |
---|
548 | with 1 - 4 * k >= 0 |
---|
549 | volume1 = (1- sqrt(1- 4*k))/2 |
---|
550 | volume2 = (1+ sqrt(1- 4*k))/2 |
---|
551 | |
---|
552 | q_star: the invariant value included extrapolation is applied |
---|
553 | unit 1/A^(3)*1/cm |
---|
554 | q_star = self.get_qstar() |
---|
555 | |
---|
556 | the result returned will be 0<= volume <= 1 |
---|
557 | |
---|
558 | @param contrast: contrast value provides by the user of type float. |
---|
559 | contrast unit is 1/A^(2)= 10^(16)cm^(2) |
---|
560 | @return: volume fraction |
---|
561 | @note: volume fraction must have no unit |
---|
562 | """ |
---|
563 | if contrast < 0: |
---|
564 | raise ValueError, "contrast must be greater than zero" |
---|
565 | |
---|
566 | if self._qstar is None: |
---|
567 | self._qstar = self.get_qstar() |
---|
568 | |
---|
569 | if self._qstar < 0: |
---|
570 | raise RuntimeError, "invariant must be greater than zero" |
---|
571 | |
---|
572 | print "self._qstar",self._qstar |
---|
573 | # Compute intermediate constant |
---|
574 | k = 1.e-8 * self._qstar/(2 * (math.pi * math.fabs(float(contrast)))**2) |
---|
575 | #Check discriminant value |
---|
576 | discrim = 1 - 4 * k |
---|
577 | print "discrim",k,discrim |
---|
578 | # Compute volume fraction |
---|
579 | if discrim < 0: |
---|
580 | raise RuntimeError, "could not compute the volume fraction: negative discriminant" |
---|
581 | elif discrim ==0: |
---|
582 | return 1/2 |
---|
583 | else: |
---|
584 | volume1 = 0.5 *(1 - math.sqrt(discrim)) |
---|
585 | volume2 = 0.5 *(1 + math.sqrt(discrim)) |
---|
586 | |
---|
587 | if 0 <= volume1 and volume1 <= 1: |
---|
588 | return volume1 |
---|
589 | elif 0 <= volume2 and volume2 <= 1: |
---|
590 | return volume2 |
---|
591 | raise RuntimeError, "could not compute the volume fraction: inconsistent results" |
---|
592 | |
---|
593 | def _get_qstar_low(self): |
---|
594 | """ |
---|
595 | Compute the invariant for extrapolated data at low q range. |
---|
596 | |
---|
597 | Implementation: |
---|
598 | data = self.get_extra_data_low() |
---|
599 | return self._get_qstar() |
---|
600 | |
---|
601 | @return q_star: the invariant for data extrapolated at low q. |
---|
602 | """ |
---|
603 | data = self._get_extra_data_low() |
---|
604 | return self._get_qstar(data=data) |
---|
605 | |
---|
606 | def _get_qstar_high(self): |
---|
607 | """ |
---|
608 | Compute the invariant for extrapolated data at high q range. |
---|
609 | |
---|
610 | Implementation: |
---|
611 | data = self.get_extra_data_high() |
---|
612 | return self._get_qstar() |
---|
613 | |
---|
614 | @return q_star: the invariant for data extrapolated at high q. |
---|
615 | """ |
---|
616 | data = self._get_extra_data_high() |
---|
617 | return self._get_qstar( data=data) |
---|
618 | |
---|
619 | def _get_extra_data_low(self): |
---|
620 | """ |
---|
621 | This method creates a new data set from the invariant calculator. |
---|
622 | |
---|
623 | It will use the extrapolation parameters kept as private data members. |
---|
624 | |
---|
625 | self._low_extrapolation_npts is the number of data points to use in to fit. |
---|
626 | self._low_extrapolation_function will be used as the fit function. |
---|
627 | |
---|
628 | |
---|
629 | |
---|
630 | It takes npts first points of data, fits them with a given model |
---|
631 | then uses the new parameters resulting from the fit to create a new data set. |
---|
632 | |
---|
633 | The new data first point is Q_MINIMUM. |
