1 | #GSASIImath - major mathematics routines |
---|
2 | ########### SVN repository information ################### |
---|
3 | # $Date: 2012-01-13 11:48:53 -0600 (Fri, 13 Jan 2012) $ |
---|
4 | # $Author: vondreele & toby $ |
---|
5 | # $Revision: 451 $ |
---|
6 | # $URL: https://subversion.xor.aps.anl.gov/pyGSAS/trunk/GSASIImath.py $ |
---|
7 | # $Id: GSASIImath.py 451 2012-01-13 17:48:53Z vondreele $ |
---|
8 | ########### SVN repository information ################### |
---|
9 | import sys |
---|
10 | import os |
---|
11 | import os.path as ospath |
---|
12 | import numpy as np |
---|
13 | import numpy.linalg as nl |
---|
14 | import cPickle |
---|
15 | import time |
---|
16 | import math |
---|
17 | import GSASIIpath |
---|
18 | import GSASIIspc as G2spc |
---|
19 | import scipy.optimize as so |
---|
20 | import scipy.linalg as sl |
---|
21 | |
---|
22 | sind = lambda x: np.sin(x*np.pi/180.) |
---|
23 | cosd = lambda x: np.cos(x*np.pi/180.) |
---|
24 | tand = lambda x: np.tan(x*np.pi/180.) |
---|
25 | asind = lambda x: 180.*np.arcsin(x)/np.pi |
---|
26 | acosd = lambda x: 180.*np.arccos(x)/np.pi |
---|
27 | atan2d = lambda y,x: 180.*np.arctan2(y,x)/np.pi |
---|
28 | |
---|
29 | def HessianLSQ(func,x0,Hess,args=(),ftol=1.49012e-8,xtol=1.49012e-8, maxcyc=0): |
---|
30 | |
---|
31 | """ |
---|
32 | Minimize the sum of squares of a set of equations. |
---|
33 | |
---|
34 | :: |
---|
35 | |
---|
36 | Nobs |
---|
37 | x = arg min(sum(func(y)**2,axis=0)) |
---|
38 | y=0 |
---|
39 | |
---|
40 | Parameters |
---|
41 | ---------- |
---|
42 | func : callable |
---|
43 | should take at least one (possibly length N vector) argument and |
---|
44 | returns M floating point numbers. |
---|
45 | x0 : ndarray |
---|
46 | The starting estimate for the minimization of length N |
---|
47 | Hess : callable |
---|
48 | A required function or method to compute the weighted vector and Hessian for func. |
---|
49 | It must be a symmetric NxN array |
---|
50 | args : tuple |
---|
51 | Any extra arguments to func are placed in this tuple. |
---|
52 | ftol : float |
---|
53 | Relative error desired in the sum of squares. |
---|
54 | xtol : float |
---|
55 | Relative error desired in the approximate solution. |
---|
56 | maxcyc : int |
---|
57 | The maximum number of cycles of refinement to execute, if -1 refine |
---|
58 | until other limits are met (ftol, xtol) |
---|
59 | |
---|
60 | Returns |
---|
61 | ------- |
---|
62 | x : ndarray |
---|
63 | The solution (or the result of the last iteration for an unsuccessful |
---|
64 | call). |
---|
65 | cov_x : ndarray |
---|
66 | Uses the fjac and ipvt optional outputs to construct an |
---|
67 | estimate of the jacobian around the solution. ``None`` if a |
---|
68 | singular matrix encountered (indicates very flat curvature in |
---|
69 | some direction). This matrix must be multiplied by the |
---|
70 | residual standard deviation to get the covariance of the |
---|
71 | parameter estimates -- see curve_fit. |
---|
72 | infodict : dict |
---|
73 | a dictionary of optional outputs with the key s:: |
---|
74 | |
---|
75 | - 'fvec' : the function evaluated at the output |
---|
76 | |
---|
77 | |
---|
78 | Notes |
---|
79 | ----- |
---|
80 | |
---|
81 | """ |
---|
82 | |
---|
83 | x0 = np.array(x0, ndmin=1) #might be redundant? |
---|
84 | n = len(x0) |
---|
85 | if type(args) != type(()): |
---|
86 | args = (args,) |
---|
87 | |
---|
88 | icycle = 0 |
---|
89 | One = np.ones((n,n)) |
---|
90 | lam = 0.001 |
---|
91 | lamMax = lam |
---|
92 | nfev = 0 |
---|
93 | while icycle < maxcyc: |
---|
94 | lamMax = max(lamMax,lam) |
---|
95 | M = func(x0,*args) |
---|
96 | nfev += 1 |
---|
97 | chisq0 = np.sum(M**2) |
---|
98 | Yvec,Amat = Hess(x0,*args) |
---|
99 | Adiag = np.sqrt(np.diag(Amat)) |
---|
100 | if 0.0 in Adiag: #hard singularity in matrix |
---|
101 | psing = list(np.where(Adiag == 0.)[0]) |
---|
102 | return [x0,None,{'num cyc':icycle,'fvec':M,'nfev':nfev,'lamMax':lamMax,'psing':psing}] |
---|
103 | Anorm = np.outer(Adiag,Adiag) |
---|
104 | Yvec /= Adiag |
---|
105 | Amat /= Anorm |
---|
106 | while True: |
---|
107 | Lam = np.eye(Amat.shape[0])*lam |
---|
108 | Amatlam = Amat*(One+Lam) |
---|
109 | try: |
---|
110 | Xvec = nl.solve(Amatlam,Yvec) |
---|
111 | except LinAlgError: |
---|
112 | psing = list(np.where(np.diag(nl.gr(Amatlam)[1]) < 1.e-14)[0]) |
---|
113 | return [x0,None,{'num cyc':icycle,'fvec':M,'nfev':nfev,'lamMax':lamMax,'psing':psing}] |
---|
114 | Xvec /= Adiag |
---|
115 | M2 = func(x0+Xvec,*args) |
---|
116 | nfev += 1 |
---|
117 | chisq1 = np.sum(M2**2) |
---|
118 | if chisq1 > chisq0: |
---|
119 | lam *= 10. |
---|
120 | else: |
---|
121 | x0 += Xvec |
---|
122 | lam /= 10. |
---|
123 | break |
---|
124 | if (chisq0-chisq1)/chisq0 < ftol: |
---|
125 | break |
---|
126 | icycle += 1 |
---|
127 | M = func(x0,*args) |
---|
128 | nfev += 1 |
---|
129 | Yvec,Amat = Hess(x0,*args) |
---|
130 | try: |
---|
131 | Bmat = nl.inv(Amat) |
---|
132 | return [x0,Bmat,{'num cyc':icycle,'fvec':M,'nfev':nfev,'lamMax':lamMax,'psing':[]}] |
---|
133 | except nl.LinAlgError: |
---|
134 | psing = [] |
---|
135 | if maxcyc: |
---|
136 | psing = list(np.where(np.diag(nl.qr(Amat)[1]) < 1.e-14)[0]) |
---|
137 | return [x0,None,{'num cyc':icycle,'fvec':M,'nfev':nfev,'lamMax':lamMax,'psing':psing}] |
---|
138 | |
---|
139 | def getVCov(varyNames,varyList,covMatrix): |
---|
140 | vcov = np.zeros((len(varyNames),len(varyNames))) |
---|
141 | for i1,name1 in enumerate(varyNames): |
---|
142 | for i2,name2 in enumerate(varyNames): |
---|
143 | try: |
---|
144 | vcov[i1][i2] = covMatrix[varyList.index(name1)][varyList.index(name2)] |