---|
634 | |
---|
635 | The last point of the new data is the first point of the original data. |
---|
636 | the number of q points of this data is INTEGRATION_NSTEPS. |
---|
637 | |
---|
638 | @return: a new data of type Data1D |
---|
639 | """ |
---|
640 | # Data boundaries for fiiting |
---|
641 | qmin = self._data.x[0] |
---|
642 | qmax = self._data.x[self._low_extrapolation_npts] |
---|
643 | |
---|
644 | try: |
---|
645 | # fit the data with a model to get the appropriate parameters |
---|
646 | a, b = self._fit(function=self._low_extrapolation_function, |
---|
647 | qmin=qmin, qmax=qmax) |
---|
648 | except: |
---|
649 | raise |
---|
650 | return None |
---|
651 | |
---|
652 | #create new Data1D to compute the invariant |
---|
653 | new_x = numpy.linspace(start=Q_MINIMUM, |
---|
654 | stop=qmin, |
---|
655 | num=INTEGRATION_NSTEPS, |
---|
656 | endpoint=True) |
---|
657 | new_y = self._low_extrapolation_function(x=new_x, |
---|
658 | scale=a, |
---|
659 | radius=b) |
---|
660 | dxl = None |
---|
661 | dxw = None |
---|
662 | if self._data.dxl is not None: |
---|
663 | dxl = numpy.ones(1, INTEGRATION_NSTEPS) |
---|
664 | dxl = dxl * self._data.dxl[0] |
---|
665 | if self._data.dxw is not None: |
---|
666 | dwl = numpy.ones(1, INTEGRATION_NSTEPS) |
---|
667 | dwl = dwl * self._data.dwl[0] |
---|
668 | |
---|
669 | data_min = LoaderData1D(x=new_x, y=new_y) |
---|
670 | data_min.dxl = dxl |
---|
671 | data_min.dxw = dxw |
---|
672 | self._data.clone_without_data( clone= data_min) |
---|
673 | |
---|
674 | return data_min |
---|
675 | |
---|
676 | def _get_extra_data_high(self): |
---|
677 | """ |
---|
678 | This method creates a new data from the invariant calculator. |
---|
679 | |
---|
680 | It takes npts last points of data, fits them with a given model |
---|
681 | (for this function only power_law will be use), then uses |
---|
682 | the new parameters resulting from the fit to create a new data set. |
---|
683 | The first point is the last point of data. |
---|
684 | The last point of the new data is Q_MAXIMUM. |
---|
685 | The number of q points of this data is INTEGRATION_NSTEPS. |
---|
686 | |
---|
687 | |
---|
688 | @return: a new data of type Data1D |
---|
689 | """ |
---|
690 | # Data boundaries for fiiting |
---|
691 | x_len = len(self._data.x) - 1 |
---|
692 | qmin = self._data.x[x_len - self._high_extrapolation_npts] |
---|
693 | qmax = self._data.x[x_len] |
---|
694 | |
---|
695 | try: |
---|
696 | # fit the data with a model to get the appropriate parameters |
---|
697 | a, b = self._fit(function=self._high_extrapolation_function, |
---|
698 | qmin=qmin, qmax=qmax) |
---|
699 | except: |
---|
700 | raise |
---|
701 | return None |
---|
702 | |
---|
703 | #create new Data1D to compute the invariant |
---|
704 | new_x = numpy.linspace(start=qmax, |
---|
705 | stop=Q_MAXIMUM, |
---|
706 | num=INTEGRATION_NSTEPS, |
---|
707 | endpoint=True) |
---|
708 | new_y = self._high_extrapolation_function(x=new_x, |
---|
709 | scale=a, |
---|
710 | power=b) |
---|
711 | dxl = None |
---|
712 | dxw = None |
---|
713 | if self._data.dxl is not None: |
---|
714 | dxl = numpy.ones(1, INTEGRATION_NSTEPS) |
---|
715 | dxl = dxl * self._data.dxl[0] |
---|
716 | if self._data.dxw is not None: |
---|
717 | dwl = numpy.ones(1, INTEGRATION_NSTEPS) |
---|
718 | dwl = dwl * self._data.dwl[0] |
---|
719 | |
---|
720 | data_max = LoaderData1D(x=new_x, y=new_y) |
---|
721 | data_max.dxl = dxl |
---|