---|
145 | except ValueError: |
---|
146 | vcov[i1][i2] = 0.0 |
---|
147 | return vcov |
---|
148 | |
---|
149 | def getDistDerv(Oxyz,Txyz,Amat,Tunit,Top,SGData): |
---|
150 | |
---|
151 | def calcDist(Ox,Tx,U,inv,C,M,T,Amat): |
---|
152 | TxT = inv*(np.inner(M,Tx)+T)+C+U |
---|
153 | return np.sqrt(np.sum(np.inner(Amat,(TxT-Ox))**2)) |
---|
154 | |
---|
155 | inv = Top/abs(Top) |
---|
156 | cent = abs(Top)/100 |
---|
157 | op = abs(Top)%100-1 |
---|
158 | M,T = SGData['SGOps'][op] |
---|
159 | C = SGData['SGCen'][cent] |
---|
160 | dx = .00001 |
---|
161 | deriv = np.zeros(6) |
---|
162 | for i in [0,1,2]: |
---|
163 | Oxyz[i] += dx |
---|
164 | d0 = calcDist(Oxyz,Txyz,Tunit,inv,C,M,T,Amat) |
---|
165 | Oxyz[i] -= 2*dx |
---|
166 | deriv[i] = (calcDist(Oxyz,Txyz,Tunit,inv,C,M,T,Amat)-d0)/(2.*dx) |
---|
167 | Oxyz[i] += dx |
---|
168 | Txyz[i] += dx |
---|
169 | d0 = calcDist(Oxyz,Txyz,Tunit,inv,C,M,T,Amat) |
---|
170 | Txyz[i] -= 2*dx |
---|
171 | deriv[i+3] = (calcDist(Oxyz,Txyz,Tunit,inv,C,M,T,Amat)-d0)/(2.*dx) |
---|
172 | Txyz[i] += dx |
---|
173 | return deriv |
---|
174 | |
---|
175 | def getAngSig(VA,VB,Amat,SGData,covData={}): |
---|
176 | |
---|
177 | def calcVec(Ox,Tx,U,inv,C,M,T,Amat): |
---|
178 | TxT = inv*(np.inner(M,Tx)+T)+C |
---|
179 | TxT = G2spc.MoveToUnitCell(TxT)+U |
---|
180 | return np.inner(Amat,(TxT-Ox)) |
---|
181 | |
---|
182 | def calcAngle(Ox,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat): |
---|
183 | VecA = calcVec(Ox,TxA,unitA,invA,CA,MA,TA,Amat) |
---|
184 | VecA /= np.sqrt(np.sum(VecA**2)) |
---|
185 | VecB = calcVec(Ox,TxB,unitB,invB,CB,MB,TB,Amat) |
---|
186 | VecB /= np.sqrt(np.sum(VecB**2)) |
---|
187 | edge = VecB-VecA |
---|
188 | edge = np.sum(edge**2) |
---|
189 | angle = (2.-edge)/2. |
---|
190 | angle = max(angle,-1.) |
---|
191 | return acosd(angle) |
---|
192 | |
---|
193 | OxAN,OxA,TxAN,TxA,unitA,TopA = VA |
---|
194 | OxBN,OxB,TxBN,TxB,unitB,TopB = VB |
---|
195 | invA = invB = 1 |
---|
196 | invA = TopA/abs(TopA) |
---|
197 | invB = TopB/abs(TopB) |
---|
198 | centA = abs(TopA)/100 |
---|
199 | centB = abs(TopB)/100 |
---|
200 | opA = abs(TopA)%100-1 |
---|
201 | opB = abs(TopB)%100-1 |
---|
202 | MA,TA = SGData['SGOps'][opA] |
---|
203 | MB,TB = SGData['SGOps'][opB] |
---|
204 | CA = SGData['SGCen'][centA] |
---|
205 | CB = SGData['SGCen'][centB] |
---|
206 | if 'covMatrix' in covData: |
---|
207 | covMatrix = covData['covMatrix'] |
---|
208 | varyList = covData['varyList'] |
---|
209 | AngVcov = getVCov(OxAN+TxAN+TxBN,varyList,covMatrix) |
---|
210 | dx = .00001 |
---|
211 | dadx = np.zeros(9) |
---|
212 | Ang = calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat) |
---|
213 | for i in [0,1,2]: |
---|
214 | OxA[i] += dx |
---|
215 | a0 = calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat) |
---|
216 | OxA[i] -= 2*dx |
---|
217 | dadx[i] = (calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat)-a0)/dx |
---|
218 | OxA[i] += dx |
---|
219 | |
---|
220 | TxA[i] += dx |
---|
221 | a0 = calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat) |
---|
222 | TxA[i] -= 2*dx |
---|
223 | dadx[i+3] = (calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat)-a0)/dx |
---|
224 | TxA[i] += dx |
---|
225 | |
---|
226 | TxB[i] += dx |
---|
227 | a0 = calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat) |
---|
228 | TxB[i] -= 2*dx |
---|
229 | dadx[i+6] = (calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat)-a0)/dx |
---|
230 | TxB[i] += dx |
---|
231 | |
---|
232 | sigAng = np.sqrt(np.inner(dadx,np.inner(AngVcov,dadx))) |
---|
233 | if sigAng < 0.01: |
---|
234 | sigAng = 0.0 |
---|
235 | return Ang,sigAng |
---|
236 | else: |
---|
237 | return calcAngle(OxA,TxA,TxB,unitA,unitB,invA,CA,MA,TA,invB,CB,MB,TB,Amat),0.0 |
---|
238 | |
---|
239 | def GetDistSig(Oatoms,Atoms,Amat,SGData,covData={}): |
---|
240 | |
---|
241 | def calcDist(Atoms,SyOps,Amat): |
---|
242 | XYZ = [] |
---|
243 | for i,atom in enumerate(Atoms): |
---|
244 | Inv,M,T,C,U = SyOps[i] |
---|
245 | XYZ.append(np.array(atom[1:4])) |
---|
246 | XYZ[-1] = Inv*(np.inner(M,np.array(XYZ[-1]))+T)+C+U |
---|
247 | XYZ[-1] = np.inner(Amat,XYZ[-1]).T |
---|
248 | V1 = XYZ[1]-XYZ[0] |
---|
249 | return np.sqrt(np.sum(V1**2)) |
---|
250 | |
---|
251 | Inv = [] |
---|
252 | SyOps = [] |
---|
253 | names = [] |
---|
254 | for i,atom in enumerate(Oatoms): |
---|
255 | names += atom[-1] |
---|
256 | Op,unit = Atoms[i][-1] |
---|
257 | inv = Op/abs(Op) |
---|
258 | m,t = SGData['SGOps'][abs(Op)%100-1] |
---|
259 | c = SGData['SGCen'][abs(Op)/100] |
---|
260 | SyOps.append([inv,m,t,c,unit]) |
---|
261 | Dist = calcDist(Oatoms,SyOps,Amat) |
---|
262 | |
---|
263 | sig = -0.001 |
---|
264 | if 'covMatrix' in covData: |
---|
265 | parmNames = [] |
---|
266 | dx = .00001 |
---|
267 | dadx = np.zeros(6) |
---|
268 | for i in range(6): |
---|
269 | ia = i/3 |
---|
270 | ix = i%3 |
---|
271 | Oatoms[ia][ix+1] += dx |
---|
272 | a0 = calcDist(Oatoms,SyOps,Amat) |
---|
273 | Oatoms[ia][ix+1] -= 2*dx |
---|
274 | dadx[i] = (calcDist(Oatoms,SyOps,Amat)-a0)/(2.*dx) |
---|
275 | covMatrix = covData['covMatrix'] |
---|
276 | varyList = covData['varyList'] |
---|
277 | DistVcov = getVCov(names,varyList,covMatrix) |
---|
278 | sig = np.sqrt(np.inner(dadx,np.inner(DistVcov,dadx))) |
---|
279 | if sig < 0.001: |
---|
280 | sig = -0.001 |
---|
281 | |
---|
282 | return Dist,sig |
---|
283 | |
---|
284 | def GetAngleSig(Oatoms,Atoms,Amat,SGData,covData={}): |
---|
285 | |
---|
286 | def calcAngle(Atoms,SyOps,Amat): |
---|
287 | XYZ = [] |
---|
288 | for i,atom in enumerate(Atoms): |
---|
289 | Inv,M,T,C,U = SyOps[i] |