722 | data_max.dxw = dxw |
---|
723 | self._data.clone_without_data( clone=data_max) |
---|
724 | |
---|
725 | return data_max |
---|
726 | |
---|
727 | def get_qstar_with_error(self, extrapolation=None): |
---|
728 | """ |
---|
729 | Compute the invariant uncertainty. |
---|
730 | This uncertainty computation depends on whether or not the data is |
---|
731 | smeared. |
---|
732 | @return: invariant, the invariant uncertainty |
---|
733 | return self._get_qstar(), self._get_qstar_smear_uncertainty() |
---|
734 | """ |
---|
735 | if self._qstar is None: |
---|
736 | self._qstar = self.get_qstar(extrapolation=extrapolation) |
---|
737 | if self._qstar_err is None: |
---|
738 | self._qstar_err = self._get_qstar_smear_uncertainty() |
---|
739 | |
---|
740 | return self._qstar, self._qstar_err |
---|
741 | |
---|
742 | def get_volume_fraction_with_error(self, contrast): |
---|
743 | """ |
---|
744 | Compute uncertainty on volume value as well as the volume fraction |
---|
745 | This uncertainty is given by the following equation: |
---|
746 | dV = 0.5 * (4*k* dq_star) /(2* math.sqrt(1-k* q_star)) |
---|
747 | |
---|
748 | for k = 10^(8)*q_star/(2*(pi*|contrast|)**2) |
---|
749 | |
---|
750 | q_star: the invariant value including extrapolated value if existing |
---|
751 | dq_star: the invariant uncertainty |
---|
752 | dV: the volume uncertainty |
---|
753 | @param contrast: contrast value |
---|
754 | @return: V, dV = self.get_volume_fraction_with_error(contrast), dV |
---|
755 | """ |
---|
756 | self._qstar, self._qstar_err = self.get_qstar_with_error() |
---|
757 | |
---|
758 | volume = self.get_volume_fraction(contrast) |
---|
759 | if self._qstar < 0: |
---|
760 | raise ValueError, "invariant must be greater than zero" |
---|
761 | |
---|
762 | k = 1.e-8 * self._qstar /(2*(math.pi* math.fabs(float(contrast)))**2) |
---|
763 | #check value inside the sqrt function |
---|
764 | value = 1 - k * self._qstar |
---|
765 | if (1 - k * self._qstar) <= 0: |
---|
766 | raise ValueError, "Cannot compute incertainty on volume" |
---|
767 | # Compute uncertainty |
---|
768 | uncertainty = (0.5 * 4 * k * self._qstar_err)/(2 * math.sqrt(1- k * self._qstar)) |
---|
769 | |
---|
770 | return volume, math.fabs(uncertainty) |
---|
771 | |
---|
772 | def get_surface_with_error(self, contrast, porod_const): |
---|
773 | """ |
---|
774 | Compute uncertainty of the surface value as well as thesurface value |
---|
775 | this uncertainty is given as follow: |
---|
776 | |
---|
777 | dS = porod_const *2*pi[( dV -2*V*dV)/q_star |
---|
778 | + dq_star(v-v**2) |
---|
779 | |
---|
780 | q_star: the invariant value including extrapolated value if existing |
---|
781 | dq_star: the invariant uncertainty |
---|
782 | V: the volume fraction value |
---|
783 | dV: the volume uncertainty |
---|
784 | |
---|
785 | @param contrast: contrast value |
---|
786 | @param porod_const: porod constant value |
---|
787 | @return S, dS: the surface, with its uncertainty |
---|
788 | """ |
---|
789 | v, dv = self.get_volume_fraction_with_error(contrast) |
---|
790 | self._qstar, self._qstar_err = self.get_qstar_with_error() |
---|
791 | if self._qstar <= 0: |
---|
792 | raise ValueError, "invariant must be greater than zero" |
---|
793 | ds = porod_const * 2 * math.pi * (( dv - 2 * v * dv)/ self._qstar\ |
---|
794 | + self._qstar_err * ( v - v**2)) |
---|
795 | s = self.get_surface(contrast=contrast, porod_const=porod_const) |
---|
796 | return s, ds |
---|