---|
290 | XYZ.append(np.array(atom[1:4])) |
---|
291 | XYZ[-1] = Inv*(np.inner(M,np.array(XYZ[-1]))+T)+C+U |
---|
292 | XYZ[-1] = np.inner(Amat,XYZ[-1]).T |
---|
293 | V1 = XYZ[1]-XYZ[0] |
---|
294 | V1 /= np.sqrt(np.sum(V1**2)) |
---|
295 | V2 = XYZ[1]-XYZ[2] |
---|
296 | V2 /= np.sqrt(np.sum(V2**2)) |
---|
297 | V3 = V2-V1 |
---|
298 | cang = min(1.,max((2.-np.sum(V3**2))/2.,-1.)) |
---|
299 | return acosd(cang) |
---|
300 | |
---|
301 | Inv = [] |
---|
302 | SyOps = [] |
---|
303 | names = [] |
---|
304 | for i,atom in enumerate(Oatoms): |
---|
305 | names += atom[-1] |
---|
306 | Op,unit = Atoms[i][-1] |
---|
307 | inv = Op/abs(Op) |
---|
308 | m,t = SGData['SGOps'][abs(Op)%100-1] |
---|
309 | c = SGData['SGCen'][abs(Op)/100] |
---|
310 | SyOps.append([inv,m,t,c,unit]) |
---|
311 | Angle = calcAngle(Oatoms,SyOps,Amat) |
---|
312 | |
---|
313 | sig = -0.01 |
---|
314 | if 'covMatrix' in covData: |
---|
315 | parmNames = [] |
---|
316 | dx = .00001 |
---|
317 | dadx = np.zeros(9) |
---|
318 | for i in range(9): |
---|
319 | ia = i/3 |
---|
320 | ix = i%3 |
---|
321 | Oatoms[ia][ix+1] += dx |
---|
322 | a0 = calcAngle(Oatoms,SyOps,Amat) |
---|
323 | Oatoms[ia][ix+1] -= 2*dx |
---|
324 | dadx[i] = (calcAngle(Oatoms,SyOps,Amat)-a0)/(2.*dx) |
---|
325 | covMatrix = covData['covMatrix'] |
---|
326 | varyList = covData['varyList'] |
---|
327 | AngVcov = getVCov(names,varyList,covMatrix) |
---|
328 | sig = np.sqrt(np.inner(dadx,np.inner(AngVcov,dadx))) |
---|
329 | if sig < 0.01: |
---|
330 | sig = -0.01 |
---|
331 | |
---|
332 | return Angle,sig |
---|
333 | |
---|
334 | def GetTorsionSig(Oatoms,Atoms,Amat,SGData,covData={}): |
---|
335 | |
---|
336 | def calcTorsion(Atoms,SyOps,Amat): |
---|
337 | |
---|
338 | XYZ = [] |
---|
339 | for i,atom in enumerate(Atoms): |
---|
340 | Inv,M,T,C,U = SyOps[i] |
---|
341 | XYZ.append(np.array(atom[1:4])) |
---|
342 | XYZ[-1] = Inv*(np.inner(M,np.array(XYZ[-1]))+T)+C+U |
---|
343 | XYZ[-1] = np.inner(Amat,XYZ[-1]).T |
---|
344 | V1 = XYZ[1]-XYZ[0] |
---|
345 | V2 = XYZ[2]-XYZ[1] |
---|
346 | V3 = XYZ[3]-XYZ[2] |
---|
347 | V1 /= np.sqrt(np.sum(V1**2)) |
---|
348 | V2 /= np.sqrt(np.sum(V2**2)) |
---|
349 | V3 /= np.sqrt(np.sum(V3**2)) |
---|
350 | M = np.array([V1,V2,V3]) |
---|
351 | D = nl.det(M) |
---|
352 | Ang = 1.0 |
---|
353 | P12 = np.dot(V1,V2) |
---|
354 | P13 = np.dot(V1,V3) |
---|
355 | P23 = np.dot(V2,V3) |
---|
356 | Tors = acosd((P12*P23-P13)/(np.sqrt(1.-P12**2)*np.sqrt(1.-P23**2)))*D/abs(D) |
---|
357 | return Tors |
---|
358 | |
---|
359 | Inv = [] |
---|
360 | SyOps = [] |
---|
361 | names = [] |
---|
362 | for i,atom in enumerate(Oatoms): |
---|
363 | names += atom[-1] |
---|
364 | Op,unit = Atoms[i][-1] |
---|
365 | inv = Op/abs(Op) |
---|
366 | m,t = SGData['SGOps'][abs(Op)%100-1] |
---|
367 | c = SGData['SGCen'][abs(Op)/100] |
---|
368 | SyOps.append([inv,m,t,c,unit]) |
---|
369 | Tors = calcTorsion(Oatoms,SyOps,Amat) |
---|
370 | |
---|
371 | sig = -0.01 |
---|
372 | if 'covMatrix' in covData: |
---|
373 | parmNames = [] |
---|
374 | dx = .00001 |
---|
375 | dadx = np.zeros(12) |
---|
376 | for i in range(12): |
---|
377 | ia = i/3 |
---|
378 | ix = i%3 |
---|
379 | Oatoms[ia][ix+1] += dx |
---|
380 | a0 = calcTorsion(Oatoms,SyOps,Amat) |
---|
381 | Oatoms[ia][ix+1] -= 2*dx |
---|
382 | dadx[i] = (calcTorsion(Oatoms,SyOps,Amat)-a0)/(2.*dx) |
---|
383 | covMatrix = covData['covMatrix'] |
---|
384 | varyList = covData['varyList'] |
---|
385 | TorVcov = getVCov(names,varyList,covMatrix) |
---|
386 | sig = np.sqrt(np.inner(dadx,np.inner(TorVcov,dadx))) |
---|
387 | if sig < 0.01: |
---|
388 | sig = -0.01 |
---|
389 | |
---|
390 | return Tors,sig |
---|
391 | |
---|
392 | def GetDATSig(Oatoms,Atoms,Amat,SGData,covData={}): |
---|
393 | |
---|
394 | def calcDist(Atoms,SyOps,Amat): |
---|
395 | XYZ = [] |
---|
396 | for i,atom in enumerate(Atoms): |
---|
397 | Inv,M,T,C,U = SyOps[i] |
---|
398 | XYZ.append(np.array(atom[1:4])) |
---|
399 | XYZ[-1] = Inv*(np.inner(M,np.array(XYZ[-1]))+T)+C+U |
---|
400 | XYZ[-1] = np.inner(Amat,XYZ[-1]).T |
---|
401 | V1 = XYZ[1]-XYZ[0] |
---|
402 | return np.sqrt(np.sum(V1**2)) |
---|
403 | |
---|
404 | def calcAngle(Atoms,SyOps,Amat): |
---|
405 | XYZ = [] |
---|
406 | for i,atom in enumerate(Atoms): |
---|
407 | Inv,M,T,C,U = SyOps[i] |
---|
408 | XYZ.append(np.array(atom[1:4])) |
---|
409 | XYZ[-1] = Inv*(np.inner(M,np.array(XYZ[-1]))+T)+C+U |
---|
410 | XYZ[-1] = np.inner(Amat,XYZ[-1]).T |
---|
411 | V1 = XYZ[1]-XYZ[0] |
---|
412 | V1 /= np.sqrt(np.sum(V1**2)) |
---|
413 | V2 = XYZ[1]-XYZ[2] |
---|
414 | V2 /= np.sqrt(np.sum(V2**2)) |
---|
415 | V3 = V2-V1 |
---|
416 | cang = min(1.,max((2.-np.sum(V3**2))/2.,-1.)) |
---|
417 | return acosd(cang) |
---|
418 | |
---|
419 | def calcTorsion(Atoms,SyOps,Amat): |
---|
420 | |
---|
421 | XYZ = [] |
---|
422 | for i,atom in enumerate(Atoms): |
---|
423 | Inv,M,T,C,U = SyOps[i] |
---|
424 | XYZ.append(np.array(atom[1:4])) |
---|
425 | XYZ[-1] = Inv*(np.inner(M,np.array(XYZ[-1]))+T)+C+U |
---|
426 | XYZ[-1] = np.inner(Amat,XYZ[-1]).T |
---|
427 | V1 = XYZ[1]-XYZ[0] |
---|
428 | V2 = XYZ[2]-XYZ[1] |
---|
429 | V3 = XYZ[3]-XYZ[2] |
---|
430 | V1 /= np.sqrt(np.sum(V1**2)) |
---|
431 | V2 /= np.sqrt(np.sum(V2**2)) |
---|
432 | V3 /= np.sqrt(np.sum(V3**2)) |
---|
433 | M = np.array([V1,V2,V3]) |
---|
434 | D = nl.det(M) |
---|
435 | Ang = 1.0 |
---|
436 | P12 = np.dot(V1,V2) |
---|
437 | P13 = np.dot(V1,V3) |
---|
438 | P23 = np.dot(V2,V3) |
---|
439 | Tors = acosd((P12*P23-P13)/(np.sqrt(1.-P12**2)*np.sqrt(1.-P23**2)))*D/abs(D) |
---|
440 | return Tors |
---|
441 | |
---|
442 | Inv = [] |
---|
443 | SyOps = [] |
---|
444 | names = [] |
---|
445 | for i,atom in enumerate(Oatoms): |
---|
446 | names += atom[-1] |
---|
447 | Op,unit = Atoms[i][-1] |
---|
448 | inv = Op/abs(Op) |
---|
449 | m,t = SGData['SGOps'][abs(Op)%100-1] |
---|
450 | c = SGData['SGCen'][abs(Op)/100] |
---|
451 | SyOps.append([inv,m,t,c,unit]) |
---|
452 | M = len(Oatoms) |
---|
453 | if M == 2: |
---|
454 | Val = calcDist(Oatoms,SyOps,Amat) |
---|
455 | elif M == 3: |
---|
456 | Val = calcAngle(Oatoms,SyOps,Amat) |
---|
457 | else: |
---|
458 | Val = calcTorsion(Oatoms,SyOps,Amat) |
---|
459 | |
---|
460 | sigVals = [-0.001,-0.01,-0.01] |
---|
461 | sig = sigVals[M-3] |
---|
462 | if 'covMatrix' in covData: |
---|
463 | parmNames = [] |
---|
464 | dx = .00001 |
---|
465 | N = M*3 |
---|
466 | dadx = np.zeros(N) |
---|
467 | for i in range(N): |
---|
468 | ia = i/3 |
---|
469 | ix = i%3 |
---|
470 | Oatoms[ia][ix+1] += dx |
---|
471 | if M == 2: |
---|
472 | a0 = calcDist(Oatoms,SyOps,Amat) |
---|
473 | elif M == 3: |
---|
474 | a0 = calcAngle(Oatoms,SyOps,Amat) |
---|
475 | else: |
---|
476 | a0 = calcTorsion(Oatoms,SyOps,Amat) |
---|
477 | Oatoms[ia][ix+1] -= 2*dx |
---|
478 | if M == 2: |
---|
479 | dadx[i] = (calcDist(Oatoms,SyOps,Amat)-a0)/(2.*dx) |
---|
480 | elif M == 3: |
---|
481 | dadx[i] = (calcAngle(Oatoms,SyOps,Amat)-a0)/(2.*dx) |
---|
482 | else: |
---|
483 | dadx[i] = (calcTorsion(Oatoms,SyOps,Amat)-a0)/(2.*dx) |
---|
484 | covMatrix = covData['covMatrix'] |
---|
485 | varyList = covData['varyList'] |
---|
486 | Vcov = getVCov(names,varyList,covMatrix) |
---|
487 | sig = np.sqrt(np.inner(dadx,np.inner(Vcov,dadx))) |
---|
488 | if sig < sigVals[M-3]: |
---|
489 | sig = sigVals[M-3] |
---|
490 | |
---|
491 | return Val,sig |
---|
492 | |
---|
493 | |
---|
494 | def ValEsd(value,esd=0,nTZ=False): #NOT complete - don't use |
---|
495 | # returns value(esd) string; nTZ=True for no trailing zeros |
---|
496 | # use esd < 0 for level of precision shown e.g. esd=-0.01 gives 2 places beyond decimal |
---|
497 | #get the 2 significant digits in the esd |
---|
498 | edig = lambda esd: int(round(10**(math.log10(esd) % 1+1))) |
---|
499 | #get the number of digits to represent them |
---|
500 | epl = lambda esd: 2+int(1.545-math.log10(10*edig(esd))) |
---|
501 | |
---|
502 | mdec = lambda esd: -int(round(math.log10(abs(esd))))+1 |
---|
503 | ndec = lambda esd: int(1.545-math.log10(abs(esd))) |
---|
504 | if esd > 0: |
---|
505 | fmt = '"%.'+str(ndec(esd))+'f(%d)"' |
---|
506 | return str(fmt%(value,int(round(esd*10**(mdec(esd)))))).strip('"') |
---|
507 | elif esd < 0: |
---|
508 | return str(round(value,mdec(esd)-1)) |
---|
509 | else: |
---|
510 | text = str("%f"%(value)) |
---|
511 | if nTZ: |
---|
512 | return text.rstrip('0') |
---|
513 | else: |
---|
514 | return text |
---|
515 | |
---|
516 | |
---|